VxWorks Kernel Programmer's Guide, 6.6 Programmers Guide
User Manual:
Open the PDF directly: View PDF 
.
Page Count: 866
| Download | |
| Open PDF In Browser | View PDF |
VxWorks Kernel Programmer's Guide, 6.6
VxWorks
®
KERNEL PROGRAMMER'S GUIDE
6.6
Copyright © 2007 Wind River Systems, Inc.
All rights reserved. No part of this publication may be reproduced or transmitted in any
form or by any means without the prior written permission of Wind River Systems, Inc.
Wind River, Tornado, and VxWorks are registered trademarks of Wind River Systems, Inc.
The Wind River logo is a trademark of Wind River Systems, Inc. Any third-party
trademarks referenced are the property of their respective owners. For further information
regarding Wind River trademarks, please see:
http://www.windriver.com/company/terms/trademark.html
This product may include software licensed to Wind River by third parties. Relevant
notices (if any) are provided in your product installation at the following location:
installDir/product_name/3rd_party_licensor_notice.pdf.
Wind River may refer to third-party documentation by listing publications or providing
links to third-party Web sites for informational purposes. Wind River accepts no
responsibility for the information provided in such third-party documentation.
Corporate Headquarters
Wind River Systems, Inc.
500 Wind River Way
Alameda, CA 94501-1153
U.S.A.
toll free (U.S.): (800) 545-WIND
telephone: (510) 748-4100
facsimile: (510) 749-2010
For additional contact information, please visit the Wind River URL:
http://www.windriver.com
For information on how to contact Customer Support, please visit the following URL:
http://www.windriver.com/support
VxWorks Kernel Programmer's Guide, 6.6
14 Nov 07
Part #: DOC-16075-ND-00
Contents
PART I: CORE TECHNOLOGIES
1
2
Overview ...............................................................................................
3
1.1
Introduction .............................................................................................................
3
1.2
Related Documentation Resources .....................................................................
4
1.3
VxWorks Configuration and Build .....................................................................
5
Kernel ....................................................................................................
7
2.1
Introduction .............................................................................................................
7
2.2
Kernel Architecture ................................................................................................
9
2.2.1
Operating System Facilities ....................................................................
10
2.2.2
System Tasks .............................................................................................
11
2.3
System Startup ........................................................................................................
13
2.4
VxWorks Configuration ........................................................................................
14
2.4.1
VxWorks Image Types ............................................................................
15
2.4.2
VxWorks Components ............................................................................
17
2.4.3
Device Driver Selection ...........................................................................
22
iii
VxWorks
Kernel Programmer's Guide, 6.6
2.4.4
VxWorks Configuration Profiles ............................................................
22
2.4.5
Small VxWorks Configuration Profiles ................................................
24
How Small Configuration Profiles Create Smaller Systems ..............
Configuring and Building Small VxWorks Configurations ...............
Optimization of Small VxWorks Profile Systems ................................
Minimal Kernel Profile ............................................................................
Basic Kernel Profile ..................................................................................
Basic OS Profile .........................................................................................
25
27
27
28
33
35
Customizing VxWorks Code ..................................................................
39
System Clock Modification .....................................................................
Hardware Initialization Customization ................................................
Other Customization ...............................................................................
39
39
39
Power Management ...............................................................................................
40
2.5.1
Power Management for IA Architecture ..............................................
41
ACPI Processor Power and Performance States ..................................
ACPI Thermal Management ...................................................................
VxWorks Power Management Facilities ...............................................
Configuring VxWorks With Power Management Facilities ..............
Power Management and System Performance ....................................
41
43
43
49
49
Power Management for Other Architectures .......................................
49
Kernel Applications ...............................................................................................
51
2.6.1
C and C++ Libraries .................................................................................
52
2.6.2
Application Structure ..............................................................................
52
2.6.3
VxWorks Header Files .............................................................................
53
2.6.4
Static Instantiation of Kernel Objects ....................................................
56
2.6.5
Applications and VxWorks Kernel Component Requirements ........
62
2.6.6
Building Kernel Application Modules ..................................................
62
2.6.7
Downloading Kernel Application Object Modules to a Target .........
64
2.6.8
Linking Kernel Application Object Modules with VxWorks ............
64
2.6.9
Image Size Considerations ......................................................................
65
2.6.10
Configuring VxWorks to Run Applications Automatically ..............
66
2.4.6
2.5
2.5.2
2.6
iv
Contents
2.7
Custom Kernel Libraries .......................................................................................
67
2.8
Custom VxWorks Components and CDFs ........................................................
67
2.8.1
Creating and Modifying Components ..................................................
68
Defining a Component ............................................................................
Modifying a Component .........................................................................
68
74
2.8.2
CDF Precedence and CDF Installation ..................................................
75
2.8.3
Testing New Components ......................................................................
78
2.8.4
Component Description Language .......................................................
79
2.8.5
CDF Naming Conventions .....................................................................
80
2.8.6
Component CDF Object ..........................................................................
81
Component Properties ............................................................................
Component Template ..............................................................................
82
85
Parameter CDF Object .............................................................................
87
Parameter Properties ...............................................................................
Parameter Template .................................................................................
87
88
Initialization Group CDF Object ............................................................
88
Initialization Group Properties ..............................................................
Initialization Group Template ................................................................
89
90
Bundle CDF Object ...................................................................................
90
Bundle Properties .....................................................................................
90
Profile CDF Object ...................................................................................
91
Profile Properties ......................................................................................
Profile Template .......................................................................................
91
91
Folder CDF Object ....................................................................................
92
Folder Properties ......................................................................................
Folder Template .......................................................................................
93
94
Selection CDF Object ...............................................................................
94
Selection Properties ..................................................................................
Selection Template ...................................................................................
95
95
CDF Template ...........................................................................................
96
2.8.7
2.8.8
2.8.9
2.8.10
2.8.11
2.8.12
2.8.13
v
VxWorks
Kernel Programmer's Guide, 6.6
2.9
Custom System Calls ............................................................................................. 100
2.9.1
How System Calls Work ......................................................................... 100
2.9.2
System Call Requirements ...................................................................... 101
System Call Naming Rules .....................................................................
System Call Numbering Rules ...............................................................
System Call Argument Rules ..................................................................
System Call Return Value Rules ............................................................
2.9.3
101
102
103
104
System Call Handler Requirements ...................................................... 105
System Call Handler Naming Rules ...................................................... 106
System Call Handler Argument Validation ......................................... 106
System Call Handler Error Reporting ................................................... 107
2.9.4
Adding System Calls ............................................................................... 107
Adding System Calls Statically .............................................................. 107
Adding System Calls Dynamically ........................................................ 112
2.10
2.9.5
Monitoring And Debugging System Calls ........................................... 114
2.9.6
Documenting Custom System Calls ...................................................... 116
Custom Scheduler .................................................................................................. 118
2.10.1
Requirements for a Custom Scheduler ................................................. 118
Code Requirements .................................................................................. 118
Configuration Requirements .................................................................. 121
2.10.2
Traditional VxWorks Scheduler ............................................................. 123
Scheduler Initialization ............................................................................ 123
Multi-way Queue Structure .................................................................... 124
Q_CLASS Operators ................................................................................ 126
3
Boot Loader .......................................................................................... 131
3.1
Introduction ............................................................................................................. 132
3.2
Using a Default Boot Loader ................................................................................ 132
3.3
Boot Loader Image Types ..................................................................................... 133
3.4
Boot Loader Shell ................................................................................................... 135
vi
Contents
3.4.1
3.5
Boot Loader Shell Commands ................................................................ 136
Boot Loader Parameters ......................................................................................... 140
3.5.1
Displaying Current Boot Parameters .................................................... 140
3.5.2
Description of Boot Loader Parameters ................................................ 141
3.5.3
Changing Boot Loader Parameters Interactively ................................ 144
3.6
Rebooting VxWorks ............................................................................................... 145
3.7
Customizing and Building Boot Loaders .......................................................... 146
3.7.1
Configuring Boot Loaders ....................................................................... 146
3.7.2
Boot Loader Components ....................................................................... 146
3.7.3
Configuring Boot Loader Parameters Statically .................................. 147
3.7.4
Enabling Networking for Non-Boot Interfaces .................................... 148
3.7.5
Selecting a Boot Device ............................................................................ 148
3.7.6
Reconfiguring Memory ........................................................................... 149
Persistent Memory Region ...................................................................... 150
3.7.7
4
Building Boot Loaders ............................................................................. 151
3.8
Installing Boot Loaders ......................................................................................... 152
3.9
Booting From a Network ...................................................................................... 152
3.10
Booting From a Target File System .................................................................... 154
3.11
Booting From the Host File System Using TSFS ............................................. 155
Multitasking .......................................................................................... 157
4.1
Introduction ............................................................................................................. 159
4.2
Tasks and Multitasking ....................................................................................... 160
4.2.1
Task States and Transitions .................................................................... 161
vii
VxWorks
Kernel Programmer's Guide, 6.6
4.3
Task Scheduling ..................................................................................................... 166
4.3.1
Task Priorities ........................................................................................... 166
4.3.2
Task Scheduling Control ......................................................................... 167
Task Priority .............................................................................................. 167
Preemption Locks ..................................................................................... 167
4.3.3
VxWorks Traditional Scheduler ............................................................. 168
Priority-Based Preemptive Scheduling ................................................. 169
Round-Robin Scheduling ........................................................................ 169
4.4
Task Creation and Management ......................................................................... 171
4.4.1
Task Creation and Activation ................................................................. 172
4.4.2
Task Creation Options ............................................................................. 173
4.4.3
Task Stack .................................................................................................. 175
Task Stack Protection ............................................................................... 176
4.5
4.4.4
Task Names and IDs ................................................................................ 177
4.4.5
Task Information ...................................................................................... 179
4.4.6
Task Deletion and Deletion Safety ......................................................... 180
4.4.7
Task Execution Control ........................................................................... 181
4.4.8
Tasking Extensions ................................................................................... 182
Task Error Status: errno ......................................................................................... 184
4.5.1
Layered Definitions of errno .................................................................. 185
4.5.2
A Separate errno Value for Each Task .................................................. 185
4.5.3
Error Return Convention ........................................................................ 186
4.5.4
Assignment of Error Status Values ........................................................ 186
4.6
Task Exception Handling ...................................................................................... 187
4.7
Shared Code and Reentrancy ............................................................................... 187
4.7.1
Dynamic Stack Variables ......................................................................... 189
4.7.2
Guarded Global and Static Variables .................................................... 190
4.7.3
Task-Specific Variables ........................................................................... 190
viii
Contents
Thread-Local Variables: __thread Storage Class ................................. 190
taskVarLib and Task Variables .............................................................. 191
4.7.4
Multiple Tasks with the Same Main Routine ....................................... 192
4.8
Intertask and Interprocess Communication ...................................................... 193
4.9
Public and Private Objects ................................................................................... 194
4.9.1
Creating and Naming Public and Private Objects .............................. 194
4.9.2
Object Ownership and Resource Reclamation ..................................... 195
4.10
Shared Data Structures .......................................................................................... 196
4.11
Mutual Exclusion .................................................................................................... 196
4.12
4.11.1
Interrupt Locks and Latency .................................................................. 197
4.11.2
Preemptive Locks and Latency .............................................................. 197
Semaphores ............................................................................................................. 198
4.12.1
Semaphore Control .................................................................................. 200
Static Instantiation of Semaphores ........................................................ 201
4.12.2
Binary Semaphores .................................................................................. 201
Mutual Exclusion ..................................................................................... 203
Synchronization ........................................................................................ 203
4.12.3
Mutual-Exclusion Semaphores .............................................................. 205
Priority Inversion ..................................................................................... 205
Deletion Safety .......................................................................................... 207
Recursive Resource Access ..................................................................... 207
4.12.4
Counting Semaphores ............................................................................. 208
4.12.5
Read/Write Semaphores ........................................................................ 209
Specification of Read or Write Mode .................................................... 210
Precedence for Write Access Operations .............................................. 211
Read/Write Semaphores and System Performance ........................... 211
4.12.6
Special Semaphore Options .................................................................... 211
Timeouts .................................................................................................... 212
Queues ....................................................................................................... 212
ix
VxWorks
Kernel Programmer's Guide, 6.6
4.12.7
4.13
Semaphores and VxWorks Events ......................................................... 213
Message Queues ..................................................................................................... 213
4.13.1
VxWorks Message Queue Routines ...................................................... 215
4.13.2
Displaying Message Queue Attributes ................................................. 217
4.13.3
Servers and Clients with Message Queues ........................................... 217
4.13.4
Message Queues and VxWorks Events ................................................. 218
4.14
Pipes .......................................................................................................................... 218
4.15
VxWorks Events ...................................................................................................... 219
4.15.1
Preparing a Task to Receive Events ....................................................... 220
4.15.2
Sending Events to a Task ........................................................................ 221
4.15.3
Accessing Event Flags .............................................................................. 223
4.15.4
Events Routines ........................................................................................ 224
4.15.5
Task Events Register ................................................................................ 224
4.15.6
Show Routines and Events ..................................................................... 225
4.16
Message Channels ................................................................................................. 226
4.17
Network Communication ..................................................................................... 226
4.18
Signals ..................................................................................................................... 226
4.18.1
Configuring VxWorks for Signals ......................................................... 228
4.18.2
Basic Signal Routines ............................................................................... 229
4.18.3
Queued Signal Routines ......................................................................... 230
4.18.4
Signal Events ............................................................................................. 236
4.18.5
Signal Handlers ........................................................................................ 236
4.19
Watchdog Timers ................................................................................................... 239
4.20
Interrupt Service Routines .................................................................................... 241
4.20.1
x
Connecting Routines to Interrupts ........................................................ 242
Contents
4.20.2
Interrupt Stack .......................................................................................... 242
Filling Interrupt Stacks ............................................................................ 243
Interrupt Stack Protection ....................................................................... 243
5
4.20.3
Writing and Debugging ISRs .................................................................. 244
4.20.4
ISRs and the Kernel Work Queue .......................................................... 244
4.20.5
Special Limitations of ISRs ...................................................................... 245
4.20.6
Exceptions at Interrupt Level ................................................................. 248
4.20.7
Reserving High Interrupt Levels ........................................................... 248
4.20.8
Additional Restrictions for ISRs at High Interrupt Levels ................. 249
4.20.9
Interrupt-to-Task Communication ........................................................ 249
POSIX Facilities .................................................................................... 251
5.1
Introduction ............................................................................................................. 252
5.2
Configuring VxWorks with POSIX Facilities ................................................... 253
5.2.1
VxWorks Components for POSIX Facilities ......................................... 253
5.3
General POSIX Support ........................................................................................ 255
5.4
POSIX Header Files ............................................................................................... 257
5.5
POSIX Namespace .................................................................................................. 259
5.6
POSIX Clocks and Timers .................................................................................... 259
5.7
POSIX Asynchronous I/O ..................................................................................... 263
5.8
POSIX Advisory File Locking .............................................................................. 263
5.9
POSIX Page-Locking Interface ............................................................................ 264
5.10
POSIX Threads ........................................................................................................ 264
5.10.1
POSIX Thread Attributes ........................................................................ 265
5.10.2
VxWorks-Specific Pthread Attributes ................................................... 265
5.10.3
Specifying Attributes when Creating Pthreads ................................... 266
xi
VxWorks
Kernel Programmer's Guide, 6.6
5.11
5.10.4
POSIX Thread Creation and Management ........................................... 268
5.10.5
POSIX Thread Attribute Access ............................................................. 269
5.10.6
POSIX Thread Private Data .................................................................... 270
5.10.7
POSIX Thread Cancellation .................................................................... 270
POSIX Thread Mutexes and Condition Variables ........................................... 273
5.11.1
Thread Mutexes ........................................................................................ 273
Protocol Mutex Attribute ........................................................................ 274
Priority Ceiling Mutex Attribute ............................................................ 275
5.11.2
5.12
5.13
5.14
Condition Variables ................................................................................. 275
POSIX and VxWorks Scheduling ........................................................................ 277
5.12.1
Differences in POSIX and VxWorks Scheduling ................................. 279
5.12.2
POSIX and VxWorks Priority Numbering ........................................... 279
5.12.3
Default Scheduling Policy ....................................................................... 280
5.12.4
VxWorks Traditional Scheduler ............................................................. 281
5.12.5
POSIX Threads Scheduler ....................................................................... 282
5.12.6
POSIX Scheduling Routines .................................................................... 286
5.12.7
Getting Scheduling Parameters: Priority Limits and Time Slice ....... 287
POSIX Semaphores ................................................................................................ 289
5.13.1
Comparison of POSIX and VxWorks Semaphores .............................. 290
5.13.2
Using Unnamed Semaphores ................................................................. 291
5.13.3
Using Named Semaphores ..................................................................... 294
POSIX Message Queues ........................................................................................ 299
5.14.1
Comparison of POSIX and VxWorks Message Queues ...................... 300
5.14.2
POSIX Message Queue Attributes ......................................................... 300
5.14.3
Displaying Message Queue Attributes ................................................. 303
5.14.4
Communicating Through a Message Queue ....................................... 303
5.14.5
Notification of Message Arrival ............................................................ 307
xii
Contents
6
5.15
POSIX Signals ......................................................................................................... 313
5.16
POSIX Memory Management .............................................................................. 313
Memory Management .......................................................................... 315
6.1
Introduction ............................................................................................................. 316
6.2
Configuring VxWorks With Memory Management Facilities ...................... 317
6.3
System Memory Maps ........................................................................................... 317
6.3.1
System Memory Map Without Process Support ................................. 318
6.3.2
System Memory Map with Process Support ........................................ 321
6.3.3
System Memory Map with Processes Running ................................... 323
6.4
Shell Commands .................................................................................................... 327
6.5
System RAM Autosizing ...................................................................................... 327
6.6
Reserved Memory .................................................................................................. 328
6.7
Kernel Heap and Memory Partition Management .......................................... 329
6.8
6.9
6.7.1
Configuring the Kernel Heap and the Memory Partition Manager . 329
6.7.2
Basic Heap and Memory Partition Manager ........................................ 330
6.7.3
Full Heap and Memory Partition Manager .......................................... 330
Memory Error Detection ....................................................................................... 331
6.8.1
Heap and Partition Memory Instrumentation ..................................... 331
6.8.2
Compiler Instrumentation ...................................................................... 338
Virtual Memory Management ............................................................................. 343
6.9.1
Configuring Virtual Memory Management ......................................... 344
xiii
VxWorks
Kernel Programmer's Guide, 6.6
6.9.2
Managing Virtual Memory Programmatically .................................... 346
Modifying Page States .............................................................................
Making Memory Non-Writable .............................................................
Invalidating Memory Pages ....................................................................
Locking TLB Entries .................................................................................
Page Size Optimization ...........................................................................
Setting Page States in ISRs ......................................................................
6.9.3
6.10
6.11
346
348
350
350
351
351
Troubleshooting ....................................................................................... 352
Additional Memory Protection Features ........................................................... 353
6.10.1
Configuring VxWorks for Additional Memory Protection ................ 354
6.10.2
Stack Overrun and Underrun Detection ............................................... 354
6.10.3
Non-Executable Task Stack ..................................................................... 355
6.10.4
Text Segment Write Protection ............................................................... 355
6.10.5
Exception Vector Table Write Protection .............................................. 355
Processes Without MMU Support ....................................................................... 355
Configuring VxWorks With Process Support for Systems Without an
MMU ........................................................................................... 356
7
I/O System ............................................................................................. 359
7.1
Introduction ............................................................................................................. 360
7.2
Configuring VxWorks With I/O Facilities ......................................................... 362
7.3
Files, Devices, and Drivers ................................................................................... 363
7.3.1
7.4
Filenames and the Default Device ......................................................... 363
Basic I/O ................................................................................................................... 365
7.4.1
File Descriptors ......................................................................................... 365
File Descriptor Table ................................................................................ 366
7.4.2
Standard Input, Standard Output, and Standard Error ..................... 366
7.4.3
Standard I/O Redirection ....................................................................... 367
Issues with Standard I/O Redirection .................................................. 368
xiv
Contents
7.5
7.6
7.7
7.4.4
Open and Close ........................................................................................ 369
7.4.5
Create and Remove .................................................................................. 372
7.4.6
Read and Write ......................................................................................... 372
7.4.7
File Truncation .......................................................................................... 373
7.4.8
I/O Control ............................................................................................... 373
7.4.9
Pending on Multiple File Descriptors with select( ) ............................ 374
7.4.10
POSIX File System Routines ................................................................... 378
Buffered I/O: stdio ................................................................................................. 378
7.5.1
Using stdio ................................................................................................ 379
7.5.2
Standard Input, Standard Output, and Standard Error ..................... 380
Other Formatted I/O .............................................................................................. 380
7.6.1
Special Cases: printf( ), sprintf( ), and sscanf( ) ................................... 380
7.6.2
Additional Routines: printErr( ) and fdprintf( ) .................................. 381
7.6.3
Message Logging ..................................................................................... 381
Asynchronous Input/Output ................................................................................ 381
7.7.1
The POSIX AIO Routines ........................................................................ 382
7.7.2
AIO Control Block .................................................................................... 384
7.7.3
Using AIO .................................................................................................. 385
AIO with Periodic Checks for Completion .......................................... 385
Alternatives for Testing AIO Completion ............................................ 388
7.8
Devices in VxWorks ............................................................................................... 391
7.8.1
Serial I/O Devices: Terminal and Pseudo-Terminal Devices ............ 392
tty Options .................................................................................................
Raw Mode and Line Mode .....................................................................
tty Special Characters ..............................................................................
I/O Control Functions .............................................................................
xv
392
393
394
396
VxWorks
Kernel Programmer's Guide, 6.6
7.8.2
Pipe Devices .............................................................................................. 396
Creating Pipes ........................................................................................... 397
Writing to Pipes from ISRs ..................................................................... 397
I/O Control Functions ............................................................................. 397
7.8.3
Pseudo I/O Device ................................................................................... 398
I/O Control Functions ............................................................................. 398
7.8.4
Network File System (NFS) Devices ...................................................... 399
Mounting a Remote NFS File System from VxWorks ........................ 399
I/O Control Functions for NFS Clients ................................................. 400
7.8.5
Non-NFS Network Devices .................................................................... 400
Creating Network Devices ...................................................................... 401
I/O Control Functions ............................................................................. 401
7.8.6
Null Devices ............................................................................................. 402
7.8.7
Sockets ........................................................................................................ 402
7.8.8
Extended Block Device Facility: XBD .................................................... 402
XBD Disk Partition Manager .................................................................. 403
XBD Block Device Wrapper .................................................................... 404
XBD TRFS Component ............................................................................ 405
7.8.9
Transaction-Based Reliable File System Facility: TRFS ...................... 405
Configuring VxWorks With TRFS .........................................................
Automatic Instantiation of TRFS ............................................................
Formatting a Device for TRFS ................................................................
Using TRFS in Applications ....................................................................
7.8.10
406
406
406
408
Block Devices ............................................................................................ 410
XBD RAM Disk ......................................................................................... 411
SCSI Drivers .............................................................................................. 412
7.9
Differences Between VxWorks and Host System I/O ..................................... 422
7.10
Internal I/O System Structure .............................................................................. 423
7.10.1
Drivers ....................................................................................................... 426
The Driver Table and Installing Drivers ............................................... 427
Example of Installing a Driver ............................................................... 428
xvi
Contents
7.10.2
Devices ....................................................................................................... 429
The Device List and Adding Devices .................................................... 429
Example of Adding Devices ................................................................... 430
Deleting Devices ....................................................................................... 431
7.10.3
File Descriptors ......................................................................................... 435
File Descriptor Table ................................................................................
Example of Opening a File ......................................................................
Example of Reading Data from the File ................................................
Example of Closing a File ........................................................................
Implementing select( ) .............................................................................
Cache Coherency ......................................................................................
8
435
436
439
441
441
445
7.11
PCMCIA ................................................................................................................... 450
7.12
Peripheral Component Interconnect: PCI ......................................................... 450
Local File Systems ............................................................................... 451
8.1
Introduction ............................................................................................................. 452
8.2
File System Monitor .............................................................................................. 455
Device Insertion Events ........................................................................... 456
XBD Name Mapping Facility ................................................................. 457
8.3
Virtual Root File System: VRFS .......................................................................... 457
8.4
Highly Reliable File System: HRFS .................................................................... 459
8.4.1
Configuring VxWorks for HRFS ............................................................ 459
8.4.2
Configuring HRFS ................................................................................... 460
8.4.3
Creating an HRFS File System .............................................................. 461
Overview of HRFS File System Creation .............................................. 461
HRFS File System Creation Steps .......................................................... 462
8.4.4
HRFS, ATA, and RAM Disk Examples ................................................. 463
8.4.5
Transactional Operations and Commit Policies ................................ 469
8.4.6
Configuring Transaction Points at Runtime ....................................... 471
8.4.7
File Access Time Stamps ......................................................................... 473
xvii
VxWorks
Kernel Programmer's Guide, 6.6
8.4.8
Maximum Number of Files and Directories ........................................ 473
8.4.9
Working with Directories ........................................................................ 473
Creating Subdirectories ........................................................................... 473
Removing Subdirectories ........................................................................ 474
Reading Directory Entries ....................................................................... 474
8.4.10
Working with Files ................................................................................... 474
File I/O Routines ...................................................................................... 474
File Linking and Unlinking ..................................................................... 475
File Permissions ........................................................................................ 475
8.4.11
Crash Recovery and Volume Consistency ........................................... 475
Crash Recovery ......................................................................................... 475
Consistency Checking .............................................................................. 476
8.4.12
8.5
I/O Control Functions Supported by HRFS ........................................ 476
MS-DOS-Compatible File System: dosFs .......................................................... 478
8.5.1
Configuring VxWorks for dosFs ............................................................ 478
8.5.2
Configuring dosFs ................................................................................... 480
8.5.3
Creating a dosFs File System .................................................................. 482
Overview of dosFs File System Creation .............................................. 482
dosFs File System Creation Steps ........................................................... 482
8.5.4
dosFs, ATA Disk, and RAM Disk Examples ........................................ 486
8.5.5
Working with Volumes and Disks ......................................................... 491
Accessing Volume Configuration Information .................................... 492
Synchronizing Volumes .......................................................................... 492
8.5.6
Working with Directories ........................................................................ 492
Creating Subdirectories ........................................................................... 492
Removing Subdirectories ........................................................................ 493
Reading Directory Entries ....................................................................... 493
8.5.7
Working with Files ................................................................................... 493
File I/O Routines ...................................................................................... 493
File Attributes ........................................................................................... 493
8.5.8
Disk Space Allocation Options ............................................................... 496
Choosing an Allocation Method ............................................................ 496
xviii
Contents
Using Cluster Group Allocation ............................................................ 497
Using Absolutely Contiguous Allocation ............................................. 497
8.6
8.7
8.8
8.9
8.5.9
Crash Recovery and Volume Consistency ........................................... 499
8.5.10
I/O Control Functions Supported by dosFsLib ................................... 500
8.5.11
Booting from a Local dosFs File System Using SCSI .......................... 503
Raw File System: rawFs ......................................................................................... 505
8.6.1
Configuring VxWorks for rawFs ........................................................... 505
8.6.2
Creating a rawFs File System ................................................................. 506
8.6.3
Mounting rawFs Volumes ...................................................................... 508
8.6.4
rawFs File I/O .......................................................................................... 509
8.6.5
I/O Control Functions Supported by rawFsLib .................................. 509
CD-ROM File System: cdromFs .......................................................................... 510
8.7.1
Configuring VxWorks for cdromFs ....................................................... 511
8.7.2
Creating and Using cdromFs .................................................................. 512
8.7.3
I/O Control Functions Supported by cdromFsLib ............................. 515
8.7.4
Version Numbers ..................................................................................... 516
Read-Only Memory File System: ROMFS ........................................................ 516
8.8.1
Configuring VxWorks with ROMFS ..................................................... 517
8.8.2
Building a System With ROMFS and Files ........................................... 517
8.8.3
Accessing Files in ROMFS ...................................................................... 517
8.8.4
Using ROMFS to Start Applications Automatically ........................... 518
Target Server File System: TSFS ......................................................................... 518
Socket Support ..........................................................................................
Error Handling .........................................................................................
Configuring VxWorks for TSFS Use ......................................................
Security Considerations ..........................................................................
Using the TSFS to Boot a Target .............................................................
xix
519
520
520
520
521
VxWorks
Kernel Programmer's Guide, 6.6
9
Network File System: NFS ................................................................... 523
9.1
Introduction ............................................................................................................ 523
9.2
Configuring VxWorks for an NFS Client .......................................................... 524
Core NFS Client ........................................................................................
NFS Client All ...........................................................................................
NFS v2 Client ............................................................................................
NFS v3 Client ............................................................................................
NFS Mount All ..........................................................................................
9.3
Creating an NFS Client ......................................................................................... 529
Exporting File Systems from the Remote NFS Server ........................
Setting Your NFS Client Name, User ID, and Group ID ....................
Mounting a Remote File System ............................................................
Using ioctl( ) With Open Files from a Mounted Directory ................
9.4
530
531
531
534
Configuring VxWorks for an NFS Server .......................................................... 535
NFS Server .................................................................................................
NFS server All ...........................................................................................
NFS server V2 ...........................................................................................
NFS server V3 ...........................................................................................
9.5
524
524
525
526
529
536
537
537
537
Creating an NFS Server ......................................................................................... 540
Initializing an NFS File System for Export ........................................... 540
Exporting a File System through NFS ................................................... 540
10
Flash File System Support: TrueFFS ................................................. 543
10.1
Introduction ............................................................................................................. 543
10.2
Overview of Implementation Steps .................................................................... 545
10.3
Creating a System with TrueFFS ......................................................................... 546
10.3.1
Selecting an MTD ..................................................................................... 546
10.3.2
Identifying the Socket Driver ................................................................. 547
10.3.3
Configuring VxWorks with TrueFFS .................................................... 548
Including File System Components ....................................................... 548
xx
Contents
Including the XBD Wrapper Component .............................................
Including the Core TrueFFS Component .............................................
Including Utility Components ...............................................................
Including the MTD Component .............................................................
Including the Translation Layer Component .......................................
Adding the Socket Driver .......................................................................
549
549
549
550
550
551
10.3.4
Building the System ................................................................................ 551
10.3.5
Formatting the Flash ................................................................................ 552
Specifying the Drive Number ................................................................. 552
Specifying Format Options ..................................................................... 552
10.3.6
Reserving a Region in Flash for a Boot Image ..................................... 554
Reserving a Fallow Region ..................................................................... 554
Writing the Boot Image to Flash ............................................................ 556
11
10.3.7
Mounting the Drive ................................................................................. 556
10.3.8
Testing the Drive ...................................................................................... 557
10.4
Using TrueFFS Shell Commands ........................................................................ 558
10.5
Using TrueFFS With HRFS .................................................................................. 560
10.5.1
TrueFFS With HRFS Code Example ...................................................... 560
10.5.2
TrueFFS With HRFS Shell Command Example ................................... 560
Error Detection and Reporting ............................................................ 563
11.1
Introduction ............................................................................................................. 563
11.2
Configuring Error Detection and Reporting Facilities ................................... 564
11.2.1
Configuring VxWorks ............................................................................. 565
11.2.2
Configuring the Persistent Memory Region ........................................ 565
11.2.3
Configuring Responses to Fatal Errors ................................................. 566
11.3
Error Records ........................................................................................................... 566
11.4
Displaying and Clearing Error Records ............................................................. 568
xxi
VxWorks
Kernel Programmer's Guide, 6.6
11.5
Fatal Error Handling Options .............................................................................. 569
11.5.1
Configuring VxWorks with Error Handling Options ......................... 570
11.5.2
Setting the System Debug Flag ............................................................... 571
Setting the Debug Flag Statically ........................................................... 571
Setting the Debug Flag Interactively ..................................................... 572
Setting the Debug Flag Programmatically ............................................ 572
12
11.6
Using Error Reporting APIs in Application Code ........................................... 572
11.7
Sample Error Record .............................................................................................. 573
Target Tools .......................................................................................... 575
12.1
Introduction ............................................................................................................. 576
12.2
Kernel Shell ............................................................................................................. 577
12.2.1
C Interpreter and Command Interpreter .............................................. 577
Switching Between Interpreters ............................................................ 578
Interpreter Commands and References ............................................... 578
12.2.2
Kernel and Host Shell Differences ......................................................... 578
12.2.3
Configuring VxWorks With the Kernel Shell ....................................... 581
Required Components ............................................................................. 581
Optional Components ............................................................................. 582
12.2.4
Configuring the Kernel Shell .................................................................. 584
12.2.5
Starting the Kernel Shell .......................................................................... 585
12.2.6
Using Kernel Shell Help .......................................................................... 585
12.2.7
Using Kernel Shell Control Characters ................................................. 586
12.2.8
Kernel Shell History ................................................................................. 586
12.2.9
Defining Kernel Shell Command Aliases ............................................. 587
12.2.10 Loading and Unloading Kernel Object Modules ................................. 587
12.2.11 Debugging with the Kernel Shell ........................................................... 588
Debugging SMP Systems with the Kernel Shell .................................. 588
12.2.12 Aborting Routines Executing from the Kernel Shell ........................... 589
xxii
Contents
12.2.13 Console Login Security ............................................................................ 590
12.2.14 Using a Remote Login to the Kernel Shell ............................................ 591
Remote Login With telnet and rlogin .................................................... 591
Remote Login Security ............................................................................ 592
12.2.15 Launching a Shell Script Programmatically ......................................... 593
12.2.16 Executing Shell Commands Programmatically ................................... 593
12.2.17 Accessing Kernel Shell Data Programmatically .................................. 593
12.2.18 Using Kernel Shell Configuration Variables ........................................ 594
12.2.19 Adding Custom Commands to the Command Interpreter ............... 594
Creating A New Command .................................................................... 595
Sample Custom Commands ................................................................... 599
12.2.20 Creating a Custom Interpreter ............................................................... 599
Sample Custom Interpreter .................................................................... 602
12.3
Kernel Object-Module Loader ............................................................................. 603
12.3.1
Configuring VxWorks with the Kernel Object-Module Loader ........ 604
12.3.2
Kernel Object-Module Loader API ........................................................ 605
12.3.3
Summary List of Kernel Object-Module Loader Options .................. 606
12.3.4
Loading C++ Modules into the Kernel ................................................. 610
12.3.5
Specifying Memory Locations for Loading Objects ............................ 610
12.3.6
Guidelines and Caveats for Kernel Object-Module Loader Use ....... 611
Relocatable Object Files ...........................................................................
Linking and Reference Resolution .........................................................
Load Sequence Requirements and Caveats ..........................................
Resolving Common Symbols .................................................................
Resolving Weak Symbols ........................................................................
Stripping Symbols From Modules .........................................................
Function Calls, Relative Branches, and Load Failures ........................
Kernel Object Modules With SDA .........................................................
12.4
611
612
613
614
615
615
616
616
Kernel Symbol Tables ........................................................................................... 617
Symbol Entries .......................................................................................... 617
Symbol Updates ....................................................................................... 618
Searching the Symbol Library ................................................................ 618
xxiii
VxWorks
Kernel Programmer's Guide, 6.6
12.4.1
Configuring VxWorks with Symbol Tables ......................................... 618
Configuration for User Symbol Tables ................................................. 618
Configuration for a System Symbol Table ............................................ 619
12.4.2
Creating a Built-In System Symbol Table ............................................. 620
Generating the Symbol Information ...................................................... 620
Compiling and Linking the Symbol File ............................................... 620
12.4.3
Creating a Loadable System Symbol Table .......................................... 621
Creating the .sym File .............................................................................. 621
Loading the .sym File ............................................................................... 621
12.4.4
Using the VxWorks System Symbol Table ........................................... 622
12.4.5
Synchronizing Host and Kernel Modules List and Symbol Table .... 623
12.4.6
Creating and Using User Symbol Tables .............................................. 623
12.5
Show Routines ........................................................................................................ 624
12.6
WDB Target Agent ................................................................................................. 626
12.6.1
Configuring VxWorks with the WDB Target Agent .......................... 627
Basic WDB Configuration .......................................................................
Host-Target Communication Options ..................................................
Debugging Mode Options ......................................................................
Process Management Options ................................................................
Initialization Options ...............................................................................
Additional Options ..................................................................................
628
628
632
633
634
634
12.6.2
WDB Target Agent and VxWorks SMP ................................................ 637
12.6.3
Using the WDB Target Agent with a TIPC Network .......................... 638
Target System Configuration ................................................................. 638
Establishing a Host-Target Connection ................................................ 639
12.6.4
Scaling the WDB Target Agent ............................................................... 639
12.6.5
WDB Target Agent and Exceptions ....................................................... 640
12.6.6
Starting the WDB Target Agent Before the VxWorks Kernel ............ 640
12.6.7
Creating a Custom WDB Communication Component ..................... 642
xxiv
Contents
12.7
Common Problems ................................................................................................. 643
Kernel Shell Debugging Never Hits a Breakpoint ..............................
Insufficient Memory ................................................................................
"Relocation Does Not Fit" Error Message .............................................
Missing Symbols .......................................................................................
Kernel Object-Module Loader is Using Too Much Memory .............
Symbol Table Unavailable ......................................................................
13
643
643
644
645
645
646
C++ Development ................................................................................. 647
13.1
Introduction ............................................................................................................. 647
13.2
Configuring VxWorks for C++ ............................................................................ 648
13.3
C++ Code Requirements ....................................................................................... 649
13.4
Using C++ in Signal Handlers and ISRs .......................................................... 649
13.5
Downloadable Kernel Modules in C++ ............................................................ 650
13.6
13.5.1
Use a Single C++ Module ....................................................................... 650
13.5.2
Munching a C++ Application Module .................................................. 650
13.5.3
Calling Static Constructors and Destructors Interactively ................. 652
C++ Compiler Differences ................................................................................... 653
13.6.1
Template Instantiation ............................................................................. 654
13.6.2
Run-Time Type Information ................................................................... 656
13.7
Namespaces ............................................................................................................. 656
13.8
C++ Demo Example ............................................................................................... 657
xxv
VxWorks
Kernel Programmer's Guide, 6.6
PART II: MULTIPROCESSING TECHNOLOGIES
14
15
Overview of Multiprocessing Technologies ...................................... 661
14.1
Introduction ............................................................................................................. 661
14.2
VxWorks SMP ......................................................................................................... 662
14.3
Shared Memory Objects ........................................................................................ 662
14.4
Distributed Shared Memory ................................................................................ 662
14.5
TIPC Over Distributed Shared Memory ............................................................ 662
14.6
Message Channels .................................................................................................. 663
VxWorks SMP ....................................................................................... 665
15.1
Introduction ............................................................................................................ 666
15.2
Technology Overview .......................................................................................... 666
15.2.1
Terminology ............................................................................................. 667
15.2.2
VxWorks SMP Operating System Features ......................................... 668
15.2.3
VxWorks SMP Hardware ....................................................................... 669
15.2.4
Comparison of VxWorks SMP and AMP ............................................ 671
15.3
VxWorks SMP Configuration and Build .......................................................... 674
15.4
Booting VxWorks SMP ......................................................................................... 676
15.5
Programming for VxWorks SMP ....................................................................... 676
15.6
Spinlocks for Mutual Exclusion and Synchronization .................................. 679
15.6.1
ISR-Callable Spinlocks ............................................................................ 682
15.6.2
Task-Only Spinlocks ............................................................................... 682
15.6.3
Caveats With Regard to Spinlock Use .................................................. 683
15.6.4
Routines Restricted by Spinlock Use ..................................................... 683
xxvi
Contents
15.7
15.8
CPU-Specific Mutual Exclusion ......................................................................... 685
15.7.1
CPU-Specific Mutual Exclusion for Interrupts ................................... 685
15.7.2
CPU-Specific Mutual Exclusion for Tasks ........................................... 686
Memory Barriers ..................................................................................................... 687
Read Memory Barrier .............................................................................. 688
Write Memory Barrier ............................................................................. 689
Read/Write Memory Barrier .................................................................. 689
15.9
Atomic Memory Operations ................................................................................. 690
15.10 CPU Affinity ........................................................................................................... 691
15.10.1 Task CPU Affinity ................................................................................... 691
15.10.2 Interrupt CPU Affinity ........................................................................... 694
15.11 CPU Information and Management .................................................................. 694
CPU Information and Management Routines ..................................... 694
CPU Set Variables and Macros ............................................................... 696
15.12 Debugging SMP Code .......................................................................................... 698
15.13 Optimizing SMP Performance ............................................................................ 699
15.14 Sample Programs ................................................................................................... 702
15.15 Migrating Code to VxWorks SMP ..................................................................... 702
15.15.1 Code Migration Path ................................................................................ 703
15.15.2 Overview of Migration Issues ............................................................... 705
15.15.3 RTP Applications and SMP .................................................................... 706
15.15.4 Implicit Synchronization of Tasks ......................................................... 706
15.15.5 Synchronization and Mutual Exclusion Facilities ............................... 707
15.15.6 VxWorks SMP Variants of Uniprocessor Routines ............................. 708
cacheLib Restrictions ............................................................................... 708
vmBaseLib Restrictions ........................................................................... 709
xxvii
VxWorks
Kernel Programmer's Guide, 6.6
15.15.7 Unsupported Uniprocessor Routines and SMP Alternatives ........... 709
Interrupt Locking: intLock( ) and intUnlock( ) ....................................
Task Locking: taskLock( ) and taskUnlock( ) .......................................
Task Locking in RTPs: taskRtpLock( ) and taskRtpUnlock( ) ............
Task Variable Management: taskVarLib ..............................................
Task Local Storage: tlsLib ........................................................................
710
711
711
712
712
15.15.8 SMP CPU-Specific Variables and Uniprocessor Global Variables .... 712
SMP Per-CPU Variables .......................................................................... 713
Uniprocessor-Only Global Variables ..................................................... 714
15.15.9 Memory-Access Attributes ..................................................................... 714
15.15.10 Drivers and BSPs ...................................................................................... 715
16
Shared-Memory Objects: VxMP .......................................................... 717
16.1
Introduction ............................................................................................................. 717
16.2
Using Shared-Memory Objects ........................................................................... 718
16.2.1
Multiprocessor-Uniprocessor Portability ............................................. 719
16.2.2
Multiprocessing and Byte Ordering ...................................................... 719
16.2.3
Restrictions on Shared Memory Object Use ......................................... 720
16.2.4
Publishing Objects With the Name Database ...................................... 720
16.2.5
Shared Semaphores .................................................................................. 722
16.2.6
Shared Message Queues .......................................................................... 728
16.2.7
Shared-Memory Allocator ...................................................................... 733
Shared-Memory System Partition ..........................................................
User-Created Partitions ...........................................................................
Using the Shared-Memory System Partition ........................................
Using User-Created Partitions ...............................................................
Side Effects of Shared-Memory Partition Options ..............................
733
734
735
738
741
16.3
System Requirements ............................................................................................ 741
16.4
Performance Considerations ................................................................................ 742
xxviii
Contents
16.5
17
Configuring VxWorks for Shared Memory Objects ...................................... 744
16.5.1
Maximum Number of CPUs ................................................................... 744
16.5.2
Cache Coherency ...................................................................................... 744
16.5.3
Mailbox Interrupts and Bus Interrupts ................................................. 745
16.5.4
Shared-Memory Anchor ......................................................................... 745
16.5.5
Shared-Memory Region .......................................................................... 746
16.5.6
Numbers of Shared Memory Objects .................................................... 747
16.5.7
Dual-Port or External Memory .............................................................. 748
16.5.8
Configuration Example ........................................................................... 750
16.6
Displaying Information About Shared Memory Objects .............................. 752
16.7
Troubleshooting ..................................................................................................... 752
16.7.1
Configuration Problems .......................................................................... 753
16.7.2
Troubleshooting Techniques .................................................................. 754
Distributed Shared Memory: DSHM ................................................... 755
17.1
Introduction ............................................................................................................ 755
17.2
Technology Overview .......................................................................................... 756
17.2.1
Architecture .............................................................................................. 757
DSHM Management Service and Custom Services ............................ 758
Hardware Interface .................................................................................. 758
17.2.2
Communication Model .......................................................................... 759
Broadcasting .............................................................................................. 759
Send Operation ......................................................................................... 762
Broadcast Operation ................................................................................ 763
17.3
Configuring VxWorks for DSHM ...................................................................... 763
17.3.1
Components and Parameters ................................................................. 764
xxix
VxWorks
Kernel Programmer's Guide, 6.6
17.4
17.5
Developing Custom Services .............................................................................. 767
17.4.1
Service Numbers ..................................................................................... 768
17.4.2
DSHM Messaging Protocols and Macro Functions ........................... 769
17.4.3
DSHM Service APIs ................................................................................ 770
17.4.4
Service Code Example ............................................................................ 771
Developing a Hardware Interface ...................................................................... 777
17.5.1
Driver Initialization ................................................................................. 778
17.5.2
Callbacks .................................................................................................... 778
17.5.3
Registering with the DSHM MUX ......................................................... 782
17.5.4
Messaging Support ................................................................................. 782
Shared Memory Messaging Support ..................................................... 783
Shared Memory Lock-Less Discovery ................................................... 785
18
17.5.5
Management Service ............................................................................... 785
17.5.6
DSHM Hardware Interface APIs .......................................................... 786
Message Channels ............................................................................... 787
18.1
Introduction ............................................................................................................. 787
18.2
Message Channel Facilities .................................................................................. 789
18.3
Multi-Node Communication with TIPC ............................................................ 791
18.4
Single-Node Communication with COMP and DSI ....................................... 791
Express Messaging ................................................................................... 792
Show Routines .......................................................................................... 792
18.4.1
18.5
COMP Socket Support with DSI ............................................................ 793
Socket Name Service .............................................................................................. 795
18.5.1
Multi-Node Socket Name Service .......................................................... 796
18.5.2
snsShow( ) Example ................................................................................. 797
xxx
Contents
18.6
18.7
Socket Application Libraries ............................................................................... 799
18.6.1
SAL Server Library .................................................................................. 800
18.6.2
SAL Client Library ................................................................................... 802
Configuring VxWorks for Message Channels .................................................. 803
COMP, DSI, and SAL Components ....................................................... 803
SNS Component Options ........................................................................ 803
Show Routines .......................................................................................... 805
18.8
Comparison of Message Channels and Message Queues .............................. 806
Index .............................................................................................................. 809
xxxi
VxWorks
Kernel Programmer's Guide, 6.6
xxxii
PART I
Core Technologies
1
Overview .............................................................
3
2
Kernel ..................................................................
7
3
Boot Loader ........................................................ 131
4
Multitasking ........................................................ 157
5
POSIX Facilities .................................................. 251
6
Memory Management ......................................... 315
7
I/O System ........................................................... 359
8
Local File Systems ............................................. 451
9
Network File System: NFS ................................. 523
10
Flash File System Support: TrueFFS ................ 543
11
Error Detection and Reporting .......................... 563
12
Target Tools ........................................................ 575
13
C++ Development ............................................... 647
1
VxWorks
Kernel Programmer's Guide, 6.6
2
1
Overview
1.1 Introduction 3
1.2 Related Documentation Resources 4
1.3 VxWorks Configuration and Build 5
1.1 Introduction
This guide describes the VxWorks operating system, and how to use VxWorks
facilities in the development of real-time systems and applications. The first part,
Core Technologies, covers the following topics:
■
■
■
■
■
■
■
■
■
■
■
■
kernel facilities, kernel-based applications, and kernel customization
boot loader
multitasking facilities
POSIX facilities
memory management
I/O system
local file systems
Network File System (NFS)
flash file system support with TrueFFS
error detection and reporting
target tools, such as the kernel shell, kernel object-module loader, and target
symbol table
C++ development
3
VxWorks
Kernel Programmer's Guide, 6.6
The second part of this guide describes VxWorks multiprocessor technologies. For
an introduction to this material, see 14. Overview of Multiprocessing Technologies.
NOTE: This book provides information about facilities available in the VxWorks
kernel. For information about facilities available to real-time processes, see the
VxWorks Application Programmer’s Guide.
1.2 Related Documentation Resources
The companion volume to this book, the VxWorks Application Programmer’s Guide,
provides material specific process-based (RTP) applications and process
management.
Detailed information about VxWorks libraries and routines is provided in the
VxWorks API references. Information specific to target architectures is provided in
the VxWorks BSP references and in the VxWorks Architecture Supplement.
For information about BSP and driver development, see the VxWorks BSP
Developer’s Guide and the VxWorks Device Driver Guide.
The VxWorks networking facilities are documented in the Wind River Network
Stack for VxWorks 6 Programmer’s Guide and the VxWorks PPP Programmer’s Guide.
For information about migrating applications, BSPs, drivers, and projects from
previous versions of VxWorks and the host development environment, see the
VxWorks Migration Guide and the Wind River Workbench Migration Guide.
The Wind River IDE and command-line tools are documented in the Wind River
Workbench User’s Guide, the VxWorks Command-Line Tools User’s Guide, the Wind
River compiler and GNU compiler guides, and the Wind River tools API and
command-line references.
4
1 Overview
1.3 VxWorks Configuration and Build
1.3 VxWorks Configuration and Build
1
This document describes VxWorks features; it does not go into detail about the
mechanisms by which VxWorks-based systems and applications are configured
and built. The tools and procedures used for configuration and build are described
in the Wind River Workbench User’s Guide and the VxWorks Command-Line Tools
User’s Guide.
NOTE: In this guide, as well as in the VxWorks API references, VxWorks
components and their configuration parameters are identified by the names used
in component description files. The names take the form, for example, of
INCLUDE_FOO and NUM_FOO_FILES (for components and parameters,
respectively).
You can use these names directly to configure VxWorks using the command-line
configuration facilities.
Wind River Workbench displays descriptions of components and parameters, as
well as their names, in the Components tab of the Kernel Configuration Editor.
You can use the Find dialog to locate a component or parameter using its name or
description. To access the Find dialog from the Components tab, type CTRL+F, or
right-click and select Find.
5
VxWorks
Kernel Programmer's Guide, 6.6
6
2
Kernel
2.1 Introduction 7
2.2 Kernel Architecture 9
2.3 System Startup 13
2.4 VxWorks Configuration 14
2.5 Power Management 40
2.6 Kernel Applications 51
2.7 Custom Kernel Libraries 67
2.8 Custom VxWorks Components and CDFs 67
2.9 Custom System Calls 100
2.10 Custom Scheduler 118
2.1 Introduction
This chapter provides an overview of the VxWorks kernel architecture and
detailed discussions of those features of interest to developers who work directly
with kernel facilities. In general, kernel developers can modify and extend the
VxWorks kernel in following ways:
7
VxWorks
Kernel Programmer's Guide, 6.6
■
By reconfiguring and rebuilding VxWorks with various standard components
to suit the needs of their application development environment, as well as the
needs of their deployed products.
■
By creating kernel applications that can either be interactively downloaded
and run on a VxWorks target system, or configured to execute at boot time and
linked with the operating system image.
■
By creating custom kernel libraries that can be built into the operating system.
■
By creating custom VxWorks components—such as file systems or networking
protocols—that can be configured into VxWorks using the operating system
configuration utilities.
■
By extending the kernel system-call interface with custom APIs that should be
accessible to applications running in user space (as real-time process—RTP—
applications).
■
By creating a custom scheduler for use in place of the traditional VxWorks
scheduler or the POSIX thread scheduler.
See 2.4 VxWorks Configuration, p.14; as well as other chapters throughout this book
for information about VxWorks facilities and their use. Chapter 4. Multitasking, for
example, includes discussion of features that are available only in the kernel (such
as ISRs and watchdog timers).
Section 2.6 Kernel Applications, p.51 provides information about creating kernel
applications. For information about RTP applications, see the VxWorks Application
Programmer’s Guide: Applications and Processes.
Instructions for creating custom kernel libraries is provided in the VxWorks
Command-Line Tools User’s Guide. Only brief mention of this topic is given in this
book in 2.7 Custom Kernel Libraries, p.67.
See 2.8 Custom VxWorks Components and CDFs, p.67, 2.9 Custom System Calls, p.100,
and 2.10 Custom Scheduler, p.118 for information about extending the operating
system.
Developers can also write or port drivers and BSPs for VxWorks. These topics are
covered by other books in the VxWorks documentation set; see the VxWorks Device
Driver’s Guide and the VxWorks BSP Developer’s Guide.
8
2 Kernel
2.2 Kernel Architecture
2.2 Kernel Architecture
Historically, the VxWorks operating system provided a single memory space with
no segregation of the operating system from user applications. All tasks ran in
supervisor mode. Although this model afforded performance and flexibility when
developing applications, only skilled programming could ensure that kernel
facilities and applications coexisted in the same memory space without interfering
with one another.1
With the release of VxWorks 6.0, the operating system provides support for
real-time processes (RTPs) that includes execution of applications in user mode
and other features common to operating systems with a clear delineation between
kernel and applications. This architecture is often referred to as the process model.
VxWorks has adopted this model with a design specifically aimed to meet the
requirements of determinism and speed that are required for hard real-time
systems. (For information about VxWorks processes and developing applications
to run in processes, see VxWorks Application Programmer’s Guide: Applications and
Processes.) VxWorks 6.x provides full MMU-based protection of both kernel and
user space.
At the same time, VxWorks 6.x maintains a high level of backward compatibility
with VxWorks 5.5. Applications developed for earlier versions of VxWorks, and
designed to run in kernel space, can be migrated to VxWorks 6.x kernel space with
minimal effort (in most cases, merely re-compilation). For more information on
this topic, see the VxWorks Migration Guide.
Naturally, new applications can be designed for kernel space as well, when other
considerations outweigh the advantages of protection that executing applications
as processes affords. These considerations might include:
■
Size. The overall size of a system is smaller without components that provided
for processes and MMU support.
■
Speed. Depending on the number of system calls an application might make,
or how much I/O it is doing when running as a process in user space, it might
be faster running in the kernel.
1. The VxWorks 5.x optional product VxVMI provides write protection of text segments and
the VxWorks exception vector table, as well as an architecture-independent interface to the
CPU’s memory management unit (MMU). In addition, specialized variants of VxWorks
such as VxWorks AE and VxWorks AE653 provide memory protection, but in a manner
different from that provided in the current release.
9
2
VxWorks
Kernel Programmer's Guide, 6.6
■
Kernel-only features. Features such as watchdog timers, ISRs, and VxMP are
available only in the kernel. In some cases, however, there are alternatives for
process-based applications (POSIX timers, for example).
■
Hardware access. If the application requires direct access to hardware, it can
only do so from within the kernel.
VxWorks is flexible in terms of both the modularity of its features and its
extensibility. The operating system can be configured as a minimal kernel that
provides a task scheduler, interrupt handling, dynamic memory management,
and little else. Or, it can be configured with components for executing applications
as processes, file systems, networking, error detection and reporting, and so on.
The operating system can also be extended by adding custom components or
modules to the kernel itself (for example, for new file systems, networking
protocols, or drivers). The system call interface can then be extended by adding
custom APIs, which makes them available to process-based applications.
2.2.1 Operating System Facilities
VxWorks provides a core set of facilities that are commonly provided by the kernel
of a multitasking operating system:
■
Startup facilities for system initialization (see 2.3 System Startup, p.13).
■
Clocks and timers (see 4.19 Watchdog Timers, p.239 and 5.7 POSIX
Asynchronous I/O, p.263).
■
Exception and interrupt handling (see Exception Task, p.12, 4.6 Task Exception
Handling, p.187, 4.18 Signals, p.226, and 4.20 Interrupt Service Routines, p.241).
■
Task management (see 4.2 Tasks and Multitasking, p.160).
■
Process management (see the VxWorks Application Programmer’s Guide:
Applications and Processes).
■
A system call interface for applications executing in processes (see VxWorks
Application Programmer’s Guide: Applications and Processes and 2.9 Custom
System Calls, p.100).
■
Intertask and interprocess communication (see 4.8 Intertask and Interprocess
Communication, p.193.
■
Signals (see 4.18 Signals, p.226).
■
Resource reclamation (see VxWorks Application Programmer’s Guide:
Applications and Processes).
10
2 Kernel
2.2 Kernel Architecture
■
Memory management (see 6. Memory Management).
■
I/O system (see 7. I/O System).
■
File systems (see 8. Local File Systems).
■
NFS (see 9. Network File System: NFS).
2
In addition, the VxWorks kernel also provides:
■
The WDB target agent, which is required for using the host development tools
with VxWorks. It carries out requests transmitted from the tools (by way of the
target server) and replies with the results (see 12.6 WDB Target Agent, p.626).
■
Facilities for error detection and reporting (see 11. Error Detection and
Reporting).
■
A target-based shell for direct user interaction, with a command interpreter
and a C-language interpreter (see 12.2 Kernel Shell, p.577).
■
A specialized facilities for multi-processor intertask communication through
shared memory (see 16. Shared-Memory Objects: VxMP).
For information about basic networking facilities, see the Wind River Network Stack
for VxWorks 6 Programmer’s Guide.
2.2.2 System Tasks
Depending on its configuration, VxWorks includes a variety of system tasks,
which are always running. These are described below.
Root Task
The root task tRootTask is the first task executed by the kernel. The entry point of
the root task is usrRoot( )initializes most VxWorks facilities. It spawns such tasks
as the logging task, the exception task, the network task, and the tRlogind
daemon. Normally, the root task terminates and is deleted after all initialization
has completed. For more information tRootTask and usrRoot( ), see the VxWorks
BSP Developer’s Guide.
Logging Task
The log task, tLogTask, is used by VxWorks modules to log system messages
without having to perform I/O in the current task context. For more information,
see 7.7 Asynchronous Input/Output, p.381 and the API reference entry for logLib.
11
VxWorks
Kernel Programmer's Guide, 6.6
Exception Task
The exception task, tExcTask, supports the VxWorks exception handling package
by performing functions that cannot occur at interrupt level. It is also used for
actions that cannot be performed in the current task’s context, such as task suicide.
It must have the highest priority in the system. Do not suspend, delete, or change
the priority of this task. For more information, see the reference entry for excLib.
Network Task
The tNet0 task is the default network daemon. It handles the task-level (as
opposed to interrupt-level) processing required by the VxWorks network. For
systems that have been configured with more than one network daemon, the task
names are tNetn. The task is primarily used by network drivers. Configure
VxWorks with the INCLUDE_NET_DAEMON component to spawn the tNet0 task.
For more information on tNet0, see the Wind River Network Stack for VxWorks 6
Programmer’s Guide.
WDB Target Agent Task
The WDB target agent task, tWdbTask, is created if the target agent is set to run in
task mode. It services requests from the host tools (by way of the target server); for
information about this server, see the host development environment
documentation. Configure VxWorks with the INCLUDE_WDB component to
include the target agent. See 12.6 WDB Target Agent, p.626 for more information
about WDB.
Tasks for Optional Components
The following VxWorks system tasks are created if their components are included
in the operating system configuration.
tShellnum
If you have included the kernel shell in the VxWorks configuration, it is
spawned as a task. Any routine or task that is invoked from the kernel shell,
rather than spawned, runs in the tShellnum context.
The task name for a shell on the console is tShell0. The kernel shell is
re-entrant, and more than one shell task can run at a time (hence the number
suffix). In addition, if a user logs in remotely (using rlogin or telnet) to a
VxWorks target, the name reflects that fact as well. For example, tShellRem1.
For more information, see 12.2 Kernel Shell, p.577. Configure VxWorks with
the INCLUDE_SHELL component to include the kernel shell.
12
2 Kernel
2.3 System Startup
tRlogind
If you have included the kernel shell and the rlogin facility in the VxWorks
configuration, this daemon allows remote users to log in to VxWorks. It
accepts a remote login request from another VxWorks or host system and
spawns tRlogInTask_hexNumber and tRlogOutTask_hexNumber (for
example, tRlogInTask_5c4d0). These tasks exist as long as the remote user is
logged on. Configure VxWorks with the INCLUDE_RLOGIN component to
include the rlogin facility.
tTelnetd
If you have included the kernel shell and the telnet facility in the VxWorks
configuration, this daemon allows remote users to log in to VxWorks with
telnet. It accepts a remote login request from another VxWorks or host system
and spawns the input task tTelnetInTask_hexNumber and output task
tTelnetOutTask_hexNumber. These tasks exist as long as the remote user is
logged on. Configure VxWorks with the INCLUDE_TELNET component to
include the telnet facility.
tPortmapd
If you have included the RPC facility in the VxWorks configuration, this
daemon is RPC server that acts as a central registrar for RPC services running
on the same machine. RPC clients query the tPortmapd daemon to find out
how to contact the various servers. Configure VxWorks with the
INCLUDE_RPC component to include the portmap facility.
tJobTask
The tJobTask executes jobs—that is, function calls—on the behalf of tasks.
(The tExcTask task executes jobs on the behalf of ISRs.) It runs at priority 0
while waiting for a request, and dynamically adjusts its priority to match that
of the task that requests job execution. Configure VxWorks with the
INCLUDE_JOB_TASK component to include the job facility. For more
information see, 4.4.6 Task Deletion and Deletion Safety, p.180.
2.3 System Startup
When a VxWorks system is powered on, the boot loader copies an operating
system image into memory and directs the CPU to begin executing it. The boot
loader is most often located in ROM (although it can also be stored on a disk). The
VxWorks image can be stored on a host or network file system, as is usually the
13
2
VxWorks
Kernel Programmer's Guide, 6.6
case during development—or stored in ROM with the boot loader, as is often the
case with production units. The VxWorks boot loader is actually a scaled-down
version of VxWorks itself, whose sole purpose is to load a system image and
initiate its execution. (See 3. Boot Loader.) For more information about system
startup, see the VxWorks BSP Developer's Guide: Overview of a BSP.
2.4 VxWorks Configuration
VxWorks is a flexible, scalable operating system with numerous facilities that can
included, excluded, variously configured, and extended with customized
technologies, depending on the requirements of your applications and system, and
the stage of the development cycle.
VxWorks distributions include default system images for each supported BSP.
Each system image is a binary module that can be booted and run on a target
system. A system image consists of a set of components linked together into a
single non-relocatable object module with no unresolved external references.
The default system images are designed for the development environment. They
contain the basic set of components that are necessary to interact with the system
using host development tools. In most cases, you will find the supplied system
image adequate for initial development (provided the default drivers are
appropriate). Using a default VxWorks image, you can interactively download
and run kernel applications.
During the development cycle you may want to reconfigure and rebuild VxWorks
with components specifically selected to support your applications and
development requirements. If, for example, you configure VxWorks with the
appropriate components and initialization settings, you can link kernel
applications with VxWorks and start them automatically at boot time (see
2.6 Kernel Applications, p.51). You can also configure VxWorks with support for
process-based applications, to store them as part of the system image in ROMFS,
and to run them interactively and automatically (see VxWorks Application
Programmer’s Guide: Applications and Processes).
If the VxWorks components provided by Wind River do not provide all the
facilities required for your system, you can create custom facilities, such as new file
systems and networking protocols, and package them as components (see
2.8 Custom VxWorks Components and CDFs, p.67), add new system calls for
14
2 Kernel
2.4 VxWorks Configuration
process-based applications (see 2.9 Custom System Calls, p.100), create your own
scheduler (see 2.10 Custom Scheduler, p.118), and so on.
Finally, for production systems, you will want to reconfigure VxWorks with only
those components needed for deployed operation, and to build it as the
appropriate type of system image (see 2.4.1 VxWorks Image Types, p.15). For
production systems you will likely want to remove components required for host
development support, such as the WDB target agent and debugging components
(INCLUDE_WDB and INCLUDE_DEBUG), as well as to remove any other
operating system components not required to support your application. Other
considerations include reducing the memory requirements of the system,
speeding up boot time, and security issues.
For information about using the Workbench and command-line tools to configure
and build VxWorks, see the Wind River Workbench User’s Guide and the VxWorks
Command-Line Tools User’s Guide.
2.4.1 VxWorks Image Types
Different types of VxWorks system images can be produced for a variety of
storage, loading, and execution scenarios. Default versions of the following images
are provided in the VxWorks installation. Customized versions with different
components can also be created. Note that only one image type requires a boot
loader, and that the others are self-booting.
The various VxWorks image types, their use, and behavior are:
vxWorks
This VxWorks image type is intended for use during development and is often
referred to as downloadable. It is also useful for production systems in which the
boot loader and system image are stored on disk. In a development
environment, the image is usually stored on the host system (or a server on the
network), downloaded to the target system by the boot loader, and loaded into
RAM. The symbol table is maintained on the host (in the file vxWorks.sym),
where it is used by the host development tools. Leaving the symbol table on
the host keeps the image size down and reduces boot time. If VxWorks is
reconfigured with the INCLUDE_STANDALONE_SYM_TBL component, the
symbol table is included in the VxWorks image.
vxWorks_rom
A VxWorks image that is stored in ROM on the target. It copies itself to RAM
and then makes the processor switch execution to RAM. Because the image is
not compressed, it is larger than the other ROM-based images and therefore
15
2
VxWorks
Kernel Programmer's Guide, 6.6
has a slower startup time; but it has a faster execution time than
vxWorks_romResident.
vxWorks_romCompress
A VxWorks image that is stored in ROM on the target. It is almost entirely
compressed, but has small uncompressed portion executed by the processor
immediately after power up/reboot. This small portion is responsible for
decompressing the compressed section of the ROM image into RAM and for
making the processor switch execution to RAM. The compression of the image
allows it to be much smaller than other images. However the decompression
operation increases the boot time. It takes longer to boot than vxWorks_rom
but takes up less space than other ROM-based images. The run-time execution
is the same speed as vxWorks_rom.
vxWorks_romResident
A VxWorks image that is stored in ROM on the target. It copies only the data
segment to RAM on startup; the text segment stays in ROM. Thus it is
described as being ROM-resident. It has the fastest startup time and uses the
smallest amount of RAM, but it runs slower than the other image types
because the ROM access required for fetching instructions is slower than
fetching them from RAM. It is obviously useful for systems with constrained
memory resources.
The default VxWorks image files can be found in sub-directories under
installDir/vxworks-6.x/target/proj/projName. For example:
/home/moi/myInstallDir/vxworks-6.x/target/proj/wrSbc8260_diab/default_rom/vxWorks_rom
For many production systems it is often necessary to store a kernel application
module that is linked with VxWorks in ROM. VxWorks can be configured to
execute the application automatically at boot time. The system image can also
simply store the application module to allow for its being called by other
programs, or for interactive use by end-users (for example, diagnostic programs).
To produce a ROM-based system, you must link the module with VxWorks, and
build an image type that is suitable for ROM. See 2.6.8 Linking Kernel Application
Object Modules with VxWorks, p.64. If you wish to have the application start
automatically at boot time, you must also configure VxWorks to do so (see
2.6.10 Configuring VxWorks to Run Applications Automatically, p.66). Also see
2.6.9 Image Size Considerations, p.65.
Note that, during development, VxWorks must be configured with the WDB target
agent communication interface that is required for the type of connection used
between your host and target system (network, serial, and so on). By default, it is
configured for an Enhanced Network Driver (END) connection. For more
16
2 Kernel
2.4 VxWorks Configuration
information, see 12.6 WDB Target Agent, p.626. Also note that before you use the
host development tools such as the shell and debugger, you must start a target
server that is configured for the same mode of communication.
For information about configuring VxWorks with different operating system
facilities (components), see 2.4.3 Device Driver Selection, p.22.
If you are going to store boot image in flash, and want to user TrueFFS as well, see
10.3.6 Reserving a Region in Flash for a Boot Image, p.554.
2.4.2 VxWorks Components
A VxWorks component is the basic unit of functionality with which VxWorks can
be configured. While some components are autonomous, others may have
dependencies on other components, which must be included in the configuration
of the operating system for run-time operation. The kernel shell is an example of a
component with many dependencies. The symbol table is an example of a
component upon which other components depend (the kernel shell and module
loader; for more information, see 12. Target Tools).
The names, descriptions, and configurable features of VxWorks can be displayed
with the GUI configuration facilities in Workbench. Workbench provides facilities
for configuring VxWorks with selected components, setting component
parameters, as well as automated mechanisms for determining dependencies
between components during the configuration and build process.
The command-line operating system configuration tool—vxprj—uses the naming
convention that originated with configuration macros to identify individual
operating system components. The convention identifies components with names
that begin with INCLUDE. For example, INCLUDE_MSG_Q is the message queue
component. In addition to configuration facilities, the vxprj tool provides
associated features for listing the components included in a project, and so on.
For information about the Workbench and command-line facilities used for
configuring and building VxWorks, see the Wind River Workbench User’s Guide and
the VxWorks Command-Line Tools User’s Guide.
Textual configuration files identify components with macro names that begin with
INCLUDE, as well as with user-friendly descriptions. (For information about
configuration files, see 2.8.4 Component Description Language, p.79.)
In this book, components are identified by their macro name. The GUI
configuration facilities provide a search facility for finding individual components
in the GUI component tree based on the macro name.
17
2
VxWorks
Kernel Programmer's Guide, 6.6
Some of the commonly used VxWorks components are described in Table 2-1.
Names that end in XXX represent families of components, in which the XXX is
replaced by a suffix for individual component names. For example,
INCLUDE_CPLUS_XXX refers to a family of components that includes
INCLUDE_CPLUS_MIN and others.
Note that Table 2-1 does not include all components provided in the default
configuration of VxWorks, and that the VxWorks simulator provides more
components by default.
Table 2-1
Key VxWorks Components
Component
INCLUDE_ANSI_XXX
Default
*
Description
Various ANSI C library options
Parse boot device configuration
information
INCLUDE_BOOTLINE_INIT
INCLUDE_BOOTP
*
BOOTP support
INCLUDE_CACHE_SUPPORT
*
Cache support
INCLUDE_CPLUS
*
Bundled C++ support
INCLUDE_CPLUS_XXX
Various C++ support options
INCLUDE_DEBUG
Kernel shell debugging facilities
INCLUDE_EDR_XXX
Error detection and reporting
facilities.
INCLUDE_DOSFS
DOS-compatible file system
INCLUDE_FLOATING_POINT
*
Floating-point I/O
INCLUDE_FORMATTED_IO
*
Formatted I/O
FTP server support
INCLUDE_FTP_SERVER
INCLUDE_IO_SYSTEM
18
*
I/O system and associated interfaces.
Directory and path manipulation
APIs (for more information, see
7.2 Configuring VxWorks With I/O
Facilities, p.362)
2 Kernel
2.4 VxWorks Configuration
Table 2-1
Key VxWorks Components (cont’d)
Component
Default
Description
2
Target-resident kernel object module
loader package
INCLUDE_LOADER
INCLUDE_LOGGING
*
Logging facility
INCLUDE_MEM_MGR_BASIC
*
Core partition memory manager
INCLUDE_MEM_MGR_FULL
*
Full-featured memory manager
INCLUDE_MIB2_XXX
Various MIB-2 options
INCLUDE_MMU_BASIC
*
Bundled MMU support
INCLUDE_MSG_Q
*
Message queue support
INCLUDE_NETWORK
*
Network subsystem code
INCLUDE_NFS
Network File System (NFS)
INCLUDE_NFS_SERVER
NFS server
INCLUDE_PIPES
*
Pipe driver
INCLUDE_POSIX_XXX
Various POSIX options
INCLUDE_PROTECT_TEXT
Text segment write protection
INCLUDE_PROTECT_VEC_TABLE
Vector table write protection
INCLUDE_PROXY_CLIENT
*
Proxy ARP client support
INCLUDE_PROXY_SERVER
Proxy ARP server support
INCLUDE_RAWFS
Raw file system
INCLUDE_RLOGIN
Remote login with rlogin
INCLUDE_ROMFS
ROMFS file system
INCLUDE_RTP
Real-time process support.
INCLUDE_SCSI
SCSI support
INCLUDE_SCSI2
SCSI-2 extensions
19
VxWorks
Kernel Programmer's Guide, 6.6
Table 2-1
Key VxWorks Components (cont’d)
Component
Default
Description
INCLUDE_SECURITY
Remote login security package
INCLUDE_SELECT
Select facility
INCLUDE_SEM_BINARY
*
Binary semaphore support
INCLUDE_SEM_COUNTING
*
Counting semaphore support
INCLUDE_SEM_MUTEX
*
Mutual exclusion semaphore
support
INCLUDE_SHELL
Kernel (target) shell
INCLUDE_XXX_SHOW
Various system object show facilities
INCLUDE_SIGNALS
*
Software signal facilities
INCLUDE_SM_OBJ
Shared memory object support
(requires VxMP)
INCLUDE_SNMPD
SNMP agent
INCLUDE_SPY
Task activity monitor
INCLUDE_STDIO
*
Standard buffered I/O package
INCLUDE_SW_FP
Software floating point emulation
package
INCLUDE_SYM_TBL
Target-resident symbol table support
INCLUDE_TASK_HOOKS
*
Kernel call-out support
INCLUDE_TASK_VARS
*
Task variable support
Remote login with telnet
INCLUDE_TELNET
INCLUDE_TFTP_CLIENT
INCLUDE_TFTP_SERVER
20
*
TFTP client support
TFTP server support
2 Kernel
2.4 VxWorks Configuration
Table 2-1
Key VxWorks Components (cont’d)
Component
INCLUDE_TIMEX
Default
*
Description
2
Function execution timer
INCLUDE_TRIGGERING
Function execution timer
INCLUDE_UNLOADER
Target-resident kernel object module
unloader package
INCLUDE_VXEVENTS
VxWorks events support.
INCLUDE_WATCHDOGS
*
Watchdog support
INCLUDE_WDB
*
WDB target agent (see 12.6 WDB
Target Agent, p.626)
INCLUDE_WDB_TSFS
*
Target server file system
INCLUDE_WINDVIEW
System Viewer command server (see
the Wind River System Viewer User’s
Guide)
By default, VxWorks includes both libc and GNU libgcc, which are provided with
the INCLUDE_ALL_INTRINSICS component. If you wish to exclude one or the
other library, you can do so by reconfiguring the kernel with either
INCLUDE_DIAB_INTRINSICS or INCLUDE_GNU_INTRINSICS, respectively. Note
that these libraries are available in the kernel to enable dynamically downloading
and running kernel object modules.
21
VxWorks
Kernel Programmer's Guide, 6.6
2.4.3 Device Driver Selection
Device drivers are provided as VxWorks components that can be added to or
removed from a system using Workbench and vxprj. Some drivers are
VxBus-compliant, and others (legacy drivers) are not.
NOTE: Only VxBus-compatible drivers can be used with the symmetric
multiprocessing (SMP) configuration of VxWorks. For general information about
VxWorks SMP and about migration, see 15. VxWorks SMP and 15.15 Migrating
Code to VxWorks SMP, p.702.
Note that the component names for VxBus drivers do not have the leading
INCLUDE_ element (for example, DRV_SIO_NS16550), whereas the names for
non-VxBus drivers do (for example, INCLUDE_ELT_3C509_END).
For information about the VxBus facility, see the VxWorks Device Driver Developer’s
Guide.
2.4.4 VxWorks Configuration Profiles
In addition to components and component bundles, configuration profiles can be
used to configure a VxWorks system. Profiles provide a convenient way of
providing a base line of operating system functionality that is different from the
default configuration available with the VxWorks product installation. The
following profiles are available:
PROFILE_MINIMAL_KERNEL—Minimal VxWorks Kernel Profile
Provides the lowest level of services at which a VxWorks system can
operate. It consists of the micro-kernel, and basic CPU and BSP support.
This profile is meant to provide a very small VxWorks system that can
support multitasking and interrupt management at a very minimum, but
semaphores and watchdogs are also supported by default. (See 2.4.5 Small
VxWorks Configuration Profiles, p.24.)
PROFILE_BASIC_KERNEL—Basic VxWorks Kernel Profile
Builds on the minimal kernel profile, adding support for message queues,
task hooks, memory allocation and de-allocation, and basic I/O facilities.
Applications based on this profile can be more dynamic and feature rich
than the minimal kernel. (See 2.4.5 Small VxWorks Configuration Profiles,
p.24.)
22
2 Kernel
2.4 VxWorks Configuration
PROFILE_BASIC_OS—Basic VxWorks OS Profile
Provides a small operating system on which higher level constructs and
facilities can be built. It supports a full I/O system, file descriptors, and
related ANSI routines. It also supports task and environment variables,
signals, pipes, coprocessor management, and a ROMFS file system. (See
2.4.5 Small VxWorks Configuration Profiles, p.24.)
PROFILE_COMPATIBLE—VxWorks 5.5 Compatible Profile
Provides the minimal configuration that is compatible with VxWorks 5.5.
PROFILE_DEVELOPMENT—VxWorks Kernel Development Profile
Provides a VxWorks kernel that includes development and debugging
components.
PROFILE_ENHANCED_NET—VxWorks Enhanced Network Profile
Adds components appropriate for typical managed network client host
devices to the default profile. The primary components added are the
DHCP client and DNS resolver, the Telnet server (shell not included), and
several command-line-style configuration utilities.
PROFILE_CERT—VxWorks DO-178 Certification Profile
Provides a DO-178B Level A-certifiable API subset of the VxWorks
operating system.
PROFILE_BOOTAPP—VxWorks Boot Loader Profile
Provides a VxWorks boot loader. For more information, see
3.7 Customizing and Building Boot Loaders, p.146.
23
2
VxWorks
Kernel Programmer's Guide, 6.6
2.4.5 Small VxWorks Configuration Profiles
VxWorks can be scaled below the size of the default operating system using special
configuration profiles (sometimes referred to as source-scalable or scalable profiles).
They provide three levels of operating system functionality, starting with a
minimal kernel through a basic operating system. These profiles do not support
networking facilities. The small configuration profiles are as follows:
■
PROFILE_MINIMAL_KERNEL—The minimal kernel profile provides the
lowest level of services at which a VxWorks system can operate. It consists of
the microkernel, and basic CPU and BSP support. This profile creates a very
small VxWorks systems that can support multitasking and interrupt
management. It does not provide support for dynamic memory allocation; all
objects must be statically instantiated. For more information see Minimal
Kernel Profile, p.28.
■
PROFILE_BASIC_KERNEL—The basic kernel profile builds on the minimal
kernel profile, adding support for message queues, task hooks, dynamic
memory allocation and de-allocation, and a basic I/O system. For more
information, see Basic Kernel Profile, p.33.
■
PROFILE_BASIC_OS—The basic OS profile provides a small operating system.
It supports a full I/O system, file descriptors, and related ANSI routines. It
also supports task and environment variables, signals, pipes, coprocessor
management, and the ROMFS file system. This profile is close to a VxWorks
5.5 configuration, but without the network stack and debugging assistance
tools. It can also be viewed as a lightweight version of the default VxWorks 6.x
configuration, but without memory protection, the network stack, System
Viewer instrumentation, and debugging assistance support (the WDB target
agent). For more information, see Basic OS Profile, p.35.
Each of the small VxWorks profiles produce an operating system that is smaller
than the current default VxWorks 6.x configuration. The smallest configuration
can build an image of 100 KB or less. The profiles are not monolithic. Each can be
used as a base upon which additional functionality can be added with other
VxWorks components, using standard configuration facilities (Workbench and
vxprj). Custom facilities and applications can also be added, as with VxWorks
configurations based on components and other profiles. And, additional
measures—such as static instantiation of objects—can be used to keep the system
as lean and efficient as possible.
Figure 2-1 illustrates the relationship between the three profiles and the
functionality they provide.
24
2 Kernel
2.4 VxWorks Configuration
NOTE: The small VxWorks profiles are built from source code, so you must install
VxWorks source to use them. For this release they can only be built with the Wind
River compiler.
NOTE: For this release, the small VxWorks profiles are not available for all BSPs.
They are provided for wrSbcPowerQuiccII (for PowerPC) and integrator1136jfs
(for ARM).
How Small Configuration Profiles Create Smaller Systems
The PROFILE_MINIMAL_KERNEL, PROFILE_BASIC_KERNEL, and
PROFILE_BASIC_OS profiles all produce systems that are smaller than those that
can be created from standard VxWorks components. They can do so because they
are build from conditionally-compiled source code, which selectively eliminates
code from individual components that is not strictly required for the level of
functionality that each profile provides (as, for example, is done with taskLib).
Standard VxWorks components on the other hand, are provided as pre-compiled
libraries, which are linked into a system by way of symbolic references. This
method is convenient, and provides for quick builds, but it means that components
tend to somewhat larger and more general purpose than suits the needs of systems
with very tight requirements for size and functionary.
Small configuration profiles allow for building systems that are scaled down to
smaller sizes in a simple, predetermined manner. Each provides a subset of the
features—and in some cases elements of standard features—that are provided in
in the default VxWorks configuration. The small configuration profiles also
provide improved system performance, because they only include code that is
carefully tailored to the requirements of the system.
25
2
VxWorks
Kernel Programmer's Guide, 6.6
Figure 2-1
Small VxWorks Configuration Profiles
Basic OS
Full I/O system
select( ) support
Full-formatted I/O
ANSI stdio
ANSI stdlib
ANSI time
TTY driver
Full-featured drivers
Full-featured memory allocator
Coprocessor support
Floating-point support
Signals
Pipes
Basic Kernel
Basic I/O system
Dynamic memory allocation/deallocation
Dynamic object creation/deletion
Message queues
Exception task
Task hooks
Task restart capability
Minimal Kernel
Interrupts
System clock
Simple drivers
Cache support
ANSI ctype, string
Static allocation of
system resources
26
Multitasking
Watchdogs
Semaphores
Exception handling
Events
2 Kernel
2.4 VxWorks Configuration
Configuring and Building Small VxWorks Configurations
The small VxWorks profiles are built from source code, so you must install
VxWorks source to use them. You must also create a project using the appropriate
configuration profile and select the option for building from source. For the vxprj
command-line tool, use the -source option; for Workbench, select the source build
option during project creation.
Each small VxWorks profile has a predefined set of components that belong to it.
If you choose to change the default configuration of a profile, you should be aware
of the following guidelines and behavior:
!
■
If components are not needed, they can be removed.
■
If a component that is used in a higher-level profile is added, the profile is
automatically elevated to the next level that uses the component. For example,
if you add a message queue component to the PROFILE_MINIMAL_KERNEL
profile the project is automatically expanded to the PROFILE_BASIC_KERNEL
profile, with all the components provided by that profile.
■
If you add components that are intended to be used with any of the (source)
small profiles—such as networking components—the project defaults to using
the standard binary components. In effect, the advantage of using a small
VxWorks profile is lost.
CAUTION: In order to maintain a small VxWorks configuration profile for a
project, you must be careful with adding other components. Adding components
that are not part of one of these profiles causes the project to use the default binary
components, and produces a larger system with different functionality (even on a
per-component basis) than the profile you started with.
Optimization of Small VxWorks Profile Systems
Systems produced with small VxWorks profiles can be further optimized when
code is developed using the following methods:
■
Static instantiation of kernel objects—which is required for
PROFILE_MINIMAL_KERNEL applications—can be useful with the other
profiles as well. For information in this regard, see 2.6.4 Static Instantiation of
Kernel Objects, p.56.
27
2
VxWorks
Kernel Programmer's Guide, 6.6
■
Using the VX_GLOBAL_NO_STACK_FILL configuration parameter disables
stack filling on all tasks in the system (for information about the parameter, see
Filling Task Stacks, p.175).
Minimal Kernel Profile
The minimal kernel profile (PROFILE_MINIMAL_KERNEL) provides the most basic
level at which a VxWorks system can operate. This profile consists of the
micro-kernel and basic support code for the CPU and BSP.
The profile provides multitasking and interrupt management at a very minimum.
Support for semaphores (binary, counting, and mutexes) and watchdogs is also
present, but is optional (that is, it can be removed). A subset of the ANSI C library
is available in the minimal kernel profile. These include support for ctype, string
manipulation, and routines converting numbers to strings and vice versa. The
minimal kernel profile does not provide an I/O system or mechanisms for
dynamic memory allocation.
The profile provides the foundation for small statically-resourced kernel
applications, such as small fixed-function controller types of systems that are
traditionally implemented using micro-controllers.
Statically-resourced systems are systems in which all application and kernel
entities are pre-allocated by the compiler at system build time. There are no
memory allocations done at run-time. Therefore minimal kernel systems spend
less time initializing themselves, and have no chance of resource unavailability.
There is a much higher level of determinism in these systems because resources are
guaranteed to be available as soon as the system boots. All the application need do
is to initialize the storage appropriately before it is ready for business. Static
instantiation of kernel objects are explained in detail in 2.6.4 Static Instantiation of
Kernel Objects, p.56.
Static instantiation of system resources does have its own shortcomings. It is not
suitable for systems that are inherently dynamic in nature, that is, ones that have
widely differing amounts of loads, or those that must instantiate and destroy
objects on demand. Statically-resourced systems therefore are not suitable for all
types of applications.
Static Memory Allocation
A unique characteristic of the minimal kernel profile is that it requires static
memory allocation; that is, all memory allocation must occur at compile time.
There is no support for dynamic memory allocation (that is, malloc( ), free( ), and
28
2 Kernel
2.4 VxWorks Configuration
related routines are not supported). Both system and application components are
expected to declare storage for all objects at compile time (that is, statically). The
absence of dynamic allocation in this profile also implies that the kernel cannot
dynamically instantiate objects like tasks, semaphores and watchdogs. APIs such
as taskSpawn( ), taskDelete( ), semXCreate( ), semDelete( ), and so on, are not
available. However the same kind of objects can be instantiated statically.
Unsupported Facilities
The minimal kernel profile is a very small, limited environment that is suitable
only for small, fixed-function systems. Consequently, it is also a very limited
programming environment. Significant capabilities that are absent from a minimal
kernel, which would otherwise be present in more feature rich configurations of
VxWorks. These features are as follows:
■
■
■
■
■
■
■
■
■
■
■
■
Dynamic memory allocation and de-allocation. The ability to destroy or
terminate kernel objects.
Memory protection (that is, detection of null pointer and other invalid
memory accesses).
Support for task hooks (that is, taskHookLib).
Floating point and other coprocessor support.
I/O system.
Signals.
Processes (RTPs).
ISR Objects (that is, isrLib).
WDB target agent.
Networking.
C++ support.
System Viewer logging.
Device Drivers for Minimal Kernel Systems
Since there is no I/O system in this profile, there is no support for traditional
device access APIs like open( ), close( ), read( ), write( ), and so on. Device drivers
written for such systems must be standalone programs that manage devices
directly. DMA-safe buffers (if needed) must be allocated at compile time. Since
there is no support for malloc( ) or free( ), there is correspondingly no support for
cacheDmaMalloc( ) or cacheDmaFree( ) either.
Formatted Character String Output
The INCLUDE_FORMATTED_OUT_BASIC component supplies capability for
formatted output with routines like printf( ), sprintf( ), snprintf( ), printfExc( ),
and so on. In the minimal and basic kernel profiles, these routines are limited to
29
2
VxWorks
Kernel Programmer's Guide, 6.6
outputting only integer values and strings. The supported formats are %d, %s, %c,
%x, %X, %u, and %o only. Floating point, 64-bit, or vector type formats are not
supported with this profile.
Because there is no I/O system in the minimal kernel profile, there are no file
descriptors, and the assumption that printf( ) output is sent to file descriptor 1 is
not true for this profile. The printf( ) routine works for the formats described
above, but its output is sent to the console device through a dedicated function.
Do not attempt to make use of file descriptors, standard output, or standard error
with this profile, because they do not operate in the standard manner.
Minimal Kernel Profile Components and Libraries
The components and libraries that make up the VxWorks minimal kernel profile
are listed in Table 2-2.
.
Table 2-2
Minimal Kernel Profile Components
Component
Library
Error detection and reporting stub.
INCLUDE_EDR_STUB
INCLUDE_ANSI_CTYPE
Description
ansiCtype
ANSI ctype routines like isalpha( ),
iscntrl( ), isdigit( ), and so on.
ANSI bsearch( ) routine
INCLUDE_ANSI_BSEARCH
INCLUDE_ANSI_STDLIB_NUMBERS
ansiStdlib
ANSI stdlib string-number
conversion routines.
INCLUDE_ANSI_STRING
ansiString
Full set of ANSI string routines.
ANSI abs( ) routine.
INCLUDE_ANSI_ABS
INCLUDE_ANSI_MEMCHR
ansiString
ANSI memchr( ) routine.
INCLUDE_ANSI_MEMCPY
ansiString
ANSI memcpy( ) routine.
INCLUDE_ANSI_MEMSET
ansiString
ANSI memset( ) routine.
INCLUDE_ANSI_MEMCMP
ansiString
ANSI memcmp( ) routine.
INCLUDE_ANSI_MEMMOVE
ansiString
ANSI memmove( ) routine.
INCLUDE_ANSI_STRCAT
ansiString
ANSI strcat( ) routine.
INCLUDE_ANSI_STRNCAT
ansiString
ANSI strncat( ) routine.
30
2 Kernel
2.4 VxWorks Configuration
Table 2-2
Minimal Kernel Profile Components
Component
Library
Description
INCLUDE_ANSI_STRCMP
ansiString
ANSI strcmp( ) routine.
INCLUDE_ANSI_STRNCMP
ansiString
ANSI strncmp( ) routine.
INCLUDE_ANSI_STRCPY
ansiString
ANSI strcpy( ) routine.
INCLUDE_ANSI_STRNCPY
ansiString
ANSI strncpy( ) routine.
INCLUDE_ANSI_STRLEN
ansiString
ANSI strlen( ) routine.
INCLUDE_REBOOT_HOOKS
rebootLib
Support for reboot hooks; that is
rebootHookAdd( ).
INCLUDE_VXEVENTS
eventLib,
semEvLib,
msgQEvLib
VxWorks events support.
INCLUDE_SEM_BINARY
semBLib
Support for binary semaphores.
INCLUDE_SEM_MUTEX
semMLib
Support for mutex semaphores.
INCLUDE_SEM_COUNTING
semCLib
Support for counting semaphores.
INCLUDE_TASK_UTIL
taskUtilLib
Programmatic interface for
modifying task information.
INCLUDE_WATCHDOGS
wdLib
Support for watchdog timers.
INCLUDE_HOOKS
hookLib
Hook routine table support.
VxWorks traditional scheduler
(priority-based preemptive
scheduling).
INCLUDE_VX_TRADITIONAL_SCHEDULER
INCLUDE_FORMATTED_OUT_BASIC
2
fioLib
Support for printf( ), sprintf( ),
snprintf( ), oprintf( ), and
printErr( ) only.
No support for scanf( ) and its
variants. No support for vprintf( )
or its variants. No support for
floating point or vector formats.
31
VxWorks
Kernel Programmer's Guide, 6.6
Table 2-2
Minimal Kernel Profile Components
Component
Library
INCLUDE_BOOT_LINE_INIT
bootParseLib Parse boot device configuration
information
32
Description
2 Kernel
2.4 VxWorks Configuration
Basic Kernel Profile
The level above the minimal kernel is provided by the basic kernel profile
(PROFILE_BASIC_KERNEL). The basic kernel profile produces small VxWorks
systems that build on the minimal kernel to provide support for moderately
complex applications. Systems based on the basic kernel profile are still not much
more than a kernel But in addition to the a minimal kernel system, the basic kernel
profile offers support for the following facilities:
■
■
■
■
■
■
Basic I/O system
Inter-task communication using message queues.
Support for task hooks.
Memory allocation and free (using memPartLib).
Ability to dynamically create and delete kernel objects such as tasks,
semaphores, watchdogs and message queues (enabled by memPartLib).
Support for ANSI string routine strdup( ), which relies on malloc( ).
The most notable additions to this profile are support for basic I/O facilities,
support for message queues and task hooks, and support for memory allocation
and de-allocation. This allows applications based on this profile to be more
dynamic and feature-rich than the minimal kernel. What this profile provides,
however, is still a kernel and not an operating system. It has a full complement of
intertask communications mechanisms and other kernel features, but does not
have operating system capabilities such as memory protection, file system
support, or higher-level constructs such as pipes and so on.
Device Drivers for Basic Kernel Systems
Like the minimal kernel profile, there is no I/O system present in the basic kernel
profile. Hence device drivers for such systems must be standalone programs,
managing devices directly. Since malloc( ) and free( ) are supported,
cacheDmaMalloc( ) and cacheDmaFree( ) are available starting with this profile.
Formatted Character String Output
The very same limitations on formatted output apply to the basic kernel profile, as
are present for the minimal kernel profile. See Formatted Character String Output,
p.29.
Basic Kernel Profile Components and Libraries
In addition to the components and libraries provided by the minimal kernel profile
(listed in Table 2-2), the basic kernel profile provides those listed in Table 2-3.
33
2
VxWorks
Kernel Programmer's Guide, 6.6
Table 2-3
Basic Kernel Profile Components—in Addition to Minimal Kernel Profile
Component
Library
Description
INCLUDE_ANSI_STRDUP
ansiString
ANSI strdup( ) routine.
INCLUDE_TASK_CREATE_DELETE
taskLib
Support for taskSpawn( ),
taskCreate( ), taskDelete( ) and
exit( ).
INCLUDE_TASK_RESTART
taskLib
Support for taskRestart( ).
INCLUDE_EXC_TASK
excLib
Support for excJobAdd( ).
INCLUDE_ISR_OBJECTS
isrLib
Interrupt service routine objects
library.
INCLUDE_MSG_Q
msgQLib
Message queue support with
msgQInitialize( ),
msgQReceive( ), msgQSend( ) and
so on.
INCLUDE_MSG_Q_CREATE_DELETE
msgQLib
Message queue creation and
deletion support with
msgQCreate( ) and msgQDelete( ).
Support for msgQInfoGet( ).
INCLUDE_MSG_Q_INFO
INCLUDE_SEM_DELETE
semLib
Support for semaphore deletion
with semDelete( ).
INCLUDE_SEM_BINARY_CREATE
semBLib
Support for semBCreate( ).
INCLUDE_SEM_COUNTING_CREATE
semCLib
Support for semCCreate( ).
INCLUDE_SEM_MUTEX_CREATE
semMLib
Support for semMCreate( ).
INCLUDE_SEM_INFO
Support for semInfo( ).
INCLUDE_TASK_INFO
Support for taskInfoGet( ).
INCLUDE_TASK_HOOKS
taskHookLib Support for adding/removing
hook routines at task creation,
deletion and task switches.
INCLUDE_WATCHDOGS_CREATE_DELETE wdLib
34
Support for wdCreate( ) and
wdDelete( ).
2 Kernel
2.4 VxWorks Configuration
Table 2-3
Basic Kernel Profile Components—in Addition to Minimal Kernel Profile
Component
Library
INCLUDE_MEM_MGR_BASIC
memPartLib Memory partition manager;
malloc( ), free( ) realloc( ).
INCLUDE_HASH
hashLib
Hash table management library.
INCLUDE_LSTLIB
lstLib
Doubly linked list subroutine
library.
INCLUDE_RNG_BUF
rngLib
Ring buffer management library.
INCLUDE_POOL
poolLib
Memory pool management library.
INCLUDE_IO_BASIC
Description
2
Basic I/O functionality.
Basic OS Profile
The basic OS profile (PROFILE_BASIC_OS) builds upon the basic kernel profile to
offer a relatively simple real-time operating system. It does not, for example,
provide support for networking or real-time processes (RTPs). This configuration
is similar to a VxWorks 5.5 configuration, but without the network stack and
debugging assistance tools (the WDB target agent). The profile provides an
operating system instead of simply a kernel. The new capabilities added in this
profile are the following:
■
■
■
■
■
■
■
■
■
■
Full I/O system, which includes file system and POSIX support.
Standard I/O file descriptors and associated API support.
APIs for directory and path manipulations, and disk utilities.
Support for select( ).
TTY and pipe driver support.
Support for logging (logLib).
Support for task and environment variables (envLib, taskVarLib).
Support for coprocessor management (coprocLib) and floating point.
Full-featured memory partition manager (memLib).
Full ANSI library support. Adds support for assert( ), setjmp( ) and
longjmp( ), stdio, stdlib, and time library routines.
Device Drivers for Basic OS Systems
Device drivers for the basic OS can now use additional IO system features,
associated capabilities like select( ), and so on. File descriptor based I/O and
35
VxWorks
Kernel Programmer's Guide, 6.6
associated APIs are available. Another major addition is coprocessor support,
which typically provides support for hardware floating point operations. Vector
operations (for some PowerPC BSPs) are also available with the coprocessor
support infrastructure (coprocLib). More advanced ANSI routines are available,
that use the standard I/O system, ANSI time facilities and mathematical routines.
Formatted Character String Output
Full ANSI formatted I/O routines are available starting with the basic OS profile.
Formatted output routines like printf( ), sprintf( ), and so on can handle floating
point or vector types if applicable. Formatted input routines such as scanf( ) and
so on are also available. These routines send their output to the standard I/O file
descriptors as expected. These capabilities are available with the
INCLUDE_FORMATTED_IO component.
Basic OS Profile Components and Libraries
In addition to the components and libraries provided by the minimal kernel profile
and basic kernel profile (listed in Table 2-2 and Table 2-3), the basic OS profile
provides those listed in Table 2-4.
36
2 Kernel
2.4 VxWorks Configuration
Table 2-4
Basic OS Profile Components—in Addition to the Minimal and Basic Kernel Profiles
2
Component
Library
Description
INCLUDE_ANSI_ASSERT
ansiAssert
ANSI assert( ) routine.
INCLUDE_ANSI_LOCALE
ansiLocale
ANSI locale routines localeconv( ) and
setlocale( ).
INCLUDE_ANSI_LONGJMP
ansiSetjmp
ANSI setjmp( ) and longjmp( ) routines.
INCLUDE_ANSI_MATH
ansiMath
ANSI math routines.
INCLUDE_ANSI_STDIO
ansiStdio
ANSI stdio routines.
INCLUDE_ANSI_STDLIB
ansiStdlib
ANSI stdlib routines.
INCLUDE_ANSI_ABORT
ansiStdlib
ANSI abort( ) routine.
INCLUDE_ANSI_TIME
ansiTime
ANSI time routines
INCLUDE_ANSI_STRERROR
ansiStdio
ANSI strerror( ) routine.
INCLUDE_POSIX_CLOCKS
clockLib
POSIX clock library support for
clock_getres( ), clock_setres( ),
clock_gettime( ) and clock_settime( ).
INCLUDE_ENV_VARS
envLib
Environment variable library; getenv( ),
setenv( ) and so on.
INCLUDE_TASK_VARS
taskVarLib
Task variables support library; taskVarAdd( ),
taskVarDelete( ), taskVarGet( ), taskVarSet( )
and so on.
INCLUDE_SIGNALS
sigLib
Software signal library; support for signal( ),
kill( ) and so on.
Error detection and reporting persistent
memory region manager.
INCLUDE_EDR_PM
INCLUDE_TTY_DEV
ttyLib
TTY device driver.
INCLUDE_FLOATING_POINT
floatLib
Floating point scanning and formatting library.
INCLUDE_FORMATTED_IO
fioLib
Full formatted I/O support; printf( ), scanf( ),
and variants.
37
VxWorks
Kernel Programmer's Guide, 6.6
Table 2-4
Basic OS Profile Components—in Addition to the Minimal and Basic Kernel Profiles
Component
Library
Description
INCLUDE_POSIX_FS
fsPxLib
POSIX APIs for file systems.
INCLUDE_IO_SYSTEM
ioLib,
iosLib dirLib,
pathLib
I/O system and associated interfaces. Directory
and path manipulation API’s. For more
information, see 7.2 Configuring VxWorks With
I/O Facilities, p.362).
INCLUDE_LOGGING
logLib
Message logging support; logMsg( ),
logFdSet( ) and so on.
INCLUDE_MEM_MGR_FULL
memLib
Support for calloc( ), valloc( ), realloc( ) and so
on.
INCLUDE_PIPES
pipeDrv
Pipe device support.
INCLUDE_TYLIB
tyLib
TTY driver support library.
ROMFS (read-only memory based file system ).
INCLUDE_ROMFS
INCLUDE_SELECT
selectLib
Support for select( ) and associated API’s.
INCLUDE_STDIO
stdioLib
Support for stdioFp( ).
38
2 Kernel
2.4 VxWorks Configuration
2.4.6 Customizing VxWorks Code
VxWorks operating system code can itself be customized. This section introduces
customization of usrClock( ), hardware initialization, and more general features of
the operating system.
System Clock Modification
During system initialization at boot time, the system clock ISR—usrClock( )—is
attached to the system clock timer interrupt. For every system clock interrupt,
usrClock( ) is called to update the system tick counter and to run the scheduler.
You can add application-specific processing to usrClock( ). However, you should
keep in mind that this is executed in interrupt context, so only limited functions
can be safely called. See 4.20.5 Special Limitations of ISRs, p.245 for a list of routines
that can be safely used in interrupt context.
Long power management, if used, allows the processor to sleep for multiple ticks.
See 2.5 Power Management, p.40. The usrClock( ) routine, and therefore
tickAnnounce( ), is not called while the processor is sleeping. Instead, usrClock( )
is called only once, after the processor wakes, if at least one tick has expired.
Application code in usrClock( ) must verify the tick counter each time it is called,
to avoid losing time due to setting the processor into sleep mode.
Hardware Initialization Customization
When the application requires custom hardware, or when the application requires
custom initialization of the existing hardware, the BSP must be modified to
perform the initialization as required. What BSP modifications are required
depend on the type of hardware and the type of initialization that must be
performed. For information about adding or customizing device drivers, the
VxWorks Device Driver Developer’s Guide, specifically the introductory sections. For
information about custom modifications to BSP code, see the VxWorks BSP
Developer’s Guide.
Other Customization
The directory installDir/vxworks-6.x/target/src/usr contains the source code for
certain portions of VxWorks that you may wish to customize. For example,
39
2
VxWorks
Kernel Programmer's Guide, 6.6
usrLib.c is a common place to add target-resident routines that provide
application-specific development aids.
If you modify one of these files, an extra step is necessary before rebuilding your
VxWorks image: you must replace the modified object code in the appropriate
VxWorks archive. The makefile in installDir/vxworks-6.x/target/src/usr automates
the details. This directory is not specific to a single architecture, however, so you
must specify the value of the CPU variable with make on the command line, as
follows:
c:\installDir\vxworks-6.x\target\src\usr> make CPU=cputype TOOL=tool
This step recompiles all modified files in the directory, and replaces the
corresponding object code in the appropriate architecture-dependent directory.
After that, the next time you rebuild VxWorks, the resulting system image includes
your modified code.
The following example illustrates replacing usrLib with a modified version,
rebuilding the archives, and then rebuilding the VxWorks system image. For the
sake of conciseness, the make output is not shown. The example assumes the
pcPentium BSP; replace the BSP directory name and CPU value as appropriate for
your environment.
c:\> cd installDir\vxworks-6.1\target\src\usr
c:\installDir\vxworks-6.1\target\src\usr> copy usrLib.c usrLib.c.orig
c:\installDir\vxworks-6.1\target\src\usr> copy develDir\usrLib.c usrLib.c
c:\installDir\vxworks-6.1\target\src\usr> make CPU=PENTIUM TOOL=diab
...
c:\installDir\vxworks-6.1\target\src\usr> cd nstallDir\vxworks-6.1\target\config\pcPentium
c:\installDir\vxworks-6.1\target\config\pcPentium2> make
...
2.5 Power Management
Starting with the VxWorks 6.2 release, enhanced power management facilities are
provided for the Intel Architecture (IA). Facilities provided in earlier releases for
other architectures remain the same. The new facilities will be provided for the
other architectures in future releases. See 2.5.1 Power Management for IA
Architecture, p.41 and 2.5.2 Power Management for Other Architectures, p.49.
40
2 Kernel
2.5 Power Management
2.5.1 Power Management for IA Architecture
VxWorks power management facilities provide for managing the power and
performance states of the CPU. These facilities can be used to control CPU power
use based on the following:
■
■
■
CPU utilization
CPU temperature thresholds
task and ISR-specific performance states
The VxWorks power management facilities utilize key concepts of the Advanced
Configuration and Power Interface (ACPI) Specification, version 3.0. The ACPI
specification has not been implemented for VxWorks because of its unsuitability
for hard real-time systems and for all the architectures supported by VxWorks.
However, the ACPI specification provides useful definitions of power states and
power-state transitions, as well as of thermal management, and these definitions
have been incorporated into the design of the VxWorks power management
facilities.
ACPI Processor Power and Performance States
The ACPI 3.0 specification defines processor power states as well as the transitions
that take the processor from one state to the other. Essentially it defines the
processor power state machine. This aspect of the specification enables the
mapping of the power management features of a CPU to one of the defined states,
whether the CPU is ACPI compliant or not. For example, ACPI defines that in
power state C1, the processor does not execute instructions but the contents of the
caches remain valid (bus snooping still takes place). Many CPUs support such a
power state, but manufacturers often use different names to identify that state.
In addition to defining processor power states, ACPI defines performance states
where the CPU executes instructions, but not necessarily at its maximum
throughput. These states correspond to voltage and or frequency scaling
capabilities present on some CPUs, and provide a power management scheme
with which power can be managed even if the system is not idle.
Figure 2-2 illustrates the ACPI-defined power states that apply to the CPU power
management for VxWorks.
41
2
VxWorks
Kernel Programmer's Guide, 6.6
Figure 2-2
Fig 2-2: ACPI Power States
G0 Global System State
C0 Processor Power State
Increased
Latency,
Increased
Power
Savings
P0 Processor Performance State
C1 Processor Power State
Decreased
Performance,
Increased
Power
Savings
P1 Processor Performance State
Cn Processor Power State
Pn Processor Performance State
The G0 global system state is also known as the working state. ACPI requires that
all processor power states reside under the G0 state, which is why other G states
are not deemed relevant to this feature. The C0 to Cn states are processor power
states. Key features of these processor power are as follows:
■
In the C0 state the processor is fetching and executing instructions
■
In the C1 to Cn states the processor is not executing instructions.
■
The higher the power state number, the greater the power saving, but at the
cost greater latency in reaction to external events.
■
State transitions occur to and from the C0 state. For example, after going from
the C0 state to the C1 state the processor must transition back to the C0 state
before going to the C2 state. This is because transitions are triggered by
software and only the C0 state is allowed to execute instructions.
Under the C0 power state reside the processor performance states. In each of these
states the CPU is executing instructions but the performance and power savings in
each P-state vary. The higher the performance state number, the greater the power
saving, but the slower the rate of instruction execution. Taking the Speedstep
technology of the Pentium processor as an example, it defines various
voltage-frequency power settings that are mapped to the ACPI-defined P-states.
Note that unlike the C-state transitions, P-state transitions can occur between any
two states.
42
2 Kernel
2.5 Power Management
See the VxWorks Architecture Supplement for information about which states are
supported
2
ACPI Thermal Management
Much like the processor power management concepts, the thermal management
concepts defined in ACPI 3.0 are applicable to non ACPI-compliant hardware.
Some of these concepts are:
■
Differentiation between active cooling and passive cooling. Actively cooling a
system consists of activating a cooling device such as a fan. Passive cooling is
achieved by reducing the power consumed by a device. In this case device
includes processors and therefore this is relevant to this feature.
■
Critical Shut Down. It is the temperature threshold at which a device or system
is shut down so as to protect it from heat induced damages.
■
Notification of temperature changes. This allows a power management entity
to actively or passively manage the temperature of a system or device without
the need to poll the devices to obtain their operating temperature.
■
Processor Throttling. This is the link between thermal management and the
processor power states. ACPI equations define how to manage the
performance states of a processor so as to attempt to keep it inside a
temperature zone.
VxWorks Power Management Facilities
The architecture of the VxWorks power management facilities is composed of two
basic elements: a power management framework and a power manager. The
power management framework is effectively a layer between the power manager
and the CPU. It transfers status information from the CPU to the power manager,
and executes control of the CPU based on instructions from the power manager.
Figure 2-3 illustrates this relationship.
43
VxWorks
Kernel Programmer's Guide, 6.6
Figure 2-3
Fig 2-3: Power Management Architecture
Power Manager
events
status
control
Power Management Framework
events
status
control
CPU
The power management framework is designed to accommodate two use cases for
controlling power consumption: one minimizes the power level of the CPU based
on how much work it has to do (and its temperature); the other runs tasks and ISRs
at different performance states based on their priority.
Wind River provides the power management framework and a two power
managers—only one of which can be used at a time (see Wind River Power
Managers, p.47).
You can develop your own power manager for either of the two use cases
supported by the power management framework (see Power Management
Framework and Use Cases, p.44). The APIs provided by the power management
framework give you the control mechanisms for a power manager (see Power
Management Framework API, p.48).
Power Management Framework and Use Cases
The VxWorks power management framework is designed to serve two use cases:
one bases the control of power consumption on how much work must be done;
and the other bases the control of power consumption on task-specific
performance states.
One use case involves controlling the power consumed by the CPU based on how
much work the CPU has to do. The idea is to keep the power level of the CPU as
low as possible while preventing the system from going into overload. That is,
prevent running at 100% CPU utilization for a user-defined period of time. It is
44
2 Kernel
2.5 Power Management
quite clear that the writers of the ACPI specification had this use case in mind
while writing the document.
The second use case is based on the premise that power can be controlled by
having tasks and ISRs execute at different performance states (P-states). For
example, a task that performs work queued up by interrupts may need to run at
the P0 performance state (highest performance) while a maintenance task with no
hard deadlines can afford to run at the lowest performance state.
The first use case is more global in nature in that the entire system is running in a
certain power state. It is also a scenario in which the power consumption is
dynamically adapted to the work that is required of the processor. One of the
drawback of the approach however is that it makes it difficult to guarantee
deadlines can be met, as a piece of code is not guaranteed to run in the same
performance state on every invocation. The second use case provides a finer
granularity of control and can be more deterministic, since each task can be set to
run in the same performance state at all times. This comes at the price of increased
context switching and interrupt handling times.
NOTE: While the power management framework allows for various power
management methods to be used, the power manager itself must be designed to
ensure that it uses the capacities of the framework in a coherent manner. For
example, the framework cannot prevent contention for the CPU if both of the use
cases described above are implemented at the same time. For the same reason, only
one power manager should be included in the configuration of VxWorks (the two
power managers provided by Wind River are mutually exclusive of one another).
CPU Utilization Based Power Management
A CPU utilization based power manager is one that uses CPU utilization and CPU
temperature to control the power consumption of the CPU. There are really two
aspects to this approach. One is to transition the CPU from the C0 power state
(executing state) to one of the other C-states (non-executing states) when the
VxWorks kernel becomes idle. The other aspect is to control the performance state
(P-state) of the CPU so as to keep it inside a specified range of CPU utilization and,
optionally, inside a temperature range. In order to support a power manager using
this approach, the power management framework has the following features:
■
The framework notifies the power manager when the VxWorks kernel goes
idle.
■
The framework notifies the power manager when the VxWorks kernel comes
out of idle state.
45
2
VxWorks
Kernel Programmer's Guide, 6.6
■
The framework allows the power manager to transition the CPU from the C0
state to any of the non-executing power states: C1, C2, ...Cn. Note that the
transition back to the C0 state occurs when an external event takes place (that
is, an interrupt) and therefore this is not a state transition the framework can
allow the power manager to perform/control.
■
The framework allows the power manager to transition the CPU between
performance states (P-states) based on the CPU utilization over a user-defined
time interval. This is achieved by the framework keeping track of CPU
utilization and reporting that figure to the power manager.
■
The framework computes the CPU utilization over two user-specified time
intervals. Having two intervals makes it easier for the power manager to
implement a quick ramp up, slow ramp down policy through the performance
states. The sampling intervals can be modified dynamically.
■
The framework notifies the power manager when a CPU-utilization interval
has elapsed and provides the CPU utilization figure to the power manager at
that time.
■
The framework allows the power manager to specify a high and a low
temperature threshold for the purpose of being notified whenever the
temperature of the CPU crosses either threshold. These thresholds can be
modified dynamically. The purpose for these is to allow the power manager to
implement a cooling policy such as reducing the CPU performance state to
lower power consumption, hence lowering temperature.
The full-featured CPU utilization power manager provided by Wind River is an
example of this type of power management. See Wind River Power Managers, p.47.
Task Performance-Based Power Management
The per-task performance power manager is based on the premise that power can
be controlled by having tasks execute at different performance states (P-states). For
example, a task that performs work queued up by interrupts may need to run at
the P0 performance state (highest performance) while a maintenance task with no
hard deadlines can afford to run at the lowest performance state. In order to
support a power manager using this approach, the power management
framework has the following features:
■
The framework allows a performance state (P-state) to be assigned to each task
and allows that state to be set during context switches.
■
The framework allows a single performance state to be assigned for all
interrupts in the system so that execution of ISRs can be performed in a
performance state other than the one of the interrupted task.
46
2 Kernel
2.5 Power Management
Wind River Power Managers
Wind River provides two power managers that implement CPU-utilization-based
power management:
■
A full-featured CPU-utilization-based power manager. It manages the C-state
of the CPU when VxWorks goes idle, as well as managing the P-state of the
CPU based on CPU utilization. See CPU Utilization Based Power Management,
p.45.
■
A light version of a CPU-utilization-based power manager, which simply
manages the C-state of the CPU when VxWorks goes idle. It implements the
same power management algorithm that was provided for VxWorks 6.0 and
6.1; and is included in the default configuration of VxWorks configuration so
that the power management behavior of the operating system is the same as in
versions prior to 6.2. See 2.5.2 Power Management for Other Architectures, p.49
for more information about its features.
The components used to configure VxWorks with these power managers are listed
in Configuring VxWorks With Power Management Facilities, p.49.
47
2
VxWorks
Kernel Programmer's Guide, 6.6
Power Management Framework API
Table 2-5 describes the API provided by the power management framework.
Power managers use this API to plug into the framework. The routines are
available only in the kernel.
Table 2-5
Power Management Framework API
Routine
Description
cpuPwrCStateSet( )
Sets the CPU in a specified power state (C-state).
cpuPwrPStateSet( )
Sets the CPU in a specified performance state
(P-state).
cpuPwrPStateGet( )
Returns the performance (P-state) state of the
CPU.
cpuPwrTaskPStateSet( )
Sets the performance state (P-state) of a task.
cpuPwrTaskPStateGet( )
Gets the performance state (P-state) of a task.
cpuPwrTempThreshSet( )
Sets the CPU temperature thresholds for the CPU
(high and low).
cpuPwrTempThreshGet( )
Returns the temperature thresholds for the CPU.
cpuPwrUtilPeriodSet( )
Sets the two time intervals over which CPU
utilization is computed.
cpuPwrEventHandlersSet( ) Registers a set of handlers for the various power
management events.
For more information about the routines and the power states supported with
VxWorks, see the API reference for cpuPwrLib.
Also see the VxWorks Architecture Supplement for the mappings between the ACPI
specification C-states and P-states and the power modes supported by the CPU in
question.
48
2 Kernel
2.5 Power Management
Configuring VxWorks With Power Management Facilities
Configure VxWorks with the INCLUDE_CPU_PWR_MGMT component for the
power management framework.
Use one of the power manager components provided by Wind River:
■
■
INCLUDE_CPU_UTIL_PWR_MGR, which is full-featured
CPU-utilization-based power manager.
INCLUDE_CPU_LIGHT_PWR_MGR, which is a light version of a
CPU-utilization-based power manager.
For full descriptions of these power managers, see Wind River Power Managers,
p.47.
A custom power manager can be used in place of a Wind River power manager. It
can be included as an application module or a custom component. For more
information, see 2.6.8 Linking Kernel Application Object Modules with VxWorks, p.64
and 2.8 Custom VxWorks Components and CDFs, p.67.
Power Management and System Performance
Performance is a concern for the CPU power management framework because it
interfaces to the interrupt handling and scheduling sub-systems in a way that
affects the interrupt latency and the task context switch time.
2.5.2 Power Management for Other Architectures
NOTE: For this release, the power management facilities available for architectures
other than IA are the same as provided for VxWorks 6.0 and 6.1. These facilities are
described in this section.
The features described in 2.5.1 Power Management for IA Architecture, p.41 will be
available for other architectures in future releases.
Power management allows the processor to conserve power by entering a low
power state. While in this mode, processor register values and memory contents
are retained. This feature is implemented by putting the CPU in a non-executing
state while the kernel is idle. This has no impact on the operation of peripheral
devices except for the system timer when long sleep mode (described below) is
selected. The VxWorks power management facilities provide two modes of
operation:
49
2
VxWorks
Kernel Programmer's Guide, 6.6
■
Short Sleep Mode
In short sleep mode, the CPU is put in a low power state until the next
interrupt occurs, including the system clock interrupt. The maximum amount
of time the CPU can sleep is therefore the interval in between system clock
interrupts. The short sleep mode does not require special BSP support. It is
provided at the architecture level. See the VxWorks Architecture Supplement to
determine if this mode is supported for your CPU.
■
Long Sleep Mode
In long sleep mode, the CPU is put in a low power state until the next interrupt
occurs, excluding the system clock interrupt. This allows the CPU to sleep
until the next system event is scheduled to occur; such as a task that times out
on a semaphore take operation or a watchdog that fires. This mode requires
BSP support because the system clock interrupt source must be turned off, and
a mechanism must schedule the CPU to wake after a specified amount of time.
To provide power management support for your system, configure VxWorks with
the INCLUDE_POWER_MGMT_BSP_SUPPORT component.
For more information, see the VxWorks BSP Developer’s Guide. Also see System Clock
Modification, p.39.
50
2 Kernel
2.6 Kernel Applications
2.6 Kernel Applications
VxWorks applications that execute in the kernel are created as relocatable object
modules. They can be referred to most specifically as kernel-based application
modules, but it is often convenient to refer to them simply as kernel application
modules or kernel applications. In any case, they should not be confused with
process-based (RTP) applications.
When a kernel-based application module is built, user code is linked to the
required VxWorks libraries, and an ELF binary is produced. Kernel applications
use VxWorks facilities by including header files that define operating system
interfaces and data structures.
Kernel application modules can be either:
■
Downloaded and dynamically linked to the operating system by the object
module loader.
■
Statically linked to the operating system, making them part of the system
image.
Downloading kernel modules is useful for rapid development and debugging, as
the operating system image does not need to be rebuilt for each iteration of the
application. This method can also be used for diagnostic facilities with production
systems. Various development tools, including the debugger and the shell (host or
kernel), can be used to download and manage modules. Modules can be
downloaded to a target from any host file system for which the kernel has support
(NFS, ftp, and so on).
Kernel application modules can also be stored on the target itself in flash or ROM,
in the ROMFS file system, or on disk. Once they have been loaded into the target,
kernel application modules can be started interactively from the shell or
Workbench.
Application modules that are statically linked to the operating system can be run
interactively from the shell or Workbench. VxWorks can also be configured to start
them automatically at boot time. Static linking and automatic startup are obviously
suitable for production systems.
An application that runs in kernel space is not executed as a process; it is simply
another set of tasks running in kernel space. The kernel is not protected from any
misbehavior that a kernel application might engage in—and the applications are
similarly not protected from each other—kernel applications and the kernel run in
the same address space in supervisor mode.
51
2
VxWorks
Kernel Programmer's Guide, 6.6
Note that VxWorks can also be configured with support for applications that
execute in user space as processes. See VxWorks Application Programmer’s Guide:
Applications and Processes.
!
WARNING: If you wish to port a kernel application to a user-mode application that
executes as a real-time process (RTP), you must ensure that it meets the code
requirements of an RTP application and is compiled as such. You must also ensure
that VxWorks is configured with support for RTPs. For more information, see the
VxWorks Application Programmer's Guide: Applications and Processes, and the
VxWorks Migration Guide.
2.6.1 C and C++ Libraries
Both VxWorks native C libraries, and Dinkum C and C++ libraries, are provided
for VxWorks application development. As shown in Table 2-6, VxWorks native
libraries are used for C kernel application development, and Dinkum libraries are
used in all other cases.
Table 2-6
C and C++ Libraries
Type of Application
C Language
C++ Language
Kernel-mode application VxWorks native libraries Dinkum C++ and
embedded (abridged) C++
libraries
User-mode RTP
application
Dinkum C libraries
Dinkum C++ and
embedded (abridged) C++
libraries
The VxWorks native C libraries provide routines outside the ANSI specification.
Note that they provide no support for wide or multi-byte characters.
For more information about these libraries, see the VxWorks and Dinkum API
references. For more information about C++ facilities, see 13. C++ Development.
2.6.2 Application Structure
Kernel application code is similar to common C or C++ applications, with the
exception that it does not require a traditional main( ) routine (unlike a VxWorks
52
2 Kernel
2.6 Kernel Applications
process-based application). It simply requires an entry point routine that starts all
the tasks required to get the application running.
2
NOTE: If your kernel application includes a main( ) routine, do not assume that it
will start automatically. Kernel application modules that are downloaded or
simply stored in the system image must be started interactively (or be started by
another application that is already running). The operating system can also be
configured to start applications automatically at boot time (see a2.6.10 Configuring
VxWorks to Run Applications Automatically, p.66).
The entry-point routine performs any data initialization that is required, and starts
all the tasks that the running application uses. For example, a kernel application
might have a routine named like myAppStartUp( ), which could look something
like this:
void myAppStartUp (void)
{
runFoo();
tidThis = taskSpawn("tThis", 200, 0, STACK_SIZE,
(FUNCPTR) thisRoutine,0,0,0,0,0,0,0,0,0,0);
tidThat = taskSpawn("tThat", 220, 0, STACK_SIZE,
(FUNCPTR) thatRoutine,0,0,0,0,0,0,0,0,0,0);
tidAnother = taskSpawn("tAnother", 230, 0, STACK_SIZE,
(FUNCPTR) anotherRoutine,0,0,0,0,0,0,0,0,0,0);
return (OK);
}
For information about VxWorks tasks and multitasking, see 4. Multitasking. For
information about working with C++ see 13. C++ Development.
2.6.3 VxWorks Header Files
Many kernel applications make use of VxWorks operating system facilities or
utility libraries. This usually requires that the source module refer to VxWorks
header files. The following sections discuss the use of VxWorks header files.
VxWorks header files supply ANSI C function prototype declarations for all global
VxWorks routines. VxWorks provides all header files specified by the ANSI
X3.159-1989 standard.
VxWorks system header files are in the directory installDir/vxworks-6.x/target/h
and its subdirectories.
53
VxWorks
Kernel Programmer's Guide, 6.6
VxWorks Header File: vxWorks.h
The header file vxWorks.h must be included first by every kernel application
module that uses VxWorks facilities. It contains many basic definitions and types
that are used extensively by other VxWorks modules. Many other VxWorks
header files require these definitions. Include vxWorks.h with the following line:
#include
Other VxWorks Header Files
Kernel applications can include other VxWorks header files, as needed, to access
VxWorks facilities. For example, a module that uses the VxWorks linked-list
subroutine library must include the lstLib.h file with the following line:
#include
The API reference entry for each library lists all header files necessary to use that
library.
ANSI Header Files
All ANSI-specified header files are included in VxWorks. Those that are
compiler-independent or more VxWorks-specific are provided in
installDir/vxworks-6.x/target/h while a few that are compiler-dependent (for
example stddef.h and stdarg.h) are provided by the compiler installation. Each
toolchain knows how to find its own internal headers; no special compile flags are
needed.
ANSI C++ Header Files
Each compiler has its own C++ libraries and C++ headers (such as iostream and
new). The C++ headers are located in the compiler installation directory rather
than in installDir/vxworks-6.x/target/h. No special flags are required to enable the
compilers to find these headers. For more information about C++ development,
see 13. C++ Development.
NOTE: In releases prior to VxWorks 5.5, Wind River recommended the use of the
flag -nostdinc. This flag should not be used with the current release since it prevents
the compilers from finding headers such as stddef.h.
The -I Compiler Flag
By default, the compiler searches for header files first in the directory of the source
module and then in its internal subdirectories. In general,
installDir/vxworks-6.x/target/h should always be searched before the compilers’
54
2 Kernel
2.6 Kernel Applications
other internal subdirectories; to ensure this, always use the following flag for
compiling under VxWorks:
2
-I %WIND_BASE%/target/h %WIND_BASE%/target/h/wrn/coreip
Some header files are located in subdirectories. To refer to header files in these
subdirectories, be sure to specify the subdirectory name in the include statement,
so that the files can be located with a single -I specifier. For example:
#include
#include
VxWorks Nested Header Files
Some VxWorks facilities make use of other, lower-level VxWorks facilities. For
example, the tty management facility uses the ring buffer subroutine library. The
tty header file tyLib.h uses definitions that are supplied by the ring buffer header
file rngLib.h.
It would be inconvenient to require you to be aware of such include-file
interdependencies and ordering. Instead, all VxWorks header files explicitly
include all prerequisite header files. Thus, tyLib.h itself contains an include of
rngLib.h. (The one exception is the basic VxWorks header file vxWorks.h, which
all other header files assume is already included.)
Generally, explicit inclusion of prerequisite header files can pose a problem: a
header file could get included more than once and generate fatal compilation
errors (because the C preprocessor regards duplicate definitions as potential
sources of conflict). However, all VxWorks header files contain conditional
compilation statements and definitions that ensure that their text is included only
once, no matter how many times they are specified by include statements. Thus, a
kernel application module can include just those header files it needs directly,
without regard to interdependencies or ordering, and no conflicts will arise.
VxWorks Private Header Files
Some elements of VxWorks are internal details that may change and so should not
be referenced in a kernel application. The only supported uses of a module’s
facilities are through the public definitions in the header file, and through the
module’s subroutine interfaces. Your adherence ensures that your application
code is not affected by internal changes in the implementation of a VxWorks
module.
Some header files mark internal details using HIDDEN comments:
/* HIDDEN */
...
/* END HIDDEN */
55
VxWorks
Kernel Programmer's Guide, 6.6
Internal details are also hidden with private header files: files that are stored in the
directory installDir/vxworks-6.x/target/h/private. The naming conventions for
these files parallel those in installDir/vxworks-6.x/target/h with the library name
followed by P.h. For example, the private header file for semLib is
installDir/vxworks-6.x/target/h/private/semLibP.h.
2.6.4 Static Instantiation of Kernel Objects
The VxWorks APIs have a long established convention for the creation and
deletion of kernel entities. Objects such as tasks, semaphores, message queues and
watchdogs are instantiated using their respective creation APIs (for example,
taskSpawn( ), semXCreate( ), and so on) and deleted using their respective delete
APIs (for example, msgQDelete( ), wdDelete( ), and so on.). Object creation is a
two-step process: first the memory for the object is allocated from the system,
which is then initialized appropriately before the object is considered usable.
Object deletion involves invalidation of the object, followed by freeing its memory
back to the system. Thus, object creation and deletion are dependent on dynamic
memory allocation, usually through the malloc( ) and free( ) routines.
Dynamic creation and deletion of objects at run-time is a convenient programming
paradigm, though it has certain disadvantages for some real-time critical
applications. First, the allocation of memory from the system cannot always be
guaranteed. Should the system run out of memory the application cannot create
the resources it must have function. The application must then resort to a suitable
error recovery process if any exists, or abort in some fashion. Second, dynamic
allocation of memory is a relatively slow operation that may potentially block the
calling task. This makes dynamic allocation non-deterministic in performance.
Static instantiation of objects is a faster, more deterministic alternative to dynamic
creation. In static instantiation, the object is declared as a compile time variable.
Thus the compiler allocates storage for the object in the program being compiled.
No more allocation need be done. At run-time the objects memory is available
immediately at startup for the initialization step. Initialization of pre-allocated
memory is much more deterministic and faster than dynamic creation. Such static
declaration of objects cannot fail, unless the program itself is too large to fit in the
systems memory.
Many applications are suited to exploit static instantiation of objects in varying
degrees. Most applications require some resources to be created, that last for the
lifetime of the application. These resources are never deleted. In lieu of the latter,
objects that last for the lifetime of the application are ideally suited for static (that
is, compile time) allocation. To the extent that they are instantiated statically
56
2 Kernel
2.6 Kernel Applications
(which we shall see below), the application is that much more fail safe and fast to
launch.
See 2.4.5 Small VxWorks Configuration Profiles, p.24 for information about operating
system profiles of particular relevance for static instantiation.
Dynamic Instantiation of an Object
struct my_object * pMyObj;
...
pMyObj = (struct my_object *) malloc (sizeof (struct my_object));
if (pMyObj != NULL)
{
objectInit (pMyOjb);
return (OK);
}
else
{
/* failure path */
return (ERROR);
}
Static Instantiation of an Object
struct my_object myObj;
...
objectInit (&myOjb);
/* myObj now ready for use */
Static instantiation of objects has the following advantages:
■
The application logic is made simpler by not having to consider the case when
dynamic allocation fails.
■
Compile time declaration of objects does not take up space in the executable
file or flash memory. If an object is merely declared at compile time but not
initialized, it is placed by the compiler in the un-initialized data section (also
known as the bss section). Un-initialized data is required by the ANSI C
standard to be of value zero. Hence the un-initialized data section (the bss
section) of a program does not occupy any space in an executable file or in
VxWorks ROM images. Un-initialized data does contribute to the programs
run-time footprint in memory, but so does dynamic allocation. The program
will not consume any more memory footprint than it did with dynamic
allocation of objects.
Using static instantiation whenever possible is more robust, deterministic and fast.
Static instantiation of objects is therefore much better suited for real-time
applications. On the other hand some applications are inherently dynamic in
nature. For these, dynamic creation and deletion is always available.
57
2
VxWorks
Kernel Programmer's Guide, 6.6
Static Instantiation of Objects
Kernel objects such as tasks, semaphores, message queues and watchdogs can be
instantiated statically using the same principles outlined above.
Normally these objects are created using the appropriate create routines for that
type of object, and deleted using the appropriate delete routine. As mentioned
before, creation and deletion involve dynamic memory allocation and free
respectively.
Static instantiation of objects is a two-step process. First the object to be created is
declared, usually at global scope. Next the declared object is initialized using an
initialization routine, which makes it ready for use. In contrast, dynamic creation
with create routines is a one-step process. Static instantiation of objects is thus a
little less convenient, but more deterministic. Users can choose the style that suits
their purpose.
NOTE: Static instantiation should only be used for objects that are kernel-resident.
It is not meant to be used to create objects in a process (RTP).
The following sections describe an alternative static instantiation method for each
of these entities.
Scope Of Static Declarations
The macros declaring kernel objects (that is VX_BINARY_SEMAPHORE, VX_TASK,
and so on) are usually declared as global variables. Since all these kernel objects
are used for inter-task communication and synchronization, their IDs are the
means by which other tasks use these objects. Hence global objects and global IDs
are the common method by which these objects are accessed and used. However it
is not always necessary that they be global. An object declaration can also be done
at function scope provided the object stays in a valid scope for the duration of its
use.
Static Instantiation of Tasks
The taskSpawn( ) routine has been the standard method for instantiating tasks.
This API relies on dynamic allocations. In order to instantiate tasks statically
several macros have been provided to emulate the dynamic instantiation
capability provided by taskSpawn( ) and related routines.
The VX_TASK macro declares a task object at compilation time. It takes two
arguments: the task name and its stack size. When calling taskSpawn( ) the name
may be a NULL pointer, but when using the VX_TASK macro, a name is
58
2 Kernel
2.6 Kernel Applications
mandatory. The stack size must evaluate to a non-zero integer value and must be
a compile-time constant.
The VX_TASK_INSTANTIATE macro is the static equivalent of the taskSpawn( )
routine. It initializes and schedules the task, making it run according to its priority.
VX_TASK_INSTANTIATE evaluates to the task ID of the spawned task if it was
successful, or ERROR if not.
The following example illustrate spawning tasks statically:
#include
#include
VX_TASK(myTask,4096);
int myTaskId;
STATUS initializeFunction (void)
{
myTaskId = VX_TASK_INSTANTIATE(myTask, 100, 0, 4096, pEntry, \
0,1,2,3,4,5,6,7,8,9);
if (myTaskId != ERROR)
return (OK);
else
return (ERROR);
}
/* instantiation succeeded */
Sometimes users may prefer to initialize a task, but keep it suspended until needed
later. This can be achieved by using the VX_TASK_INITIALIZE macro, as
illustrated below. Since the task is left suspended, users are responsible for calling
taskActivate( ) in order to run the task.
#include
#include
VX_TASK(myTask,4096);
int myTaskId;
STATUS initializeFunction (void)
{
myTaskId = VX_TASK_INITIALIZE(myTask, 100, 0, 4096, pEntry, \
0,1,2,3,4,5,6,7,8,9);
if (myTaskId != NULL)
{
taskActivate (myTaskId);
return (OK);
}
else
return (ERROR);
}
59
2
VxWorks
Kernel Programmer's Guide, 6.6
It is the programmer’s responsibility to pass the same name to
VX_TASK_INSTANTIATE as was used in the VX_TASK declaration, which is
myTask in this case. The arguments to VX_TASK_INSTANTIATE and their
meaning are the same as those passed to taskSpawn( ). This makes the usage of
VX_TASK_INSTANTIATE consistent with taskSpawn( ). Please note the backslash
that continues the argument list on the succeeding line. This backslash character is
crucial if the arguments span more than one line. This is to ensure correct macro
expansion.
For more information, see the API reference for taskLib.
Static Instantiation Of Semaphores
The macros VX_BINARY_SEMAPHORE, VX_COUNTING_SEMAPHORE and
VX_MUTEX_SEMAPHORE are used to declare a semaphore of type binary,
counting, and mutex respectively. These macros take the semaphore name as an
argument. The declared semaphores are initialized by calling routines
semBInitialize( ), semCInitialize( ) and semMInitialize( ) respectively.
The three semXInitialize( ) routines are the equivalents of their respective
semXCreate( ) routines, the only difference being that the semaphore same name
used in the associated VX_XXX_SEMAPHORE be passed to the semXInitialize( )
routines. The return value from the semXInitialize( ) routines is a semaphore ID
that is then used to perform all operations on the semaphores.
The following example illustrates static instantiation of a binary semaphore:
#include
#include
VX_BINARY_SEMAPHORE(mySemB);
SEM_ID mySemBId;
/* declare the semaphore */
/* semaphore ID for further operations */
STATUS initializeFunction (void)
{
if ((mySemBId = semBInitialize (mysemB, options, 0)) == NULL)
return (ERROR);
/* initialization failed */
else
return (OK);
}
For more information, see the API reference for semLib.
Static Instantiation of Message Queues
The macro VX_MSG_Q is used to declare a message queue at compile time. It takes
three parameters: the name, the maximum number of messages in the message
queue, and the maximum size of each message.
60
2 Kernel
2.6 Kernel Applications
After this declaration, the msgQInitialize( ) routine is used to initialize the
message queue and make it ready for use.
The following example illustrates static instantiation of a message queue:
#include
#include
VX_MSG_Q(myMsgQ,100,16);
MSG_Q_ID myMsgQId;
/* declare the msgQ */
/* MsgQ ID to send/receive messages */
STATUS initializeFunction (void)
{
if ((myMsgQId = msgQInitialize (myMsgQ, 100, 16, options)) == NULL)
return (ERROR);
/* initialization failed */
else
return (OK);
}
As with other static instantiation macros it is crucial to pass exactly the same name
used in the VX_MSG_Q declaration to the msgQInitialize( ) routine, or else
compilation errors will result. Also it is crucial to pass exactly the same values for
the message size and maximum number of messages as was passed to the
VX_MSG_Q macro.
For more information, see the API reference for msgQLib.
Static Instantiation of Watchdogs
The macro VX_WDOG is used to declare a watchdog at compile time. It takes one
parameter, the name of the watchdog. After this declaration, the routine
wdInitialize( ) is used to initialize the watchdog and make it ready for use.
#include
#include
VX_WDOG(myWdog);
WDOG_ID myWdogId;
/* declare the watchdog */
/* watchdog ID for further operations */
STATUS initializeFunction (void)
{
if ((myWdogId = wdInitialize (myWdog)) == NULL)
return (ERROR);
/* initialization failed */
else
return (OK);
}
As with other static instantiation macros it is crucial to pass exactly the same name
used in the VX_WDOG declaration to the wdInitialize( ) routine, or else
compilation errors will result.
For more information, see the API reference for wdLib.
61
2
VxWorks
Kernel Programmer's Guide, 6.6
2.6.5 Applications and VxWorks Kernel Component Requirements
VxWorks is a highly configurable operating system. When kernel application
modules are built independently of the operating system (see 2.6.6 Building Kernel
Application Modules, p.62), the build process cannot determine if the instance of
VxWorks on which the application will eventually run has been configured with
all of the components that the application requires (for example, networking and
file systems). It is, therefore, useful for application code to check for errors
indicating that kernel facilities are not available (that is, check the return values of
API calls) and to respond appropriately.
When kernel application modules are linked with the operating system, the build
system generates errors with regard to missing components. Both Workbench and
the vxprj command-line tool also provide a mechanisms for checking
dependencies and for reconfiguring VxWorks accordingly.
2.6.6 Building Kernel Application Modules
The VxWorks environment provides simple mechanisms for building kernel
application modules, including a useful set of default makefile rules. Both the IDE
and command line can be used to build applications. For command-line use, the
wrenv utility program can be used to open a command shell with the appropriate
host environment variables set. See VxWorks Command-Line Tools User's Guide:
Creating a Development Shell with wrenv and the VxWorks Command-Line Tools User’s
Guide.
Using Makefile Include Files for Kernel Application Modules
You can make use of the VxWorks makefile structure to put together your own
application makefiles quickly and efficiently. If you build your application directly
in a BSP directory (or in a copy of one), you can use the makefile in that BSP, by
specifying variable definitions that include the components of your application.
You can specify values for these variables either from the make command line, or
from your own makefiles (when you take advantage of the predefined VxWorks
make include files).
ADDED_CFLAGS
Application-specific compiler options for C programs.
ADDED_C++FLAGS
Application-specific compiler options for C++ programs.
62
2 Kernel
2.6 Kernel Applications
Additional variables can be used to link kernel application modules with the
VxWorks image; see 2.6.8 Linking Kernel Application Object Modules with VxWorks,
p.64.
For more information about makefiles, see the VxWorks Command-Line Tools User’s
Guide.
You can also take advantage of the makefile structure if you develop kernel
application modules in separate directories. Example 2-1 illustrates the general
scheme. Include the makefile headers that specify variables, and list the object
modules you want built as dependencies of a target. This simple scheme is usually
sufficient, because the makefile variables are carefully designed to fit into the
default rules that make knows about.2
Example 2-1
Skeleton Makefile for Kernel Applications
#
#
#
#
#
#
Makefile - makefile for ...
Copyright ...
DESCRIPTION
This file specifies how to build ...
## It is often convenient to override the following with "make CPU=..."
CPU
= cputype
TOOL
= diab
include $(WIND_BASE)/target/h/make/defs.bsp
## Only redefine make definitions below this point, or your definitions
## will be overwritten by the makefile stubs above.
exe : myApp.o
For information about build options, see the VxWorks Architecture Supplement for
the target architecture in question. For information about using makefiles to build
applications, see the VxWorks Command-Line Tools User’s Guide.
Statically Linking Kernel Application Modules
In general, kernel application modules do not need to be linked before being
downloaded to the target. However, when several modules cross reference each
other they should be linked to form a single module. With C++ code, this linking
should be done before the munch step. (See 13.5.2 Munching a C++ Application
Module, p.650.)
2. However, if you are working with C++, it may be also convenient to copy the .cpp.out rule
from installDir/vxworks-6.x/target/h/make/rules.bsp into your application’s makefile.
63
2
VxWorks
Kernel Programmer's Guide, 6.6
The following example is a command to link several modules, using the Wind
River linker for the PowerPC family of processors:
c:\> dld -o applic.o -r applic1.o applic2.o applic3.o
Similarly, this example uses the GNU linker:
c:\> ldppc -o applic.o -r applic1.o applic2.o applic3.o
In either case, the command creates the object module applic.o from the object
modules applic1.o, applic2.o, and applic3.o. The -r option is required, because the
object-module output must be left in relocatable form so that it can be downloaded
and linked to the target VxWorks image.
Any VxWorks facilities called by the kernel application modules are reported by
the linker as unresolved externals. These are resolved by the loader when the
module is loaded into VxWorks memory.
!
WARNING: Do not link each kernel application module with the VxWorks libraries.
Doing this defeats the load-time linking feature of the loader, and wastes space by
writing multiple copies of VxWorks system modules on the target.
2.6.7 Downloading Kernel Application Object Modules to a Target
Kernel application object modules can be downloaded from Workbench or from
the kernel shell. Once a module has been loaded into target memory, any
subroutine in the module can be invoked, tasks can be spawned, debugging
facilities employed with the modules, and so on. It is often useful to make use of a
startup routine to run the application (see 2.6.2 Application Structure, p.52).
For information about using the kernel shell and module loader, see 12.2 Kernel
Shell, p.577 and 12.3 Kernel Object-Module Loader, p.603. For information about
using theWorkbench, see the Wind River Workbench User’s Guide.
2.6.8 Linking Kernel Application Object Modules with VxWorks
In order to produce complete systems that include kernel application modules, the
modules must be statically linked with the VxWorks image. The makefile
EXTRA_MODULES variable can be used to do so. It can be used from the command
line as follows:
% make EXTRA_MODULES="foo.o"
64
2 Kernel
2.6 Kernel Applications
For more information about using makefile variables, see 2.6.6 Building Kernel
Application Modules, p.62.
To include your kernel application modules in the system image using a makefile,
identify the names of the application object modules (with the .o suffix) with
EXTRA_MODULES. For example, to link the module myMod.o with the operating
system, add a line like the following:
EXTRA_MODULES = myMod.o
Building the system image with the module linked in is the final part of this step.
In the project directory, execute the following command:
c:\myVxProj\osProj> make vxWorks
For information about how to have kernel applications start automatically at boot
time, see 2.6.10 Configuring VxWorks to Run Applications Automatically, p.66.
2.6.9 Image Size Considerations
The size of the system image is often an important consideration, particularly
when kernel application modules are linked with the operating system. This is true
whether the image is loaded by a boot loader or is self-booting (see 2.4.1 VxWorks
Image Types, p.15).
!
CAUTION: For ROM-based images, ensure that ROM_SIZE configuration
parameter reflects the capacity of the ROMs used.
Boot Loader and Downloadable Image
Generally, VxWorks boot loader code is copied to a start address in RAM above
the constant RAM_HIGH_ADRS, and the boot loader in turn copies the
downloaded system image starting at RAM_LOW_ADRS. The values of these
constants are architecture dependent, but in any case the system image must not
exceed the space between the two. Otherwise the system image will overwrite the
boot loader code while downloading, potentially killing the booting process.
To help avoid this, the last command executed when you build a new VxWorks
image is vxsize, which shows the size of the new executable image and how much
space (if any) is left in the area below the space used for boot ROM code:
vxsize 386 -v 00100000 00020000 vxWorks
vxWorks: 612328(t) + 69456(d) + 34736(b) = 716520 (235720 bytes left)
(In this output, t stands for text segment, d for data segment, and b for bss.)
65
2
VxWorks
Kernel Programmer's Guide, 6.6
Make sure that RAM_HIGH_ADRS is less than LOCAL_MEM_SIZE. If your new
image is too large, vxsize issues a warning. In this case, you can reconfigure the
boot loader to copy the boot ROM code to a sufficiently high memory address by
increasing the value of RAM_HIGH_ADRS in config.h and in the BSP’s makefile
(both values must agree). Then rebuild the boot loader. For more information, see
Persistent Memory Region, p.150.
Self-Booting Image
For self-booting images, the data segment of a ROM-resident VxWorks system is
loaded at RAM_LOW_ADRS (defined in the makefile) to minimize fragmentation.
For a CPU board with limited memory (under 1 MB of RAM), make sure that
RAM_HIGH_ADRS is less than LOCAL_MEM_SIZE by a margin sufficient to
accommodate the data segment. Note that RAM_HIGH_ADRS is defined in both
the BSP makefile and config.h (both values must agree).
2.6.10 Configuring VxWorks to Run Applications Automatically
VxWorks can be configured to start kernel applications automatically at boot time.
To do so, perform the following steps:
1.
Configure VxWorks with the INCLUDE_USER_APPL component.
2.
Add a call to the application’s entry-point routine in the usrAppInit( ) routine
stub, which is in installDir/vxworks-6.x/target/proj/projDir/usrAppInit.c.
Assuming, for example, that the application entry point routine
myAppStartUp( ) starts all the required application tasks, you would add a
call to that routine in the usrAppInit( ) stub as follows:
void usrAppInit (void)
{
#ifdef USER_APPL_INIT
USER_APPL_INIT;
#endif
/* for backwards compatibility */
myAppStartUp();
}
3.
66
Link the kernel-base application object modules with the kernel image (see
2.6.8 Linking Kernel Application Object Modules with VxWorks, p.64).
2 Kernel
2.7 Custom Kernel Libraries
2.7 Custom Kernel Libraries
For information about creating custom kernel libraries, see the VxWorks
Command-Line Tools User’s Guide.
2.8 Custom VxWorks Components and CDFs
A VxWorks component is the basic unit of functionality with which VxWorks can
be configured. VxWorks kernel components are described in Component
Description Files (CDFs), using the Component Description Language (CDL). (For
introductory information about the components provided with VxWorks, see
2.4.2 VxWorks Components, p.17.)
Wind River recommends using CDFs for developing extensions to the VxWorks
operating system. If you wish to develop components that make APIs available to
RTP applications (which run in user mode as real-time processes), also see
2.9 Custom System Calls, p.100.
NOTE: Functionality can be added to VxWorks in the form of kernel modules that
do not have to be defined as VxWorks components (see 2.6 Kernel Applications,
p.51). However, in order to make use of either Workbench or the vxprj
command-line configuration facilities, to define dependencies between
components, and so on, extensions to the operating system should be developed
using CDFs.
A CDF identifies the binary and source code elements that make up the
component, its configuration parameters, relationship to other components, and
so on. CDFs also define information about how components are displayed in the
Workbench kernel configuration facility, and they can be used to group
components into predefined sets to facilitate VxWorks configuration. While some
components are autonomous, some have dependencies on other components,
which must be included in the configuration of the operating system for run-time
operation.
Both Workbench and the vxprj command-line facility use CDFs for configuring the
operating system with selected components, for setting component parameter
values, and so on. Workbench also uses information in CDFs to display the names
and descriptions of components, and to provide links to online help. For
67
2
VxWorks
Kernel Programmer's Guide, 6.6
information about Workbench and the command-line facilities used for
configuring and building VxWorks, see the Wind River Workbench User’s Guide and
the VxWorks Command-Line Tools User’s Guide.
The first of the sections that follow provide procedural instructions for creating,
installing, and testing a CDF; then additional sections provide reference
information about CDL and its constituent elements.
2.8.1 Creating and Modifying Components
This section provides an instructional example of how to create a component or to
modify an existing one. The example is generally useful in its illustration of the
process, but does not make exhaustive use of CDL objects and properties.
You must follow the CDL conventions when creating a CDF. For detailed
information, see the sections referenced below in the instructions, as well as
2.8.13 CDF Template, p.96.
After creating a CDF, you must place the file in the proper path to ensure that the
configuration facility reads the information and includes the component in the
hierarchy of components. For information in this regard, see 2.8.2 CDF Precedence
and CDF Installation, p.75 and Adding New CDFs to the VxWorks Installation Tree,
p.77.
Information about testing a component is provided in 2.8.3 Testing New
Components, p.78.
Defining a Component
This section describes the process of defining your own component. To allow for
the greatest flexibility, there is no convention governing the order in which
properties describe a component or the sequence in which CDL objects are entered
into a component description file. The following steps taken to create the
component INCLUDE_FOO are a suggested order only; the sequence is not
mandatory. Nor is there meant to be any suggestion that you use all of the
properties and object classes described.
The naming conventions for CDL are described in 2.8.5 CDF Naming Conventions,
p.80. Note that CDF files must have a .cdf suffix.
68
2 Kernel
2.8 Custom VxWorks Components and CDFs
Step 1:
Name and Provide a Synopsis
To create your new component, first name it and provide a synopsis of its utility.
Component
NAME
SYNOPSIS
...
INCLUDE_FOO
{
foo component
this is just an example component
In the component INCLUDE_FOO, the NAME property is foo component. The
SYNOPSIS property instructively informs the user that this is just an example
component.
NAME and SYNOPSIS properties affect user presentation only; they have no
bearing on initialization sequence or dependencies.
For information about the properties of the component object (NAME and so on),
see 2.8.6 Component CDF Object, p.81.
Step 2:
Describe the Code-Related Elements
Next, describe your component’s code portions by defining any modules and
source configlettes that should be configured in during the build.
If the component INCLUDE_FOO has an object module associated with it—and not
source code—use the MODULES property to specify it. You can specify any
number of modules this way. In the following example, fooLib.o and fooShow.o
are listed:
...
MODULES
HDR_FILES
ARCHIVE
CONFIGLETTES
...
fooLib.o fooShow.o
foo.h
fooLib.a
fooConfig.c
The configuration facilities (Workbench and vxprj) automatically analyze object
module dependencies in order to calculate component dependencies. (Workbench
also offers visibility into component dependencies by graphically presenting
component relationships.)
For example, if fooLib.o has an unresolved global for logMsg( ), an automatic
dependency upon the component INCLUDE_LOGGING is detected. For
components not shipped in object module format, CDL supports the explicit
listing of component dependencies.
If the source portion contains a call to logMsg( ), the configuration facility does not
detect the dependency; instead, an explicit dependency upon
INCLUDE_LOGGING should be declared using the REQUIRES property. See
Step 6.
69
2
VxWorks
Kernel Programmer's Guide, 6.6
If your module is not located in the standard path, use the ARCHIVE property to
specify the archive name, for example, /somePath/fooLib.a. (For additional
instructions concerning the use of ARCHIVE, see Step 10.)
NOTE: Developers should create their own archives for custom components. Do
not modify Wind River archives.
Use the HDR_FILES property to specify any .h files associated with the component,
like foo.h. These files are emitted in prjConfig.c, providing the initialization
routine with a declaration.
If there is source code that should be compiled as part of the component, put it in
a .c file and specify the file name in the CONFIGLETTES property, for example,
fooConfig.c. Component parameters should be referenced in the configlette or in
the initialization routine; otherwise they have no effect.
For detailed information about the properties of the component object (MODULES
and so on), see 2.8.6 Component CDF Object, p.81.
Step 3:
Set Up Initialization
If the component must be initialized, use the INIT_RTN property of the component
object class to specify the initialization routine and its arguments to be called, as in
fooInit(arg1, arg2). If your component needs more than a single line of
initialization, create or add the initialization code to a .c file and use the
CONFIGLETTES property instead. By associating a configlette with an
initialization routine, you are securing the configlette’s place in the initialization
sequence.
...
INIT_RTN
...
fooInit(arg1, arg2);
If you are not using the MODULES property of the component object, use the
LINK_SYMS property to include your object module from a linked archive. The
system generates an unresolved reference to the symbol (fooRtn1 in this example),
causing the linker to resolve it by extracting your module from the archive.
...
LINK_SYMS
...
fooRtn1
For more information about the INIT_RTN property of the component object, see
and 2.8.6 Component CDF Object, p.81.
70
2 Kernel
2.8 Custom VxWorks Components and CDFs
Step 4:
Establish the Initialization Sequence
Initialization order is important. You can control when in the initialization
sequence your component is initialized with the _INIT_ORDER property. A
component (or initialization group) that is bound to an existing initialization
group using the _INIT_ORDER property is, by default, initialized last within that
group. This is typically the desired effect; however, you can override this behavior
by explicitly using the INIT_BEFORE property.
...
_INIT_ORDER
INIT_BEFORE
...
usrRoot
INCLUDE_USER_APPL
In the example, INCLUDE_FOO is declared a member of the usrRoot initialization
group. INIT_BEFORE has been used for fine control, and INCLUDE_FOO is
initialized before INCLUDE_USER_APPL. For more information about the
_INIT_ORDER and INIT_BEFORE properties of the component object, see and
2.8.6 Component CDF Object, p.81.
Alternatively, you can create a new initialization group and declare
INCLUDE_FOO a member; however, you would have to declare the new
initialization group a member of an existing initialization group. For information
on initialization groups, see 2.8.8 Initialization Group CDF Object, p.88.
NOTE: INIT_BEFORE only affects ordering within the initialization group. Do not
reference a component that is not in the initialization group; this has no effect.
NOTE: The INIT_AFTER property has no effect in this release.
Step 5:
Link Helpful Documentation
Specify related reference entries (in HTML format) with the HELP property.
...
HELP
...
fooMan
By default, a build automatically includes reference entries related to values
declared by the MODULES and INIT_RTN properties. In the case of
INCLUDE_FOO, in addition to fooMan.html, which is specified by HELP, the build
associates the fooLib and fooShow libraries and the fooInit( ) routine.
For more information about the HELP property of the component object, see and
2.8.6 Component CDF Object, p.81.
71
2
VxWorks
Kernel Programmer's Guide, 6.6
Step 6:
Define Dependencies
Use the REQUIRES and INCLUDE_WHEN properties to explicitly declare
dependencies among components. (See Step 2 to learn how the configuration
facility automatically configures object module-related dependencies.)
The configuration facility does not detect implicit dependencies when a
component is not shipped in object module format. Likewise, no dependencies are
detected when symbols are referenced by pointers at run-time. Both circumstances
require you to declare dependencies explicitly.
...
REQUIRES
INCLUDE_WHEN
...
INCLUDE_LOGGING
INCLUDE_POSIX_AIO INCLUDE_POSIX_MQ
In the example, REQUIRES declares that INCLUDE_LOGGING must be configured
along with INCLUDE_FOO. INCLUDE_WHEN tells the system that whenever the
components INCLUDE_POSIX_AIO and INCLUDE_POSIX_MQ are included, then
INCLUDE_FOO must also be included.
For information about properties of the component object (including REQUIRES
and so on), see 2.8.6 Component CDF Object, p.81.
NOTE: In general, the configuration facility is designed to increase flexibility when
selecting components, that is, to increase scalability; specifying a REQUIRES
relationship reduces flexibility. Be sure that using REQUIRES is the best course of
action in your situation before implementing it.
Step 7:
List Associated Parameters
In the component object, use the CFG_PARAMS property to declare all associated
parameters, for example, FOO_MAX_COUNT.
...
CFG_PARAMS
...
FOO_MAX_COUNT
For information about properties of the component object (including
CFG_PARAMS), see 2.8.6 Component CDF Object, p.81.
Step 8:
Define Parameters
For each parameter declared by CFG_PARAMS, create a parameter object to
describe it. Provide a name using the NAME property.
Use the TYPE property to specify the data type, either int, uint, bool, string, exists,
or untyped.
72
2 Kernel
2.8 Custom VxWorks Components and CDFs
Use the DEFAULT property to specify a default value for each parameter.
Parameter
NAME
TYPE
DEFAULT
}
FOO_MAX_COUNT
Foo maximum
uint
50
{
2
For information about the parameter object and its properties, see 2.8.7 Parameter
CDF Object, p.87.
!
CAUTION: A component is considered mis-configured if it contains a parameter
without an assigned value. Be sure to assign default values, unless there is no
reasonable default and you want to force the user to set it explicitly. (Other CDF
files with higher precedence, may of course be used to set the parameter value; see
CDF File Naming and Precedence, p.75.)
Step 9:
Define Group Membership
A component must be associated with either a folder or selection, otherwise it is
not visible in Workbench. Assign a component to a folder because of its logical
inclusion in the group defined by the folder, based on similar or shared
functionality, for example. By including a component in a folder, you make it
possible for the user to load it simultaneously with other components in its group
by declaring them as default values for the folder, using the folder object’s
DEFAULTS property.
...
_CHILDREN
...
FOLDER_ROOT
The _CHILDREN property declares that INCLUDE_FOO is a child component of
folder FOLDER_ROOT. The prepended underscore (“_”) serves to reverse the
relationship declared by the property CHILDREN, which means that _CHILDREN
identifies the parent. You can also use _DEFAULTS in conjunction with _CHILDREN
to specify a component as a default component of a folder.
If you think a component is becoming too complex, you can divide it into a set of
components assigned to a folder or selection object. In the following example,
INCLUDE_FOO has been specified as part of a selection. You can add or remove the
group from your configuration as a unit rather than by its individual components.
For folders, the DEFAULTS property specifies the base set of components that are
included if the group is configured without any overrides.
For selections, the DEFAULTS property specifies the components that are included
to satisfy the count (declared by the COUNT property), if you provide no
alternative values.
73
VxWorks
Kernel Programmer's Guide, 6.6
In a selection group, the COUNT property specifies the minimum and maximum
number of included components. If the user exceeds these limits the system flags
the selection as mis-configured.
Selection
NAME
SYNOPSIS
COUNT
CHILDREN
DEFAULTS
SELECT_FOO
{
Foo type
Select the type of desired FOO support
0-1
INCLUDE_FOO_TYPE1 \
INCLUDE_FOO_TYPE2 \
INCLUDE_FOO_TYPE3
INCLUDE_FOO_TYPE1
}
For information about the folder and section objects and its properties, see
2.8.11 Folder CDF Object, p.92 and 2.8.12 Selection CDF Object, p.94.
Step 10:
Create a Dummy Component
The configuration facility analyzes archives only when they are associated with
included components. This creates a chicken and egg problem: in order to know
about a particular archive, the configuration facility would need to analyze
components before they are actually added. In other words, if you add a
component declared with an ARCHIVE property, the configuration analysis is
done without knowing what the value of ARCHIVE is. So, if your component
includes an archive with several object modules, you should create a dummy
component that is always included, making it possible for the configuration
facility to know that a new archive should be read. Call such a component
INSTALL_FOO. It should contain only NAME, SYNOPSIS, and ARCHIVE
properties. The user cannot add other components from the same archive until
INSTALL_FOO is added.
For information about properties of the component object (including ARCHIVE),
see 2.8.6 Component CDF Object, p.81.
!
CAUTION: Do not alter Wind River-supplied CDFs directly. Use the naming
convention to create a file whose higher precedence overrides the default
properties of Wind River-supplied components.
Modifying a Component
Do not modify any Wind River CDF.
However, you can effectively modify the properties of existing components by
re-specifying them in another, higher priority CDF file. Third-party CDF files are
74
2 Kernel
2.8 Custom VxWorks Components and CDFs
by convention read last and therefore have the highest priority. Use the naming
convention to create a high-precedence CDF file that overrides the default
properties of Wind River components.
For detailed information, see 2.8.2 CDF Precedence and CDF Installation, p.75.
In the following example, the default number of open file descriptors
(NUM_FILES) in the standard Wind River component INCLUDE_IO_SYSTEM has
been modified. The normal default value is 50.
Parameter NUM_FILES {
DEFAULT
75
}
By adding these example lines of code to a third-party CDF file, by removing and
adding the component if it is already in the configuration, and by re-building the
project, the value of NUM_FILES is changed to 75. The original Wind River CDF
file, 00vxWorks.cdf, is not changed; the default property value is changed because
the third-party file has higher precedence. Other property values remain the same
unless specifically redefined.
!
CAUTION: Do not alter the Wind River-supplied source configlette files in
installDir/vxworks-6.x/target/config/comps/src.
!
CAUTION: Do not alter Wind River-supplied CDFs directly. Use the naming
convention to create a file whose higher precedence overrides the default
properties of Wind River-supplied components.
2.8.2 CDF Precedence and CDF Installation
More than one CDF may define a given component and its properties. The
precedence of multiple definitions is determined by a numbering scheme used
with the CDF naming convention and by the order in which directories containing
CDFs are read by the configuration facility.
CDF File Naming and Precedence
Wind River reserves the first 50 numbers, that is 00fileName.cdf through
49fileName.cdf. The remaining numbers, 50 through 99, may be used by third
parties for their components.
Higher number files have greater precedence than lower number files (when the
remainder of the name is the same); for example, the definition of the
INCLUDE_FOO component in 66comp_foo.cdf override those in 55comp_foo.cdf.
75
2
VxWorks
Kernel Programmer's Guide, 6.6
This method of setting precedence allows project, BSP, and CPU
architecture-specific overrides of generic components or parameters.
For example, if a BSP provider wanted to change the maximum number of files
that can be open (NUM_FILES) to 75 from the default value of 50, it can be done in
a BSP-specific CDF file with a higher file number that has the following entry in it:
Parameter NUM_FILES {
DEFAULT 75
}
CDF Directories and Precedence
The component description files are read at two points by the configuration
facility:
■
When a project is created.
■
After component description files are changed and the project build occurs.
The order in which CDFs are read is significant. If more than one file describes the
same property of the same component, the one read last overrides all earlier ones.
The intent is to allow component description files to have some precedence level.
Files read later have higher precedence than those read earlier.
Precedence is established in two complementary ways:
■
CDF files reside in certain directories, and those directories are read in a
specified order.
■
Within one of these directories, CDFs are read in alphanumeric order.
The configuration facility sources all .cdf files in any of the following directories.
These directories are read in the order in which they are presented:
1.
installDir/vxworks-6.x/target/config/comps/vxWorks
Contains all generic VxWorks components.
2.
installDir/vxworks-6.x/target/config/comps/vxWorks/arch/arch
Contains all architecture-specific VxWorks components (or component
overrides).
3.
installDir/vxworks-6.x/target/config/bspName
Contains all BSP-specific components.
4.
the project directory
Contains all other components.
76
2 Kernel
2.8 Custom VxWorks Components and CDFs
Within a directory, to control alphanumeric order, a two digit number is
prepended to each .cdf file to determine the order of precedence within a given
directory. See 2.8.5 CDF Naming Conventions, p.80.
Elements of Components in a VxWorks Installation Tree
Wind River delivers the parts of its components in the following locations:
■
Source code modules are usually found in the
installDir/vxworks-6.x/target/src or target/config directories.
■
Headers are found in installDir/vxworks-6.x/target/h; object modules are
delivered in installDir/vxworks-6.x/target/lib/objARCH.
■
Component description files are in
installDir/vxworks-6.x/target/config/comps/vxWorks.
■
Component configlettes (source fragments) are in
installDir/vxworks-6.x/target/config/comps/src.
Third parties are not limited to this arrangement, and the location of component
elements can be fully described in the component description file.
Adding New CDFs to the VxWorks Installation Tree
If you have created a new CDF, you must place it in the appropriate path, based
on the nature of the contents and the desired level of precedence. Your choices of
paths (as described in CDF Directories and Precedence, p.76) are as follows:
■
installDir/vxworks-6.x/target/config/comps/vxWorks for generic VxWorks
component descriptions only
■
installDir/vxworks-6.x/target/config/comps/vxWorks/arch/arch for
architecture-specific VxWorks component descriptions
■
installDir/vxworks-6.x/target/config/config/bspName for board-specific
component descriptions
■
the project directory for all other files
Wind River recommends that third parties place their component source code and
object elements in a directory, such as
installDir/vxworks-6.x/target/config/vendorName. The location of the component
description file (CDF) depends on where in the system the components should be
integrated.
To be able to integrate a new general-purpose VxWorks component into the
system, the CDF must be located in
77
2
VxWorks
Kernel Programmer's Guide, 6.6
installDir/vxworks-6.x/target/config/comps/vxWorks. If it is a BSP-specific
component, the file should be located in the BSP directory. If it is specific to a single
project, it should be located in the project directory
(installDir/vxworks-6.x/target/proj/projectName).
Be sure to follow the proper naming and numbering conventions, which are
described in 2.8.5 CDF Naming Conventions, p.80.
2.8.3 Testing New Components
There are several tests that can run to verify that components have been written
correctly:
■
Check Syntax and Semantics
This vxprj command provides the most basic test:
vxprj component check [projectFile] [component ... ]
For example:
% vxprj component check MyProject.wpj
If no project file is specified, vxprj looks for a .wpj file in the current directory.
If no component is specified, vxprj checks every component in the project. This
command invokes the cmpTest routine, which tests for syntactical and
semantic errors
Based on test output, make any required modifications. Keep running the
script until you have removed the errors.
■
Check Component Dependencies
You can test for scalability bugs in your component by running a second vxprj
command, which has the following syntax:
vxprj component dependencies [projectFile] component [component ... ]
For example, the following command displays a list of components required
by INCLUDE_OBJ_LIB, as well as those that require the component:
% vxprj component dependencies INCLUDE_OBJ_LIB
If no project file is specified, vxprj looks for a .wpj file in the current directory.
■
Check the Component Hierarchy in Workbench
Verify that selections, folders, and new components you have added are
properly included by making a visual check of the Workbench component
hierarchy.
78
2 Kernel
2.8 Custom VxWorks Components and CDFs
Look at how your new elements appear in the folder tree. Check the
parameters associated with a component and their parameter default values.
If you have added a folder containing components, and have included that
folder in your configuration, the Workbench component hierarchy should
display in boldface all components listed as defaults for that folder (that is,
values for the DEFAULTS property).
■
Build and Boot the System
Verify that the resulting image builds and boots.
2.8.4 Component Description Language
The component description language (CDL) is used in component description files
(CDFs) to describe software components. Each CDF has the suffix .cdf. A single file
can define more than one component.
The Component Descriptor Language has several classes of objects, each of which
has various properties.
Three of the object classes are used to define a software component, its
configuration options, and its initialization routine. These objects are discussed in
the following sections:
■
2.8.6 Component CDF Object, p.81
■
2.8.7 Parameter CDF Object, p.87
■
2.8.8 Initialization Group CDF Object, p.88
Two classes of objects are used to group components into a single unit for ease of
configuration. It is described in the following section:
■
2.8.9 Bundle CDF Object, p.90
■
2.8.10 Profile CDF Object, p.91
Two additional classes of objects are used to control the display of components in
the Workbench kernel configuration facility. They are described in the following
sections:
■
2.8.11 Folder CDF Object, p.92
■
2.8.12 Selection CDF Object, p.94
For information about naming conventions, see 2.8.5 CDF Naming Conventions,
p.80.
79
2
VxWorks
Kernel Programmer's Guide, 6.6
Based on the operating system components selected by the user, the configuration
facilities (Workbench or vxprj) create the system configuration files prjComps.h,
prjParams.h, and prjConfig.c, which are used in building the specified system.
For information about how the configuration and code generation facilities work,
see the Wind River Workbench User’s Guide and the VxWorks Command-Line Tools
User’s Guide.
2.8.5 CDF Naming Conventions
Follow these conventions when creating CDFs, where FOO (or Foo) is the variable
element of the naming convention:
■
All bundles names are of the form BUNDLE_FOO.
■
All component names are of the form INCLUDE_FOO.
■
All VxBus driver component names are of the form
INCLUDE_driverType_driverName (for example, DRV_SIO_NS16550).
■
All folders names are of the form FOLDER_FOO.
■
All selection names are of the form SELECT_FOO.
■
Parameter names should not match the format of any other object type, but are
otherwise unrestricted. For example, you can use FOO_XXX, but not
INCLUDE_FOO.
■
All initialization group names should be of the form initFoo. However, Wind
River initialization groups use the form usrFoo for backwards compatibility
reasons.
■
All component description files have a .cdf suffix.
■
All .cdf file names begin with two decimal digits; for example,
00comp_foo.cdf. These first two digits control the order in which .cdf files are
read within a directory. See 2.8.2 CDF Precedence and CDF Installation, p.75 for
more information.
Note that more than one component can be defined in a single CDF.
New component description files should be independent of existing files, with two
exceptions:
■
80
New component objects should be bound into an existing folder or selection
for GUI display purposes.
2 Kernel
2.8 Custom VxWorks Components and CDFs
■
New component object initialization routines must be associated with an
existing initialization group.
A new component object can be bound to an existing folder or selection, and to an
existing initialization group, without modifying the existing elements. By
prepending an underscore (“_”) to certain component properties, you can reverse
the meaning of the property. For example, if there is already a folder
FOLDER_EXISTING and an initialization group initExisting, you can bind a new
component (which is defined in a different file) to it as follows:
Component
INCLUDE_FOO
...
_CHILDREN
FOLDER_EXISTING
_INIT_ORDER
initExisting
}
The property _CHILDREN has the opposite relationship to FOLDER_EXISTING as
CHILDREN; that is, _CHILDREN identifies the parent. In other words, it produces
the same effect as FOLDER_EXISTING having the component INCLUDE_FOO on
its list of children—without any modifications being made to the CDF containing
the FOLDER_EXISTING object.
Note that INIT_BEFORE can be used _INIT_ORDER to define exactly where in the
initialization group the component should be placed.
2.8.6 Component CDF Object
Components are the basic units of configurable software. They are the smallest,
scalable unit in a system. With either Workbench or the command-line vxprj
facility, a user can reconfigure VxWorks by including or excluding a component,
as well as modifying some of its characteristics. The properties of a component
include:
■
Identification of the object code (modules) and source code (configlettes) used
in the build of a project.
■
Identification of configuration parameters, which are typically preprocessor
macros used within a component’s configuration code.
■
Integration information that controls how a component is integrated into an
executable system image (for example, an initialization routine).
For information about how to group components in to a bundle of components, see
2.8.9 Bundle CDF Object, p.90.
81
2
VxWorks
Kernel Programmer's Guide, 6.6
For information about displaying information about components in the
Workbench kernel configuration facility, see 2.8.11 Folder CDF Object, p.92 and
2.8.12 Selection CDF Object, p.94.
Component Properties
The component object class defines the source and object code associated with a
component, much of the integration information, and any associated parameters.
Dependencies among components can be detected and the related components can
be automatically added by the configuration facility (GUI or CLI). It does so by
analyzing the global symbols in each object module that belongs to the component,
and then determining which other components provide the functionality.
As an example, a message logging component could be defined in CDL as follows:
Component
NAME
SYNOPSIS
MODULES
INIT_RTN
CFG_PARAMS
HDR_FILES
}
INCLUDE_LOGGING {
message logging
Provides logMsg support
logLib.o
logInit (consoleFd, MAX_LOG_MSGS);
MAX_LOG_MSGS
logLib.h
For illustrative purposes, in another configuration system using #ifdef/#endif
statements, the definition of the logging component might look like this:
#ifdef INCLUDE_LOGGING
logInit (consoleFd, MAX_LOG_MSGS);
#endif /* INCLUDE_LOGGING */
The component object includes the greatest number of properties. The more
commonly used properties are described below.
NAME
The name that appears next to the component icon in the component tree of
the Workbench kernel configuration facility.
SYNOPSIS
A brief description of the component’s functionality, which is used in
Workbench.
The next four properties are all related to the configuring of code in a project:
82
2 Kernel
2.8 Custom VxWorks Components and CDFs
MODULES
The names of the object files that constitute the component’s code (along with
any source file configlettes). For example, semShow.o is identified in the
MODULES property of the INCLUDE_SEM_SHOW component.
!
CAUTION: An object file must only be associated with one component, and its
name must be globally unique (across the entire system).
CONFIGLETTES
Configlettes are source file fragments used in conjunction with parameters
(see CFG_PARAMS, p.83 and 2.8.7 Parameter CDF Object, p.87). They are
typically used for parameter switches (for example, to specify a number, a
binary setting to true or false, and so on).
Configlette definitions may include the use of macros. For example:
CONFIGLETTES
$(MY_PARAM)\myFile.c
Where MY_PARAM may either be a build macro, or an environment variable.
The default directory for configlettes is
installDir/vxworks-6.x/target/config/comps/src.
HDR_FILES
Header files associated with your configlette code or initialization routine.
These are header files that must be included in order for your configlette or
initialization routine to compile.
ARCHIVE
The archive file in which to find object modules stored other than in the
standard location.
The following property provides configuration information:
CFG_PARAMS
A list of configuration parameters associated with the component, typically a
list of preprocessor macros. Each must be described separately by its own
parameter object. Also see CONFIGLETTES, p.83 and 2.8.7 Parameter CDF
Object, p.87.
The next group of properties control integration of the component into the system
image, including initialization and dependency information.
INIT_RTN
A one-line initialization routine. Also see INIT_BEFORE, p.84, _INIT_ORDER,
p.84, and 2.8.8 Initialization Group CDF Object, p.88.
83
2
VxWorks
Kernel Programmer's Guide, 6.6
LINK_SYMS
A list of symbols to look up in order to include components from an archive.
REQUIRES
A list of component(s) that do not otherwise have structural dependencies and
must be included if this component is included. List only those components
that cannot be detected from MODULES; that is, by their associated object files.
For example, components with configlettes only or those that reference other
components through function pointers. In this latter case, REQUIRES can
specify a selection.
INCLUDE_WHEN
Sets a dependency to automatically include the specified component(s) when
this component is included (that is, it handles nested includes).
INIT_BEFORE
Call the initialization routine of this component before the one specified by this
property. This property is effective only in conjunction with _INIT_ORDER.
Also see INIT_RTN, p.83, _INIT_ORDER, p.84, and 2.8.8 Initialization Group
CDF Object, p.88.
_INIT_ORDER
The component belongs to the specified initialization group. This property places
the specified component at the end of the INIT_ORDER property list of the
initialization group object. Also see INIT_RTN, p.83, INIT_BEFORE, p.84, and
2.8.8 Initialization Group CDF Object, p.88. Like NAME and SYNOPSIS before them,
the following properties also affect user presentation in Workbench:
HELP
List reference pages associated with the component.
_CHILDREN
This component is a child component of the specified folder (or selection). Also
see CHILDREN, p.93.
_DEFAULTS
This component is a default component of the specified folder (or selection).
This property must be used in conjunction with _CHILDREN.
Normally, the component would be listed in the container (folder or selection)
as CHILDREN & DEFAULT. The usage of the complement allows adding
components to an existing container without modifying the container's CDF.
84
2 Kernel
2.8 Custom VxWorks Components and CDFs
Component Template
Component
NAME
component {
// required for all components
name
// readable name (e.g., "foo manager").
// should be in all lower case.
SYNOPSIS
desc
// one-line description
MODULES
m1 m2 ..
//
//
//
//
//
//
//
//
//
//
//
object modules making up the service.
used to generate dependency
information.
it is important to keep this list
small,since the tool’s dependency
engine assumes that the component is
included if *any* of the modules are
dragged in by dependency. It may make
sense to split a large number of
modules into several distinct
components.
CONFIGLETTES
1 s2 ..
//
//
//
//
//
source files in the component that are
#included in the master configuration
file.
file paths are assumed to be relative
to $(WIND_BASE)/target/config/comps/src
HDR_FILES
h1 h1 ..
//
//
//
//
header files that need to be included
to use this component. Typically
contains prototypes for the
initialization routine.
CFG_PARAMS
p1 p2 ..
//
//
//
//
configuration parameters, typically
macros defined in config[All].h, that
can change the way a component works.
see Parameters, below, for more info.
INIT_RTN
init(..)
// one-line initialization routine.
// if it needs to be more than one line,
// put the code in a CONFIGLETTE.
LINK_SYMS
s1 s2 ..
//
//
//
//
//
reference these symbols in order to drag
in the component from the archive.
this tells the code generator how to
drag in components that don’t need to
be initialized.
REQUIRES
r1 r2 ..
//
//
//
//
//
//
//
//
other components required. Note:
dependencies are automatically calculated
based on a components MODULES and
LINK_SYMS. For example, because nfsLib.o
calls rpcLib.o, the tool is able to
figure out that INCLUDE_NFS requires
INCLUDE_RPC. One only needs to list
requirements that cannot be detected from
85
2
VxWorks
Kernel Programmer's Guide, 6.6
HELP
86
h1 h2 ..
//
//
//
//
MODULE dependencies. Typically only needed
for components that don’t have associated
MODULES (e.g., ones with just
configlettes).
//
//
//
//
//
//
//
//
//
//
//
reference pages associated with the
component.
The default is the MODULES and INIT_RTN
of the component. For example, if the
component has MODULE fooLib.o, then the
manual page for fooLib is automatically
associated with the component (if the
manual page exists). Similary for the
components INIT_RTN. If there are any
other relevant manual pages, they can
be specified here.
ARCHIVE
a1
//
//
//
//
//
//
//
//
//
//
//
//
//
//
//
//
//
//
//
//
//
//
//
//
archive in which to find the MODULES.
default is lib$(CPU)$(TOOL)vx.a.
file path is assumed to be relative
to $(WIND_BASE)/target/lib.
any archive listed here is
automatically added to the VxWorks
link-line when the component is
included.
Note: the tool only analyzes archives
associated with included components.
This creates a chicken-and-egg problem,
because the tool analyzes components
before they are actually added.
So if you add a component with an ARCHIVE,
analysis will be done without the ARCHIVE.
As a work-around, if a separate archive is
used, create a dummy component that
lets the tool know that a new archive
should be read. Such a component
should be called INSTALL_something.
It should contain only NAME, SYNOPSIS,
and ARCHIVE attributes. Only after the
user adds it can he or she add other
components from the archive.
INCLUDE_WHEN
c1 c2 ..
//
//
//
//
//
//
//
//
//
automatically include this component
when some other components are included.
All listed components must be included to
activate the INCLUDE_WHEN (AND
relationship). This allows, for example,
msgQShow to be included whenever msgQ and
show are included. Similarly, WDB fpp
support can be included when WDB and fpp
are included.
INIT_BEFORE
c1
// if component c1 is present, our init
// routine must be called before c1.
// Only needed for component releases.
2 Kernel
2.8 Custom VxWorks Components and CDFs
_CHILDREN
_INIT_ORDER
fname
// Component is a child of folder or
// selection fname.
gname
//
//
//
//
Component is a member of init group gname
and is added to the end of the
initialization sequence by default (see
INIT_BEFORE).
}
!
WARNING: The _INIT_ORDER and _CHILDREN elements should refer to elements
that have already been defined. If a component is a _CHILDREN element of a folder
that does not exist, it will not be displayed in the IDE.
2.8.7 Parameter CDF Object
Parameters are one of the primary means of configuring a component. Typically,
one or more parameters control a component’s features.
Also see CONFIGLETTES, p.83 and CFG_PARAMS, p.83.
For example, the following parameter defines the maximum number of log
messages in the logging queue:
Parameter
NAME
TYPE
DEFAULT
}
MAX_LOG_MSGS {
max # queued messages
uint
50
This is comparable to the C code macro assignment:
#define MAX_LOG_MSGS
50
Parameter Properties
The following properties describe a parameter:
NAME
The name that appears in the Workbench.
TYPE
The type of parameter, which can be defined as int, uint, bool, string, exists,
or untyped.
When a configuration parameter is defined as type of exists the prjParams.h
file that is automatically generated as part of creating a project will contain
either an #define or #undef statement for the parameter.
87
2
VxWorks
Kernel Programmer's Guide, 6.6
For example, the kernel configuration parameter
INCLUDE_CONSTANT_RDY_Q has a type of exists. Thus, if it's value is set to
TRUE (which is the default), the following will appear in the auto-generated
prjParams.h file:
#define INCLUDE_CONSTANT_RDY_Q
If the value is set to FALSE, the following will appear:
#undef INCLUDE_CONSTANT_RDY_Q
DEFAULT
The default value of the parameter.
Parameter Template
parameter
name
{
// readable name (e.g., "max open files")
SYNOPSIS
desc
// one-line description.
STORAGE
storage
// MACRO, REGISTRY, ... Default is MACRO.
TYPE
type
//
//
//
//
type of parameter:
int, uint, bool, string, exists.
Default is untyped.
more types will be added later.
DEFAULT
value
//
//
//
//
//
//
default value of the parameter.
default is none - in which case the user
must define a value to use the component.
for parameters of type "exists," the
value is TRUE if the macro should be
defined, or FALSE if undefined.
Parameter
NAME
}
2.8.8 Initialization Group CDF Object
A component must include information that controls how it is integrated into an
executable system image. This means that an initialization routine must be
identified, as well as the point in the initialization sequence that the routine should
be called. Initialization groups assemble related components for initialization and,
thus, define the system startup sequence. The definition of the initialization group
hierarchy is related only to run-time behavior, and has no impact on GUI
presentation.
Also see INIT_RTN, p.83, INIT_BEFORE, p.84, and _INIT_ORDER, p.84.
88
2 Kernel
2.8 Custom VxWorks Components and CDFs
An initialization group is a routine from which components and other
initialization groups are called. The code in the routine is determined by the
included components and their initialization code fragments. For example:
InitGroup
INIT_RTN
SYNOPSIS
INIT_ORDER
usrIosExtraInit {
usrIosExtraInit();
extended I/O system
INCLUDE_EXC_TASK \
INCLUDE_LOGGING \
INCLUDE_PIPES \
...
}
In this example, the components INCLUDE_EXC_TASK, INCLUDE_LOGGING, and
INCLUDE_PIPES are part of the initialization group usrIosExtraInit. If those
components are included, then their initialization routines are called in the order
specified as part of this initialization group.
The code in an initialization group is synthesized by the configuration facility into
the file prjConfig.c, which always part of every BSP project. To see a system’s
initialization sequence, build the project and examine the generated code in
prjConfig.c.
Initialization Group Properties
The following properties describe an initialization group:
NAME
The NAME property can be used to provide a short name for the initialization
group; however, it does not appear anywhere in this release of the GUI.
SYNOPSIS
The SYNOPSIS property is used to provide a brief, readable definition of the
initialization group.
INIT_RTN
The initialization routine that initializes an associated component(s).
INIT_ORDER
Components and initialization groups belonging to this initialization group
listed in the order in which they are to be initialized.
89
2
VxWorks
Kernel Programmer's Guide, 6.6
Initialization Group Template
group
rtn(..)
{
// initialization routine definition
INIT_ORDER
i1 i2 ..
// ordered list of init groups and
// components that belong to this init group
INIT_AFTER
i2 i2 ..
// Only needed for component releases.
InitGroup
INIT_RTN
}
2.8.9 Bundle CDF Object
Bundle object can be used to associate components that are often used together,
which facilitates configuration of the operating system with generic sets of
facilities.
For example, the network kernel shell (BUNDLE_NET_SHELL) includes all the
components are required to use the kernel shell with a network symbol table:
Bundle BUNDLE_NET_SHELL {
NAME network kernel shell
SYNOPSIS Kernel shell tool with networking symbol table
HELP shell windsh tgtsvr
COMPONENTS INCLUDE_SHELL \
INCLUDE_LOADER \
INCLUDE_DISK_UTIL \
INCLUDE_SHOW_ROUTINES \
INCLUDE_STAT_SYM_TBL \
INCLUDE_DEBUG \
INCLUDE_UNLOADER \
INCLUDE_MEM_SHOW \
INCLUDE_SYM_TBL_SHOW \
INCLUDE_CPLUS \
INCLUDE_NET_SYM_TBL
_CHILDREN FOLDER_BUNDLES
}
For information about components, see 2.8.6 Component CDF Object, p.81.
Bundle Properties
NAME
The name of the bundle as it should appear in Workbench.
SYNOPSIS
A description of the bundle’s functionality, as it should appear in Workbench.
90
2 Kernel
2.8 Custom VxWorks Components and CDFs
HELP
HTML help topics that the bundle is related to.
2
COMPONENTS
A list of components to add to the kernel configuration when this bundle is
added.
_CHILDREN
This bundle is a child of the specified folder.
2.8.10 Profile CDF Object
A profile defines a set of components to be used as a starting point to create a
project. The profile can only be used at creation time. When used, the profile
replace any default configuration. Components defined by the profile are added to
the project as if they were added by the user. Therefore, after creation of the
project, there is no trace of the profile in the project. A profile, as such, cannot be
removed. Profiles can be based on other profiles and add to them. There is a special
base profile defined as BSP_DEFAULT that represents the default set of components
defined by each BSP. If BSP_DEFAULT is not listed in PROFILES, then none of the
BSP default components will be included in the project and the profile will replace
the default configuration.
Profile Properties
NAME
The name of the profile as it should appear in Workbench.
SYNOPSIS
A brief description of the profiles functionality, which is used in Workbench.
PROFILES
A list of base profiles.
COMPONENTS
A list of components to add to the kernel configuration when this profile is
added.
Profile Template
Profile
NAME
profileName {
nameForGui
91
VxWorks
Kernel Programmer's Guide, 6.6
SYNOPSIS
PROFILES
COMPONENTS
description
p1 p2 p3...
comp1 comp2 comp3...
}
2.8.11 Folder CDF Object
A folder is used to control the display of components in the Workbench kernel
configuration facility.
The Workbench component hierarchy uses folders to group components logically.
Instead of presenting all components in a flat list, it presents them hierarchically.
For example, top-level folders might organize components under headings such as
network, drivers, OS, and so on.
Folder objects provide a directory-type hierarchy for grouping components that
are logically related. Folders allows for the graphical presentation of useful
information about each component, such as:
■
any hierarchical grouping it has with related components
■
its associated dependencies on other components
■
its configurable properties
■
its integration into an existing system (for component releases)
Folder information only affects user presentation. It is not related to initialization
group hierarchy, and thereby the startup sequence, which depend solely on a
component’s initialization group. Folders can contain one or more components,
selections, and other folders; they do not have parameter or initialization group
objects associated with them.
Folders can contain more than one component. For example, the ANSI
functionality contained in a folder might be described as follows:
92
2 Kernel
2.8 Custom VxWorks Components and CDFs
Folder
FOLDER_ANSI {
NAME
SYNOPSIS
CHILDREN
DEFAULTS
ANSI C components (libc)
ANSI libraries
INCLUDE_ANSI_ASSERT
\
INCLUDE_ANSI_CTYPE
\
INCLUDE_ANSI_LOCALE
\
INCLUDE_ANSI_MATH
\
INCLUDE_ANSI_STDIO
\
INCLUDE_ANSI_STDLIB
\
INCLUDE_ANSI_STRING
\
INCLUDE_ANSI_TIME
\
INCLUDE_ANSI_STDIO_EXTRA
INCLUDE_ANSI_ASSERT INCLUDE_ANSI_CTYPE \
INCLUDE_ANSI_MATH INCLUDE_ANSI_STDIO \
INCLUDE_ANSI_STDLIB INCLUDE_ANSI_STRING \
INCLUDE_ANSI_TIME
}
Folders offer great flexibility in grouping components. They allow groups to
include more than just a default set of components that are included together.
Components can be added and removed from the configuration individually (as
defined by the CHILDREN property of the folder object).
Folder Properties
The following properties describe a folder:
NAME
The name that appears next to the folder icon in the Workbench component
hierarchy.
SYNOPSIS
A brief description of the folder.
CHILDREN
Components, folders, and selections belonging to this folder are called
children. Also see _CHILDREN, p.84.
DEFAULTS
The default component(s) that would be included if the folder were added
without any overrides. Folder groupings can affect configuration when a
folder is added, because components specified by a folder’s DEFAULTS
property are added all at once.
93
2
VxWorks
Kernel Programmer's Guide, 6.6
NOTE: Use folder objects only when creating a new grouping of components (new
or existing). Do not modify existing folders in order to include new components.
CDL accommodates that by prepending an underscore to a property name, for
example, _CHILDREN.
Folder Template
folder
name
{
// readable name (e.g., "foo libraries").
SYNOPSIS
desc
// one-line description
CHILDREN
i1 i2 ..
// containers and components
// that belong to this container.
DEFAULTS
i1 i2 ..
//
//
//
//
//
//
//
//
//
Folder
NAME
default CHILDREN.
if the folder represents a complex
subsystem (such as the WDB agent),
this is used to suggest to the user
which components in the folder are
considered "default." That way the user
can add the whole subsystem at once,
and a reasonable set of subcomponents
will be chosen.
}
2.8.12 Selection CDF Object
A selection is used to display alternative options in the Workbench kernel
configuration facility.
They are similar to folders, but they are components that implement a common
interface, for example, serial drivers, the timestamp mechanism, and the WDB
communication interface. These components provide alternatives for the same
service, and one or more can be selected for configuration in a single project. The
selections CDL class is comparable to #ifdef/#else constructs in other
configuration systems.
Selection information, like that for folders, is for user presentation only. It is not
related to initialization group hierarchy, and thereby the startup sequence.
Selections contain one or more components only.
Selections behave like folders, except they add a count for a range of available
components; for example, a selection containing three components might tell the
user only one can be configured at a time, or perhaps two of three. Because of the
94
2 Kernel
2.8 Custom VxWorks Components and CDFs
count, selections do not contain folders or other selections; nor do they have
parameter or initialization group objects associated with them.
2
Selection Properties
The following properties describe a selection:
NAME
A readable name, the one that appears next to the selection icon in the
Workbench component hierarchy.
SYNOPSIS
A brief description of the selection.
COUNT
Set a minimum and maximum count from available options.
CHILDREN
Components from which to select.
DEFAULTS
The default component(s), depending on the count.
For example, the following example provides for selection from a set of timestamp
drivers, for use with the timestamp component:
Selection
NAME
COUNT
CHILDREN
DEFAULTS
SELECT_TIMESTAMP {
select timestamping
1-1
INCLUDE_SYS_TIMESTAMP
\
INCLUDE_USER_TIMESTAMP \
INCLUDE_SEQ_TIMESTAMP
INCLUDE_SEQ_TIMESTAMP
}
There are three timestamp drivers available, as indicated by the three values for
the CHILDREN property. The COUNT permits a choice of one, that is, a minimum
and a maximum of 1.
Selection Template
Selection
NAME
SYNOPSIS
selection
name
{
// readable name (for example , "foo
// communication path")
desc
// one-line description
95
VxWorks
Kernel Programmer's Guide, 6.6
COUNT
min-max
// range of allowed subcomponents.
// 1-1 means exactly one.
// 1- means one or more.
CHILDREN
i1 i2 ..
// components from which to select
DEFAULTS
i1 i2 ..
// default CHILDREN.
// this is not used for anything except to
// to suggest to the user which components
// in the selection we consider "default."
}
2.8.13 CDF Template
This template includes all objects and properties that can be used in a CDF.
Generally, CDFs use only a portion of them.
Bundle MY_BUNDLE
{
COMPONENTS :
HELP :
NAME :
SYNOPSIS :
_CHILDREN :
}
Component MY_COMPONENT
{
ARCHIVE :
CFG_PARAMS :
CONFIGLETTES :
ENTRY_POINTS :
EXCLUDE_WHEN :
HDR_FILES :
HELP :
HIDE_UNLESS :
INCLUDE_WHEN :
INIT_AFTER :
INIT_BEFORE :
INIT_RTN :
LINK_DATASYMS :
LINK_SYMS :
MODULES :
NAME :
PREF_DOMAIN :
PROJECT :
PROTOTYPE :
REQUIRES :
SHUTDOWN_RTN :
SYNOPSIS :
TERM_RTN :
USES :
_CHILDREN :
96
2 Kernel
2.8 Custom VxWorks Components and CDFs
_COMPONENTS :
_DEFAULTS :
_EXCLUDE_WHEN :
_HIDE_UNLESS :
_INCLUDE_WHEN :
_INIT_AFTER :
_INIT_BEFORE :
_INIT_ORDER :
_LINK_DATASYMS :
_LINK_SYMS :
_REQUIRES :
_USES :
}
2
EntryPoint MY_ENTRYPOINT
{
NAME :
PRIVILEGED :
SYNOPSIS :
TYPE :
_ENTRY_POINTS :
}
EntryPointType MY_ENTRYPOINTTYPE
{
SYNOPSIS :
_TYPE :
}
Folder MY_FOLDER
{
CHILDREN :
DEFAULTS :
HELP :
NAME :
SYNOPSIS :
_CHILDREN :
_DEFAULTS :
}
InitGroup MY_INITGROUP
{
HELP :
INIT_AFTER :
INIT_BEFORE :
INIT_ORDER :
INIT_RTN :
NAME :
PROTOTYPE :
SHUTDOWN_RTN :
SYNOPSIS :
TERM_RTN :
_INIT_AFTER :
_INIT_BEFORE :
_INIT_ORDER :
}
97
VxWorks
Kernel Programmer's Guide, 6.6
Module MY_MODULE
{
ENTRY_POINTS :
NAME :
SRC_PATH_NAME :
_MODULES :
}
Parameter MY_PARAMETER
{
DEFAULT :
HELP :
NAME :
STORAGE :
SYNOPSIS :
TYPE :
VALUE :
_CFG_PARAMS :
}
Profile MY_PROFILE
{
COMPONENTS :
HELP :
NAME :
SYNOPSIS :
_CHILDREN :
}
Selection MY_SELECTION
{
ARCHIVE :
CFG_PARAMS :
CHILDREN :
CONFIGLETTES :
COUNT :
DEFAULTS :
HDR_FILES :
HELP :
HIDE_UNLESS :
INIT_AFTER :
INIT_BEFORE :
INIT_RTN :
LINK_DATASYMS :
LINK_SYMS :
MODULES :
NAME :
PROTOTYPE :
REQUIRES :
SHUTDOWN_RTN :
SYNOPSIS :
USES :
_CHILDREN :
_DEFAULTS :
_HIDE_UNLESS :
98
2 Kernel
2.8 Custom VxWorks Components and CDFs
_INIT_AFTER :
_INIT_BEFORE :
_INIT_ORDER :
_LINK_DATASYMS :
_LINK_SYMS :
_REQUIRES :
_USES :
}
2
Symbol MY_SYMBOL
{
_LINK_DATASYMS :
_LINK_SYMS :
}
99
VxWorks
Kernel Programmer's Guide, 6.6
2.9 Custom System Calls
The VxWorks system call interface provides kernel services for applications that
are executed as processes in user space. The interface can be easily extended
developers who wish to add custom system calls to the operating system to
support special needs of their applications. (See the VxWorks Application
Programmer’s Guide: Applications and Processes for information about user-space
applications.)
Initially, the developer’s main tasks in extending the system call interface are
designing the custom system call in accordance with the naming, numbering, and
argument guidelines, and then writing the system call handler to support that
design. See 2.9.2 System Call Requirements, p.101 and 2.9.3 System Call Handler
Requirements, p.105.
The system call interface can then be extended either statically and dynamically.
Static extension involves the use of configuration files and build system facilities
to create a VxWorks system image that includes the new system call functionality.
Dynamic extension involves using the host or kernel shell, and kernel object
module loader, to download a development version of the system call handler to
the kernel. See 2.9.4 Adding System Calls, p.107 and 2.9.5 Monitoring And Debugging
System Calls, p.114.
2.9.1 How System Calls Work
System calls are C-callable routines. They are implemented as short pieces of
assembly code called system call stubs. The stubs execute a trap instruction, which
switches execution mode from user mode to kernel mode. All stubs are identical
to each other except for the unique system call number that they pass to the kernel
to identify the system call.
In kernel mode, a trap handler copies any system call arguments from the user
stack to the kernel stack, and then calls the system call handler.
Each system call handler is given only one argument—the address of its argument
array. Handler routines interpret the argument area as a structure whose members
are the arguments.
System call handlers may call other routines in the kernel to service the system call
request. They must validate the parameters of the system call, and return errors if
necessary.
100
2 Kernel
2.9 Custom System Calls
The architecture of the system call dispatcher allows system call handlers to be
installed at either compile time or run-time.
2
2.9.2 System Call Requirements
In order to be able to generate system calls automatically, as well as to ensure
proper run-time operation, system calls must adhere strictly to naming,
numbering, argument, and return value rules.
System Call Naming Rules
The names of various elements associated with a system call must derive their
names from that of the system call itself. It is important to adhere to this
convention in order to avoid compilation errors when using the automated
mechanisms provided for adding system calls. See Table 2-7.
Table 2-7
System Call Naming Conventions
Element
Name Convention
system call
sysCallName( )
system call stub
SYSCALL_STUB_sysCallName.s
system call handler routine
sysCallNameSc( )
system call argument structure
sysCallNameScArgs
The system call name is used by developer in system call definition files. The
system call stub is generated automatically from the information in the definition
files. The developer must write the system call handler routine, which includes the
system call argument structure.
For example, if the name of the system call is foo( ), then:
■
The system call stub is named SYSCALL_STUB_foo.s.
The stub implements the routine foo( ) in user mode.
■
The system call handler routine for system call foo must be named fooSc( ).
Routine fooSc( ) is called when an application makes a call to foo( ) in user
space. Writing a routine with this name is the kernel developer’s
101
VxWorks
Kernel Programmer's Guide, 6.6
responsibility. Unless fooSc( ) exists, an error will be generated when the
kernel is rebuilt.
■
If the foo system call takes at least one argument, the argument structure for
foo must be declared as struct fooScArgs in the system call handler.
For information about system call handler requirements, see 2.9.3 System Call
Handler Requirements, p.105. For information about adding system calls to the
operating system—both statically and dynamically—see 2.9.4 Adding System Calls,
p.107.
System Call Numbering Rules
Each system call must have a unique system call number. The system call number
is passed by the system call stub to the kernel, which then uses it to identify and
execute the appropriate system call handler.
A system call number is a concatenation of two numbers:
■
the system call group number
■
the routine number within the system call group
The group number is implemented as a ten-bit field, and the routine number as a
six-bit field. This allows for up to 1024 system call groups, each with 64 routines in
it. The total system-call number-space can therefore accommodate 65,536 system
calls.
Six system call groups—numbers 2 through 7—are reserved for customer use.
(Customers may request a formal system call group allocation from Wind River.)
All other system call groups are reserved for Wind River use.
!
WARNING: Do not use any system call group numbers other than those reserved
for customer use. Doing so may conflict with Wind River or Wind River partner
implementations of system calls.
Wind River system call group numbers and system call routine numbers are
defined in the syscallNum.def file. It should not be modified by customers.
Customer system calls group numbers and system call routine numbers are
defined in the syscallUsrNum.def file.
The Wind River system call number definition file, and a template for the customer
system call definition file are located in installDir/vxworks-6.x/target/share/h.
102
2 Kernel
2.9 Custom System Calls
A given system call group is simply a collection of related system calls offering
complementary functionality. For example, the VxWorks SCG_STANDARD group
includes system calls that are commonly found in UNIX-like (POSIX) systems, and
the SCG_VXWORKS group includes system calls that are unique to VxWorks or
that are otherwise dissimilar to UNIX-like system calls.
For information about using the system call definition files to generate system
calls, see Adding System Calls Statically, p.107.
System Call Argument Rules
System calls may only take up to eight arguments. Special consideration must be
given to 64-bit arguments on 32-bit systems. Floating point and vector-type
arguments are not permitted.
Wind River system calls are defined in the syscallApi.def file. It should not be
modified by customers.
Customer system calls are defined in the syscallUsrApi.def file. See Adding System
Calls Statically, p.107 for information about editing this file.
Number of Arguments
System calls can take up to a maximum of eight arguments (the maximum that the
trap handler can accommodate). Each argument is expected to be one native-word
in size. The size of a native-word is 32 bits for a 32-bit architecture and 64 bits for
64-bit architectures. For the great majority of system calls (which use 32 bits),
therefore, the number of words in the argument list is equal to the number of
parameters the routine takes.
In cases where more than eight arguments are required the arguments should be
packed into a structure whose address is the parameter to the system call.
64-Bit Argument Issues
64-bit arguments are permitted, but they may only be of the type long long. For
32-bit architectures, a 64-bit argument takes up two native-words on the argument
list, although it is still only one parameter to the routine.
There are other complications associated with 64-bit arguments to routines. Some
architectures require 64-bit arguments to be aligned to either even or odd
numbered registers, while some architectures have no restrictions.
It is important for system call developers to take into account the subtleties of
64-bit argument passing on 32-bit systems. The definition of a system call for
103
2
VxWorks
Kernel Programmer's Guide, 6.6
VxWorks requires identification of how many words are in the argument list, so
that the trap handler can transfer the right amount of data from the user-stack to
the kernel-stack. Alignment issues may make this less than straightforward.
Consider for example, the following routine prototypes:
int foo (int a, int b, long long c, int d);
int bar (int p, long long q, int r);
The ARM and Intel x86 architectures have no alignment constraints for 64-bit
arguments, so the size of the argument list for foo( ) would be five words, while
the size of the argument for bar( ) would be four words.
PowerPC requires long long arguments to be aligned on eight-byte boundaries.
Parameter c to routine foo( ) is already at an eight-byte offset with respect to the
start of the argument list and is hence aligned. So for PowerPC, the argument list
size for foo( ) is five words.
However, in the case of bar( ) the long long argument q is at offset four from the
first argument, and is therefore not aligned. When passing arguments to bar, the
compiler will skip one argument register and place q at offset eight so that it is
aligned. This alignment pad is ignored by the called routine, though it still
occupies space in the argument list. Hence for PowerPC, the argument list for bar
is five words long. When describing a system call such as bar( ), it is thus advised
that the argument list size be set to five for it to work correctly on all architectures.
Consult the architecture ABI documentation for more information. There are only
a few routines that take 64-bit arguments.
System Call Return Value Rules
System calls may return only a native word as a return value (that is, integer values
or pointers, and so on).
64-bit return values are not permitted directly, though they may be emulated by
using private routines. To do so, a system call must have a name prefixed by an
underscore, and it must a pointer to the return value as one of the parameters. For
example the routine:
long long get64BitValue (void)
must have a companion routine:
void _get64BitValue (long long *pReturnValue)
104
2 Kernel
2.9 Custom System Calls
Routine _get64BitValue( ) is the actual system call that should be defined in the
syscallUsrNum.def and syscallUsrApi.def files. The routine get64BitValue( ) can
then be written as follows:
long long get64BitValue (void)
{
long long value;
_get64BitValue (&value);
return value;
}
(The get64BitValue( ) routine would be written by the user and placed in a user
mode library, and the _get64BitValue( ) routine would be generated
automatically; see 2.9.4 Adding System Calls, p.107.)
The value -1 (ERROR) is the only permitted error return value from a system call.
No system call should treat -1 as a valid return value. When a return value of -1 is
generated, the operating system transfers the errno value correctly across the trap
boundary so that the user-mode code has access to it.
If NULL must be the error return value, then the system call itself must be
implemented by another routine that returns -1 as an error return. The -1 value
from the system call can then be translated to NULL by another routine in user
mode.
2.9.3 System Call Handler Requirements
System call handlers must adhere to naming conventions, and to organizational
requirements for the system call argument structure. They should validate
arguments. If an error is encountered, they set errno and return ERROR.
A system call handler typically calls one or more kernel routines that provide the
functionality required. In some cases, the code will call the public kernel API
directly; in other cases, it may do otherwise to skip the kernel level validation, and
call the underlying functionality directly.
!
CAUTION: In order to enforce isolation between kernel and user space, not all
kernel APIs may be called from a system call handler. In particular, APIs cannot be
called if their operation involves passing a user-side task ID or an RTP ID. APIs
also cannot be called to create objects in the kernel if those APIs are already directly
accessible in user space by way of the standard system call interface. Examples
include taskSpawn( ), taskCreate( ), msgQCreate( ), pthread_create( ), the
various semaphore creation routines, and so on.
105
2
VxWorks
Kernel Programmer's Guide, 6.6
System Call Handler Naming Rules
System call handlers must be named in accordance with the system call naming
conventions, which means that they must use the same name as the system call,
but with an Sc appended. For example, the foo( ) system call must be serviced by
the fooSc( ) system call handler.
All system call handlers take a single parameter, which is a pointer to their
argument structure. The argument structure must also be named in accordance
with the system call naming conventions, which means that they must use the
same name as the system call handler, but with Args appended. For example, the
argument structure for fooSc must be declared as struct fooScArgs.
For example, the write( ) system call is declared as:
int write (int fd, char * buf, int nbytes)
The system call handler routine for write is therefore named writeSc( ), and it is
declared as:
int writeSc (struct writeScArgs * pArgs)
And the argument structure is writeScArgs, which is declared as:
struct writeScArgs
{
int
fd;
char * buf;
int
nbytes;
};
See System Call Naming Rules, p.101.
System Call Handler Argument Validation
A system call handler should validate all arguments. In particular, it should:
■
Bounds-check numerical values.
■
Validate any memory addresses to ensure they are accessible within the
current memory context (that is memory within the process, and not within
the kernel).
See the scMemValidate( ) API reference entry for information on pointer
validation across system calls.
106
2 Kernel
2.9 Custom System Calls
System Call Handler Error Reporting
At the end of the system call, in the case of failure, the system call handler should
ensure errno is set appropriately, and then return -1 (ERROR). If the return value
is -1 (ERROR) the kernel errno value is then copied into the calling process’ errno.
If there is no error, simply return a value that will be copied to user mode. If the
handlers set their errno before returning ERROR, user mode code sees the same
errno value.
2.9.4 Adding System Calls
System calls can be added both statically and dynamically. This means that they
can be either configured and built into the VxWorks operating system image, or
they can be added interactively to the operating system while it is running on a
target.
Dynamic addition is useful for rapid prototyping and debugging of system calls.
Static configuration is useful for more stable development efforts, and production
systems.
Adding System Calls Statically
The process of adding system calls statically is based on the use of the
syscallUsrNum.def and syscallUsrApi.def system call definition files.
The files define the system call names and numbers, their prototypes, the system
call groups to which they belong, and (optionally) the components with which
they should be associated. The scgen utility program uses these files—along with
comparable files for standard VxWorks system calls—to generate the system call
apparatus required to work with the system call handler written by the developer.
The scgen program is integrated into the build system, and is run automatically
when the build system detects that changes have been made to
syscallUsrNum.def and syscallUsrApi.def.
The template files syscallUsrNum.def.template and syscallUsrApi.def.template
are in installDir/vxworks-6.x/target/share/h. Make copies of files in the same
directory without the .template extension, and create the appropriate entries in
them, as described below.
After you have written a system call handler, the basic steps required to add a new
system call to VxWorks are:
107
2
VxWorks
Kernel Programmer's Guide, 6.6
1.
If you are creating a new system call group, add an entry for the group to
syscallUsrNum.def. See Defining a New System Call Group, p.109. Remember
that only groups 2 through 7 are available to developers; do not use any other
group numbers. (Contact Wind River if you need to have a group formally
added to VxWorks.)
2.
Add an entry to syscallUsrNum.def to assign the system call to a system call
group, and to associate the system call name with a system call number. See
Defining a New System Call, p.110.
3.
Add an entry to syscallUsrApi.def to define the system call name and its
arguments. See Defining a New System Call, p.110.
4.
Write the system call handler routine. See 2.9.3 System Call Handler
Requirements, p.105.
5.
Rebuild the kernel-mode and user-mode source code trees
installDir/vxworks-6.x/target/src and installDir/vxworks-6.x/target/usr/src.
Use the following command in each directory:
make CPU=cpuType TOOL=toolType
This command automatically detects changes in syscallUsrNum.def and
syscallUsrApi.def, invokes the scgen utility, and then rebuilds the source
trees.
What scgen Does
Using the system call definitions in the both Wind River and the customer system
call definition files scgen generates the following:
1.
The files installDir/vxworks-6.x/target/h/syscall.h and
installDir/vxworks-6.x/target/usr/h/syscall.h. The contents of both files are
identical. They define all system call numbers and group numbers in the
system. These files provide information shared between kernel and user space
code.
2.
One system call assembly stub file for each system call. The stubs are placed
into the appropriate architecture directory under
installDir/vxworks-6.x/target/usr/src/arch for compilation into libvx.a or
libc.so.
3.
A file containing argument structures for all system calls in the system. This
file is architecture/ABI specific, and is used by the system call handlers located
in the kernel. This file is named syscallArgsArchAbi.h under
installDir/vxworks-6.x/target/h/arch/archName (for example,
installDir/vxworks-6.x/target/h/arch/ppc/syscallArgsppc.h).
108
2 Kernel
2.9 Custom System Calls
4.
A file containing a pre-initialized system call group table for all system call
groups known at compile-time. This file is
installDir/vxworks-6.x/target/h/syscallTbl.h.
All of this output is then used by the build system automatically; no user
intervention is required to build the appropriated system call infrastructure into
the system.
The scgen utility can also be run from the command line for debugging purposes.
Defining a New System Call Group
If you need to define a new system call group, add it to syscallUsrNum.def using
the following syntax:
SYSCALL_GROUP
SCG_sgcGroupName
groupNum
componentNames
Six system call groups—numbers 2 through 7—are reserved for customer use. All
other system call groups are reserved for Wind River use. (See System Call
Numbering Rules, p.102.) Group names must be unique.
!
WARNING: Do not use any system call group numbers other than those reserved
for customer use. Doing so may conflict with Wind River or Wind River partner
implementations of system calls.
Identification of component names is optional, and provides the means of
associating a system call group (all its calls) with specific operating system
components for inclusion in a VxWorks configuration. It works as follows:
■
If a component name is not defined, the system call group is always included
in the system.
■
If a component is defined, the system call group will either be included in the
system or left out of it—depending on the presence or absence of the
component. That is, if the component is included in a VxWorks configuration
by the user, then the system call group is included automatically. But if the
component is not included in the configuration, the group is likewise not
included.
The fields must be separated by one or more space characters.
For example, a new group called SCG_MY_NEW_GROUP could be defined with the
following entry (where N is the group number selected for use):
SYSCALL_GROUP
SCG_MY_NEW_GROUP
N INCLUDE_FOO
109
2
VxWorks
Kernel Programmer's Guide, 6.6
The system calls that are part of the system call group are identified below the
SYSCALL_GROUP definition line. Up to 64 system calls can be identified within
each group. See Defining a New System Call, p.110.
Defining a New System Call
To define a new system call, you must create entries in two different files:
■
One entry in syscallUsrNum.def, which assigns it to a system call group and
associates the system call name and number.
■
One entry in syscallUsrApi.def, which defines the system call name and its
arguments.
System Call Definition Syntax
To add a system call to a call group, add an entry to syscallUsrApi.def under the
appropriate system call group name, using the following syntax:
sysCallNum sysCallName
Note that it is important to add system calls to the end of a system call group; do
use numbers that have already been assigned. Reusing an existing number will
break binary compatibility with existing binaries; and all existing applications
must be recompiled. System call numbers need not be strictly sequential (that is
there can be gaps in the series for future use).
To define a system call itself, add an entry to syscallUsrApi.def using the
following syntax:
sysCallName numArgs [ argType arg1; argType arg2; argType angN; ] \
CompName INCLUDE headerFileName.h
System call definition lines can be split over multiple lines by using the backslash
character as a connector.
The name of the system call used in syscallUsrApi.def must match the name used
in syscallUsrNum.def.
When defining the number of arguments, take into consideration any 64-bit
arguments and adjust the number accordingly (for issues related to 64-bit
arguments, see System Call Argument Rules, p.103).
The arguments to the system call are described in the bracket-enclosed list. The
opening bracket must be followed by a space; and the closing bracket preceded by
one. Each argument must be followed by a semicolon and then at least one space.
If the system call does not take any arguments, nothing should be listed—not even
the bracket pair.
110
2 Kernel
2.9 Custom System Calls
More than one component name can be listed. If any of the components is included
in the operating system configuration, the system call will be included when the
system is built. (For information about custom components, see 2.8 Custom
VxWorks Components and CDFs, p.67.)
The following mistakes are commonly made when editing syscallUsrApi.def and
syscallUsrNum.def, and can confuse the scgen utility:
■
No space after the opening bracket of an argument list.
■
No space before the closing bracket of an argument list.
■
No backslash at the end of a line (if the argument list continues onto the next
line).
■
An empty pair of brackets that encloses no arguments at all. This will cause the
generated temporary C file to have a compile error.
Bear in mind that there can be no more than 64 routines in any system call group.
If the system call includes the definition of a new type in a header file, the header
file must be identified with the INCLUDE statement. The scgen utility must resolve
all types before generating the argument structures, and this is the mechanism by
which it is informed of custom definitions.
For examples of how this syntax is used, see System Call Definition Example, p.111.
Also consult the Wind River system call definitions files (syscallNum.def and
syscallApi.def), but do not modify these files.
System Call Definition Example
Assume that we want to add the custom system call myNewSyscall( ) to a new
system call group SCG_USGR0 (which is defined in syscallNum.def).
First, create syscallUsrNum.def file by copying syscallUsrNum.def.template.
Then edit the file syscallUsrNum.def, adding a system call group entry for the
appropriate group, and the system call number and name under it. System call
groups 2 through 7 are reserved for customer use; do not use any other group
numbers.
For example:
SYSCALL_GROUP
1 myNewSyscall
SCG_USER0
2
Then we must edit syscallUsrApi.def to define the system call itself.
The C prototype for myNewSyscall( ) is:
int myNewSyscall (MY_NEW_TYPE a, int b, char *c);
111
2
VxWorks
Kernel Programmer's Guide, 6.6
The call has three arguments, and a type defined in a custom header file. Assume
that we also want to implement the system call conditionally, depending on
whether or not the component INCLUDE_FOO is configured into the operating
system.
The entry in syscallUsrApi.def would therefore look like this:
INCLUDE
myNewSyscall 3 [ MY_NEW_TYPE a; int b; char *c; ] INCLUDE_FOO
Adding System Calls Dynamically
You can dynamically extend the system call interface on a target by downloading
a kernel object module that includes code for installing system call handlers as well
as the system call handler routines themselves. You do not need to modify the
system call definition files, to run scgen, or to rebuild the kernel.
This approach is useful for rapid prototyping. It would rarely be useful or
advisable with a deployed system.
System Call Installation Code
The code required to install your system call handlers in the kernel consists of:
■
an initialized table for the system call handler routines
■
a call to a system call registration routine
This code should be included in the same module with the system call handlers.
You must identify a system call group for the system calls, and it should be a group
that is otherwise unused in the target system.
Routine Table
The system call handler routine table is used to register the system call handler
routines with the system call infrastructure when the module is downloaded.
For example, if the system handler routines are testFunc0( ), testFunc1( ),
testFunc2( ), and testFunc3( ),the table should be declared as follows:
_WRS_DATA_ALIGN_BYTES(16) SYSCALL_RTN_TBL_ENTRY testScRtnTbl [] =
{
{(FUNCPTR) testFunc0,
1, "testFunc0",
0}, /* routine
{(FUNCPTR) testFunc1,
2, "testFunc0",
1}, /* routine
{(FUNCPTR) testFunc2,
3, "testFunc0",
2}, /* routine
{(FUNCPTR) testFunc3,
4, "testFunc0",
3} /* routine
}
112
0
1
2
3
*/
*/
*/
*/
2 Kernel
2.9 Custom System Calls
The _WRS_DATA_ALIGN_BYTES(16) directive instructs the compiler/linker to
align the table on a 16-byte boundary. This directive is optional, but is likely to
improve performance as it increases the chance of locating the table data on a cache
line boundary.
Building the Object Module
Build the object module containing the system call handlers and registration code
as you would any module. See 2.6.6 Building Kernel Application Modules, p.62.
Downloading the Module and Registering the System Calls
After you have built the module, download it, register it, and check that
registration has been successful:
1.
Download it to the target system with the debugger, host shell, or kernel shell.
From the shell (using the C interpreter) the module foo.o could be loaded as
follows:
-> ld < foo.o
2.
Register the new handlers with the system call infrastructure before any
system calls are routed to your new handlers. This is done by calling
syscallGroupRegister( ). For example:
-> syscallGroupRegister (2, "testGroup", 4, &testScRtnTbl, 0)
The first argument is a variable holding the group number (an integer); the
second is the group name; the second is the group name; the third is the
number of system handler routines, as defined in the table; the fourth is the
name of the table; and the last is set to that the registration does not forcibly
overwrite an existing entry. (Note that you use the ampersand address
operator with the third argument when you execute the call from the shell—
which you would not do when executing it from a program.)
It is important to check the return value from syscallGroupRegister( ) and
print an error message if an error was returned. See the API reference for
syscallGroupRegister( ) for more information.
3.
Verify that the group is registered by running syscallShow( ) from the shell
(host or kernel).
The system call infrastructure is now ready to route system calls to the newly
installed handlers.
113
2
VxWorks
Kernel Programmer's Guide, 6.6
Making System Calls from a Process
The quickest method of testing a new system call is to create and run a simple RTP
application
First, calculate the system call numbers for your new system calls. In order to do
so, use the SYSCALL_NUMBER( ) utility macro (defined in syscall.h). For example,
if you used group number 2 for your test group and the routine number for
testFunc0( ) is 0 (as described above), then the system call number for this routine
is the value returned by the following call:
SYSCALL_NUMBER (2, 0)
The system call number for testFunc1( ) is the value returned by this call:
SYSCALL_NUMBER (2, 1)
And so on.
To make the actual system calls, the application calls the syscall( ) routine. The first
eight arguments (all integers) are the arguments passed to your system call, and
the ninth argument is the system call number.
For example, to have your user-mode applications to call testFunc0( ) from
process, you should implement testFunc0( ) like this:
int testFunc0
(
int arg1,
int arg2,
int arg3,
int arg4,
int arg5
)
{
return syscall (arg1, arg2, arg3, arg4, arg5, 0, 0, 0,
SYSCALL_NUMBER(2,0));
}
Note that you must use nine arguments with syscall( ). The last argument is the
system call number, and the preceding eight are for the system call arguments. If
your routine takes less than eight arguments, you must use zeros as placeholders
for the remainder.
2.9.5 Monitoring And Debugging System Calls
This section discusses using show routines, syscallmonitor( ), and hooks for
obtaining information about, and debugging, system calls.
114
2 Kernel
2.9 Custom System Calls
If show routines are included in your VxWorks configuration (with the component
INCLUDE_SHOW_ROUTINES), the set of system calls currently available can be
displayed with the syscallShow( ) shell command with the shell’s C interpreter:
-> syscallShow
Group Name
GroupNo
-------------------------TEMPGroup
7
STANDARDGroup
8
VXWORKSGroup
9
value = 55 = 0x37 = '7'
NumRtns
------6
48
31
Rtn Tbl Addr
-----------0x001dea50
0x001deab0
0x001dedb0
-> syscallShow 9,1
System Call Group name: VXWORKSGroup
Group Number
: 9
Routines provided
:
Rtn#
Name
------------------------0
(null)
1
(null)
2
(null)
3
msgQSend
4
msgQReceive
5
_msgQOpen
6
objDelete
7
objInfoGet
8
_semTake
9
_semGive
10
_semOpen
11
semCtl
12
_taskOpen
13
taskCtl
14
taskDelay
15
rtpSpawn
16
rtpInfoGet
17
taskKill
18
taskSigqueue
19
_timer_open
20
timerCtl
21
pxOpen
22
pxClose
23
pxUnlink
24
pxCtl
25
pxMqReceive
26
pxMqSend
27
pxSemWait
28
pxSemPost
29
pipeDevCreate
30
pipeDevDelete
value = 50 = 0x32 = '2'
->
Address
---------0x00000000
0x00000000
0x00000000
0x001d9464
0x001d94ec
0x001d9540
0x001d95b8
0x001d9bf8
0x001d9684
0x001d96d0
0x001d970c
0x001d9768
0x001d98b8
0x001d99dc
0x001d99d4
0x001a2e14
0x001a2e60
0x001a2ec8
0x001a2f00
0x0018a860
0x0018a8c0
0x0018a960
0x0018acf4
0x0018ae44
0x0018b334
0x0018aea0
0x0018afcc
0x0018b1fc
0x0018b0f8
0x001971a8
0x001971c4
# Arguments
----------0
0
0
5
4
5
2
4
2
1
5
4
1
4
1
6
2
2
3
4
4
4
1
2
4
6
6
3
1
3
2
115
2
VxWorks
Kernel Programmer's Guide, 6.6
The syscallMonitor( ) routine allows truss style monitoring of system calls from
kernel mode, on a global, or per-process basis. It lists (on the console) every system
call made, and their arguments. The routine synopsis is:
syscallMonitor(level, RTP_ID)
If the level argument is set to 1, the system call monitor is turned on; if it is set to 0,
it is turned off. If the RTP_ID is set to an RTP_ID, it will monitor only the system
calls made from that process; if it is set to 0, it will monitor all system calls.
The sysCallHookLib library provides routines for adding extensions to the
VxWorks system call library with hook routines. Hook routines can be added
without modifying kernel code. The kernel provides call-outs whenever system
call groups are registered, and on entry and exit from system calls. Each hook type
is represented as an array of function pointers. For each hook type, hook functions
are called in the order they were added. For more information, see the
syscallHookLib API reference.
2.9.6 Documenting Custom System Calls
Since system calls are not functions written in C, the apigen documentation
generation utility cannot be used to generate API references from source code
comments. You can, however, create a function header in a C file that can be read
by apigen. The function header for system calls is no different from that for other
C functions.
116
2 Kernel
2.9 Custom System Calls
Here is a function header for getpid( ):
/***********************************************************************
*
* getpid - Get the process identifier for the calling process.
*
* SYNOPSIS
* \cs
* int getpid
*
(
*
void
*
)
* \ce
*
* DESCRIPTION
*
* This routine gets the process identifier for the calling process.
* The ID is guaranteed to be unique and is useful for constructing
* uniquely named entities such as temporary files etc.
*
* RETURNS: Process identifier for the calling process.
*
* ERRNO: N/A.
*
* SEE ALSO:
* .pG "Multitasking"
*
*/
No code or C declaration should follow the header. The compiler treats it as a
comment block, but apigen uses it to generate API documentation. All fields in the
header above (SYNOPSIS, DESCRIPTION, RETURNS, and so on) must to be present
in the code.
You have two choices for the location of the comments:
■
You may add system call function headers to an existing C source file (one that
has code for other functions). Be sure that this source file is part of the
DOC_FILES list in the makefile for that directory. The apigen utility will not
process it otherwise.
■
You may create a C file that contains only function headers and no C code.
Such files must be part of the DOC_FILES list in the makefile, but not part of
the OBJS list (because there is no code to compile).
For more information about the coding conventions that are required for API
documentation generation, and the apigen tool, see the VxWorks BSP Developer’s
Guide and the apigen entry in the Wind River Host Utilities API reference.
117
2
VxWorks
Kernel Programmer's Guide, 6.6
2.10 Custom Scheduler
VxWorks provides a scheduler framework that can be used to implement a custom
scheduler. This section describes the requirements for creating a custom scheduler
for VxWorks, and for configuring the operating system for its use. It also describes
key features of the traditional VxWorks scheduler that provide a useful reference
for designing a custom scheduler.
Before you consider implementing a custom scheduler, however, you should be
familiar with the schedulers provided by Wind River. They are the following:
!
■
The traditional VxWorks scheduler, which provides for priority-based
preemptive scheduling, plus a round-robin extension. The traditional
scheduler is included in VxWorks by default. For information, see
4.3.3 VxWorks Traditional Scheduler, p.168.
■
The VxWorks POSIX thread scheduler, which provides additional features for
pthreads running in processes. For information, see 5.12 POSIX and VxWorks
Scheduling, p.277.
WARNING: The scheduler is fundamental to an operating system's behavior.
VxWorks is not guaranteed to function as expected if a custom scheduler is used.
You should ensure that all VxWorks components behave appropriately when a
custom scheduler is used in place of a Wind River scheduler.
NOTE: The scheduler framework is not supported for the symmetric multiprocess-
ing (SMP) configuration of VxWorks. For general information about VxWorks
SMP and about migration, see 15. VxWorks SMP and 15.15 Migrating Code to
VxWorks SMP, p.702.
2.10.1 Requirements for a Custom Scheduler
The VxWorks scheduler framework allows you to implement a custom kernel
scheduler for VxWorks. This section describes the code and configuration
requirements for custom schedulers.
Code Requirements
A custom scheduler must manage the set of tasks that are in the READY state; that
is, the tasks that are eligible for execution. At a minimum, a custom scheduler must
118
2 Kernel
2.10 Custom Scheduler
define a ready queue structure for all ready tasks or a hook-routine that is executed
at every clock tick.
A custom scheduler may also specify other class-specific structures, manage data
in the task control block, and so on.
Q_HEAD and Q_NODE Structures
A user-specified ready queue class must define class-specific Q_NODE and
Q_HEAD structures. The size of these structures together must not be more than 16
bytes. For more information, see Multi-way Queue Structure, p.124.
Q_CLASS Structure and Associated Routines
Users must define a ready-queue class for all READY tasks. A set of routines
required by the Q_CLASS structure must be implemented. For more information
about the Q_CLASS structure, see Multi-way Queue Structure, p.124.
Task Control Block Data
For a custom scheduler that must store user specific information in tasks, the
pSchedInfo member of the task control block (TCB) may be used. The pSchedInfo
member of the TCB is a void pointer.
There are two ways to access pSchedInfo, as follows:
■
If the qNode is given, the TASK_QNODE_TO_PSCHEDINFO( ) macro may be
used to get the address of pSchedInfo. The file
installDir/vxworks-6.x/target/h/taskLib.h provides the definition of this
macro. The macro is typically used in the user-defined queue management
functions. For example:
void customQPut
(
CUSTOM_Q_HEAD
*pQHead, /* head of readyQ */
CUSTOM_NODE
*pQNode, /* mode of insert */
ULONG
key
/* key for insert */
)
{
void
**ppSchedInfo;
/* get the address to the pSchedInFo */
ppSchedInfo = (void **) TASK_QNODE_TO_PSCHEDINFO (pQNode);
}
■
If the task ID tid is given, the TASK_SCHED_INFO_SET( ) macro can be used to
set the pSchedInfo field in the TCB. The macro TASK_SCHED_INFO_GET( )
119
2
VxWorks
Kernel Programmer's Guide, 6.6
can be used for getting the value of pSchedInfo. Both macros are defined in
installDir/vxworks-6.x/target/h/taskUtilLib.h.
The custom scheduler may use pSchedInfo as the pointer to the user-specific data
structure for tasks. If so, it must allocate memory for the data structure using a task
hook routine that calls malloc( ) or memalign( ). This approach, however, makes
the task creation process less deterministic.
The memory can also be statically allocated (using global variables) for
user-specified storage, and then used it during task initialization.
Tick Hook Routine
If a custom scheduler performs operations at each tick interrupt, the
tickAnnounceHookAdd( ) routine can be used to register a hook routine that is
called at each tick interrupt. The hook routine must obey the same rules as ISRs,
because it runs in interrupt context. Any VxWorks kernel service that should not
be called in an interrupt context should not be called in this hook. For information
about restrictions on ISRs, see 4.20 Interrupt Service Routines, p.241.
The following pseudo-code example illustrates hook use:
void usrTickHook
(
int tid
/* task ID */
)
{
update the statistics information if needed;
update interrupted task's time slice if needed;
resort the interrupted task location in the ready queue if needed.
}
Custom Round-Robin Scheduling
VxWorks provides a round-robin scheduling policy based on a task hook routine.
A custom scheduler may use the VxWorks round-robin scheme by incorporating
the kernelRoundRobinHook( ) routine in the user-specific tick hook routine.
The kernelRoundRobinHook( ) routine places a task at the tail of the task list for
its priority in the ready queue, and resets its time slice, if all of the following are
true:
1.
The interrupted task has not locked preemption.
2.
The interrupted task is still in the READY state.
3.
The interrupted task has consumed its allowed time slice.
120
2 Kernel
2.10 Custom Scheduler
To take advantage of the VxWorks's implementation of round-robin scheduling,
the kernelRoundRobinInstall( ) routine should be called in the
usrCustomSchedulerInit( ) routine to install the kernelRoundRobinHook( ) call.
For more information see Modify usrCustomSchedulerInit( ) Routine, p.122.
The routine _func_kernelRoundRobinHook( ) can then be called within the user
defined hook for the round robin policy to take effect. The
_func_kernelRoundRobinHook( ) takes the task ID (tid) of the interrupted task as
its argument. The following code example takes advantage of the VxWorks
round-robin scheduling scheme:
void usrTickHook
int tid
/* task interrupted by tick */
)
{
/* statistic information */
/* call kernelRoundRobinHook() */
if (_func_kernelRoundRobinHook != NULL)
_func_kernelRoundRobinHook (tid);
/* other work */
...
}
If the custom scheduler does not take advantage of the VxWorks round-robin
scheduling policy using kernelRoundRobinHook( ), the routine
kernelTimeSlice( ) must not be used to adjust system time slice nor to enable or
disable round-robin scheduling. The kernelTimeSlice( ) routine is used to
dynamically enable round robin scheduling and to set the system time slice, or to
disable it.
For more information about VxWorks round-robin scheduling, see Round-Robin
Scheduling, p.169.
Configuration Requirements
To use a custom scheduler with VxWorks, the following must be done:
■
Configure VxWorks for use with a custom scheduler
■
Modify an initialization routine in a configuration file.
■
Link the custom ready queue management code with VxWorks.
121
2
VxWorks
Kernel Programmer's Guide, 6.6
Add INCLUDE_CUSTOM_SCHEDULER Component
To enable the custom scheduler framework, VxWorks must be configured with the
INCLUDE_CUSTOM_SCHEDULER component.
Modify usrCustomSchedulerInit( ) Routine
The usrCustomSchedulerInit( ) routine must be modified to specify the custom
ready queue structure and any hook routines that are executed at each tick
interrupt. The following code illustrates this modification:
void usrCustomSchedulerInit (void)
{
vxKernelSchedDesc.readyQClassId
vxKernelSchedDesc.readyQInitArg1
vxKernelSchedDesc.readyQInitArg2
= qUserClassId; /* user’s class ID */
= (void *) &usrReadyQBMap; /* arg1 */
= (void *) 256; /* arg2 */
tickAnnounceHookAdd ((FUNCPTR)usrTickHook);
kernelRoundRobinInstall();
}
The usrTickHook argument is the hook to be called at each tick interrupt. The
kernelRoundRobinInstall( ) call is for the user VxWorks round-robin scheduling
scheme.
The usrCustomSchedulerInit( ) routine is in the
installDir/vxworks-6.x/target/config/comps/src/usrCustomerScheduler.c file.
For information about the vxKernelSchedDesc variable, see Scheduler
Initialization, p.123. This variable must be initialized for a custom scheduler.
Link Custom Scheduler Code
There are several ways for users to link the definition and implementation of
Q_NODE, Q_HEAD, and Q_CLASS structure to VxWorks. For example, the custom
scheduler configuration file
installDir/vxworks-6.x/target/config/comps/src/usrCustomerScheduler.c can be
the placeholder for the Q_NODE and Q_HEAD type definitions and user specified
Q_CLASS implementation.
Another way is to create a new header file for Q_NODE and Q_HEAD definitions
and a new source file for Q_CLASS implementation, and then link the new object
file to VxWorks using the EXTRA_MODULES macro. For example:
EXTRA_MODULES = qUserPriLib.o
For information about using the EXTRA_MODULES macro, see 2.6.8 Linking Kernel
Application Object Modules with VxWorks, p.64.
122
2 Kernel
2.10 Custom Scheduler
2.10.2 Traditional VxWorks Scheduler
The traditional VxWorks scheduler is the default, and is included in the system
with the INCLUDE_VX_TRADITIONAL_SCHEDULER component. The traditional
scheduler has a priority-based preemptive scheduling policy. A round-robin
scheduling extension can be enabled with the kernelTimeSlice( ) routine. For
more information about these options, see 4.3.3 VxWorks Traditional Scheduler,
p.168.
This section provides information about key features of the traditional VxWorks
scheduler that can be useful in designing a custom scheduler.
Scheduler Initialization
The kernel scheduler description structure is initialized in the usrKernelInit( )
routine. The following is an example for configuring the VxWorks traditional
scheduler:
#ifdef INCLUDE_VX_TRADITIONAL_SCHEDULER
/* install the traditional priority based preemptive scheduler */
#if (VX_TRADITIONAL_SCHED_CONSTANT_RDY_Q == TRUE)
vxKernelSchedDesc.readyQClassId
= Q_PRI_BMAP;
vxKernelSchedDesc.readyQInitArg1
= (void *) &readyQBMap;
vxKernelSchedDesc.readyQInitArg2
= (void *) 256;
#else
vxKernelSchedDesc.readyQClassId
= Q_PRI_LIST;
#endif /* VX_TRADITIONAL_SCHED_CONSTANT_RDY_Q == TRUE */
#endif /* INCLUDE_VX_TRADITIONAL_SCHEDULER */
This code is from installDir/vxworks-6.x/target/config/comps/src/usrKernel.c.
The vxKernelSchedDesc variable is the kernel scheduler description structure,
which is defined in installDir/vxworks-6.x/target/h/kernelLib.h as follows:
typedef struct wind_sched_desc
{
Q_CLASS_ID readyQClassId; /* readyQ Id
*/
void *
readyQInitArg1; /* readyQ init arg 1 */
void *
readyQInitArg2; /* readyQ init arg 2 */
} WIND_SCHED_DESC;
The readyQClassId is a pointer to a ready queue class. The ready queue class is a
structure with a set of pointers to routines that manage tasks that are in the READY
state. The readyQInitArg1 and readyQInitArg2 are the input arguments for the
initRtn( ) routine of the ready queue class.
123
2
VxWorks
Kernel Programmer's Guide, 6.6
The Q_PRI_BMAP value (which is assigned to
vxKernelSchedDesc.readyQClassId) is the priority-based bit-mapped ready
queue class ID. See Multi-way Queue Structure, p.124 for more information about
the ready queue class and its associated members.
The vxKernelSchedDesc can be initialized with the user-specified ready queue
class for customized kernel schedulers. See 2.10.1 Requirements for a Custom
Scheduler, p.118 for more information how to install custom schedulers.
After the initialization of vxKernelSchedDesc variable, VxWorks invokes the
qInit( ) routine to initialize the ready queue class, as follows:
/* kernel scheduler ready queue init */
qInit (&readyQHead, vxKernelSchedDesc.readyQClassId,
(int)(vxKernelSchedDesc.readyQInitArg1),
(int)(vxKernelSchedDesc.readyQInitArg2));
The qInit( ) routine invokes vxKernelSchedDesc.readyQClassId->initRtn( ) to
set up the ready queue and the readyQHead variable (which is of type Q_HEAD).
It is described below in Multi-way Queue Structure, p.124.
Multi-way Queue Structure
The VxWorks scheduler data structure consists of Q_HEAD, Q_NODE, and
Q_CLASS elements. The type definitions of Q_HEAD and Q_NODE structures are
flexible so that they can be used for different types of ready queues.
The readyQHead variable is the head of a so-called multi-way queue, and the
aforementioned Q_PRI_BMAP queue classes comply with the multi-way queue
data structures.
The multi-way queue head structure (Q_HEAD) is defined in qLib.h as follows:
typedef struct
/* Q_HEAD */
{
Q_NODE *pFirstNode;
/* first node in queue based on key */
UINT
qPriv1;
/* use is queue type dependent */
UINT
qPriv2;
/* use is queue type dependent */
Q_CLASS *pQClass;
/* pointer to queue class */
} Q_HEAD;
The first field in the Q_HEAD contains the highest priority node.
124
2 Kernel
2.10 Custom Scheduler
NOTE: Both the qFirst( ) routine and Q_FIRST( ) macro simply read the first four
bytes of the Q_HEAD structure (the pFirstNode field) to determine the head of the
queue. There is, therefore, no need for a queue-class specific routine to determine
which node is the head of the queue.
The kernel scheduler performs a Q_FIRST( ) on readyQHead to determine which
task should be allocated to the CPU. For the Q_PRI_BMAP and Q_PRI_LIST queue
classes, this represents the highest priority ready task.
The multi-way queue node structure (Q_NODE) is also defined in qLib.h as
follows:
typedef struct
{
UINT
qPriv1;
UINT
qPriv2;
UINT
qPriv3;
UINT
qPriv4;
} Q_NODE;
/* Q_NODE */
/*
/*
/*
/*
use
use
use
use
is
is
is
is
queue
queue
queue
queue
type
type
type
type
dependent
dependent
dependent
dependent
*/
*/
*/
*/
Each task control block contains a Q_NODE structure for use by a multi-way queue
class to manage the set of ready tasks. This same Q_NODE is used to manage a task
when it is in a pend queue.
Note that a custom implementation of a multi-way queue class may define
class-specific Q_HEAD and Q_NODE structures. The size of the class-specific
structures must not exceed 16 bytes, which is the current size of both the Q_HEAD
and Q_NODE structures.
Q_CLASS Structure
The kernel interacts with a multi-way queue class through a Q_CLASS structure. A
Q_CLASS structure contains function pointers to the class-specific operators. For
example, the address of the class specific put routine is stored in the putRtn field.
As described in Scheduler Initialization, p.123, the qInit( ) routine is used to
initialize a multi-way queue head to a specified queue type. The second parameter
specifies the queue class (that is, the type of queue), and is merely a pointer to a
Q_CLASS structure. All kernel invocations of the queue class operators are
performed indirectly through the Q_CLASS structure.
The Q_CLASS structure is defined in qClass.h as follows:
125
2
VxWorks
Kernel Programmer's Guide, 6.6
typedef struct q_class
/* Q_CLASS */
{
FUNCPTR createRtn;
/* create and initialize a queue */
FUNCPTR initRtn;
/* initialize a queue */
FUNCPTR deleteRtn;
/* delete and terminate a queue */
FUNCPTR terminateRtn;
/* terminate a queue */
FUNCPTR putRtn;
/* insert a node into q with insertion key */
FUNCPTR getRtn;
/* return and remove lead node routine */
FUNCPTR removeRtn;
/* remove routine */
FUNCPTR resortRtn;
/* resort node to new priority */
FUNCPTR advanceRtn;
/* advance queue by one tick routine */
FUNCPTR getExpiredRtn; /* return and remove an expired Q_NODE */
FUNCPTR keyRtn;
/* return insertion key of node */
FUNCPTR calibrateRtn;
/* calibrate every node in queue by an offset */
FUNCPTR infoRtn;
/* return array of nodes in queue */
FUNCPTR eachRtn;
/* call a user routine for each node in queue */
FUNCPTR restoreRtn;
/* restore a node position in queue */
struct q_class *valid; /* valid == pointer to queue class */
} Q_CLASS;
The restoreRtn operator is used in VxWorks SMP, which does not support custom
schedulers. The operator must therefore be set to NULL.
The following operators are not applicable to a queue class that is used to manage
the set of ready tasks: advanceRtn, getExpiredRtn, and calibrateRtn.
The signatures of the expected Q_CLASS operators are as follows:
Q_HEAD * createRtn
STATUS
initRtn
STATUS
deleteRtn
STATUS
terminateRtn
void
putRtn
Q_NODE * getRtn
STATUS
removeRtn
void
resortRtn
ULONG
keyRtn
int
infoRtn
Q_NODE * eachRtn
(... /*
(Q_HEAD
(Q_HEAD
(Q_HEAD
(Q_HEAD
(Q_HEAD
(Q_HEAD
(Q_HEAD
(Q_HEAD
(Q_HEAD
(Q_HEAD
optional arguments */);
*pQHead, ... /* optional arguments */);
*pQHead);
*pQHead);
*pQHead, Q_NODE *pQNode, ULONG key);
*pQHead);
*pQHead, Q_NODE *pQNode);
*pQHead, Q_NODE *pQNode, ULONG newKey);
*pQHead, Q_NODE *pQNode, int keyType);
*pQHead, Q_NODE *nodeArray [ ], int maxNodes);
*pQHead, FUNCPTR routine, int routineArg);
As noted above, a custom scheduler may define class-specific Q_HEAD and
Q_NODE structures.
Q_CLASS Operators
This section provides descriptions of each Q_CLASS operator that pertains to the
management of ready tasks. Each description provides information about when
the kernel invokes the operator for managing ready tasks.
Descriptions of the advanceRtn, getExpiredRtn, and calibrateRtn operators are
not provided as they are not applicable to managing the set of ready tasks.
126
2 Kernel
2.10 Custom Scheduler
Some Q_CLASS operators are invoked within the kernel context. The operator
description indicate whether the operator is invoked within kernel context or not.
The operators that are invoked within kernel context do not have access to all
VxWorks facilities. Table 2-8 lists the routines that are available from within kernel
context.
Table 2-8
Kernel Context Routines
VxWorks Library
Available Routines
blib
All routines
fppArchLib
fppSave( ) and fppRestore( )
intLib
intContext( ), intCount( ), intVecSet( ), intVecGet( )
lstLib, dllLib, sllLib All routines except xxxCreate( ) and xxxDelete( )
!
mathALib
All routines, if fppSave( ) and fppRestore( ) are used
rngLib
All routines except rngCreate( )
taskLib
taskIdVerify( ), taskIdDefault( ),
taskIsReady( ),taskIsSuspended( )and taskTcb( )
vxLib
vxTas( )
WARNING: The use of any VxWorks APIs that are not listed Table 2-8 from an
operator that is invoked from kernel context results in unpredictable behavior.
Typically the target will hang or reboot.
createRtn
Allocates a multi-way queue head structure from the system memory pool.
The dynamically-allocated head structure is subsequently initialized.
Currently the kernel does not utilize this operator. Instead, the ready task
queue is initialized by statically allocating the head structure, and using the
initRtn operator.
initRtn
Initializes a multi-way queue head. Up to ten optional arguments can be
passed to the initRtn. The kernel initializes the ready task queue from the
usrKernelInit( ) routine as described in Scheduler Initialization, p.123. This
operator is not called from within kernel context.
127
2
VxWorks
Kernel Programmer's Guide, 6.6
deleteRtn
Deallocates (frees) the multi-way queue head. All queued nodes are lost.
Currently the kernel does not utilize this operator.
terminateRtn
Terminates a multi-way queue head. All queued nodes will be lost. Currently
the kernel does not utilize this operator.
putRtn
Inserts a node into a multi-way queue. The insertion is based on the key and
the underlying queue class. The second parameter is the Q_NODE structure
pointer of the task to be inserted into the queue. Recall that each task control
block contains a Q_NODE structure for use by a multi-way queue class to
manage the set of ready tasks.
The third parameter, (the key), is the task’s current priority. Note that a task’s
current priority may be different than a task’s normal priority due to the mutex
semaphore priority inheritance protocol.
The pFirstNode field of the Q_HEAD structure must be updated to contain the
first node in the queue (if any change has occurred).
The putRtn operator is called whenever a task becomes ready; that is, a task is
no longer suspended, pended, delayed, or stopped (or a combination thereof).
The VxWorks round-robin policy performs a removeRtn operation followed
by a putRtn when a task has exceeded its time slice. In this case, the task does
not change state. However, the expectation after performing a removeRtn
operation followed by a putRtn operation is that the task appears as the last
task in the list of tasks with the same priority, if there are any.
Performing a taskDelay(0) operation also results in a removeRtn operation
followed by a putRtn. Again, in this case the task does not change state, and
the expectation after performing a removeRtn operation followed by a putRtn
operation is that the task appears as the last task in the list of tasks with the
same priority, if there are any.
This operator is called from within kernel context.
getRtn
Removes and returns the first node in a multi-way queue. Currently the kernel
does not utilize this operator.
removeRtn
Removes the specified node from the specified multi-way queue.
128
2 Kernel
2.10 Custom Scheduler
The removeRtn operator is called whenever a task is no longer ready; that is,
it is no longer eligible for execution, since it has become suspended, pended,
delayed, or stopped (or a combination thereof).
See the discussion of the putRtn operator above for more information about
situations in which the kernel performs a removeRtn operation followed by a
putRtn without the task’s state actually changing.
This operator is called from within kernel context.
resortRtn
Resorts a node to a new position based on a new key.
The resortRtn operator is called whenever a task’s priority changes, either due
to an explicit priority change with the taskPrioritySet( ) API, or an implicit
priority change due to the mutex semaphore priority inheritance protocol.
The difference between invoking the resortRtn operator and a
removeRtn/putRtn combination is that the former operator does not change
the position of the task in the list of tasks with the same priority (if any) when
the priority is the same as the old priority.
This operator is called from within kernel context.
keyRtn
Returns the key of a node currently in a multi-way queue. The keyType
parameter determines key style on certain queue classes. Currently the kernel
does not utilize this operator.
infoRtn
Gathers information about a multi-way queue. The information consists of an
array, supplied by the caller, filled with all the node pointers currently in the
queue. Currently the kernel does not utilize this operator.
eachRtn
Calls a user-supplied routine once for each node in the multi-way queue. The
routine should be declared as follows:
BOOL routine
(
Q_NODE *pQNode,
int
arg
);
/* pointer to a queue node */
/* arbitrary user-supplied argument */
The user-supplied routine should return TRUE if qEach( ) is to continue calling
it for each entry, or FALSE if it is done, and qEach( ) can exit.
Currently the kernel does not utilize this operator.
129
2
VxWorks
Kernel Programmer's Guide, 6.6
130
3
Boot Loader
3.1 Introduction 132
3.2 Using a Default Boot Loader 132
3.3 Boot Loader Image Types 133
3.4 Boot Loader Shell 135
3.5 Boot Loader Parameters 140
3.6 Rebooting VxWorks 145
3.7 Customizing and Building Boot Loaders 146
3.8 Installing Boot Loaders 152
3.9 Booting From a Network 152
3.10 Booting From a Target File System 154
3.11 Booting From the Host File System Using TSFS 155
131
VxWorks
Kernel Programmer's Guide, 6.6
3.1 Introduction
A VxWorks boot loader is an application whose purpose is to load a VxWorks
image onto a target. It is sometimes called the VxWorks bootrom, but use of this
term is not encouraged (it conflates application and media). Like VxWorks, the
boot loader can be configured with various facilities; such a command shell for
dynamically setting boot parameters, a network loader, and a file system loader.
The same boot loaders are used for uniprocessor (UP), symmetric multiprocessor
(SMP), and asymmetric multiprocessor (AMP), configurations of VxWorks.
In a development environment, boot loaders are useful for loading a VxWorks
image from a host system, where VxWorks can be quickly modified and rebuilt.
They can also be used in production systems when the boot loader and operating
system are stored on a disk or other media.
Self-booting (standalone) VxWorks images do not require a boot loader. These
images are commonly used in production systems (stored in non-volatile devices).
For more information, see 2.4.1 VxWorks Image Types, p.15.
Usually, the boot loader is programmed in a non-volatile device (usually flash
memory or EEPROM) at an address such that it is the first code that is run by the
processor when the target is powered on or rebooted. The procedure to get the
boot loader programmed in a non-volatile device or written to a disk is dependent
on the target, and is described in the BSP reference documentation.
The VxWorks product installation includes default boot loader images for each
installed BSP. If they do not meet your needs, you can create custom boot loaders.
For example, you may need to use a different network driver to load the VxWorks
image over your development network, or you may want to remove the boot
loader shell for deployed systems.
For information beyond what is in this chapter, particularly information about
setting up a cross-development environment, see the Wind River Workbench User’s
Guide: Setting up Your Hardware.
3.2 Using a Default Boot Loader
The default boot loader is designed for a networked target and must be configured
with parameters such as your host and target network addresses, the full path and
132
3 Boot Loader
3.3 Boot Loader Image Types
name of the file to be loaded, the user name, and more. To use the default boot
loader, you must interactively change the default parameters using the boot loader
shell so that the loader can find the VxWorks image on the host and load it onto
the target.
After you have entered boot loader parameters, the target can be booted with the
VxWorks image. For most targets, the new settings are saved (in a non-volatile
device or to disk) so you can reboot the target without resetting them.
You interact with the boot loader shell at a terminal console that is usually
established by connecting a serial port of the target to a serial port on the host and
starting a terminal application on the host. For information about the setup
required to establish communication over the serial port of a particular target, see
the reference documentation for the BSP in question.
When you apply power to the target (or each time it is reset), the target runs the
boot loader (from ROM, flash, disk, or other media). During the boot process, the
target uses its serial port to communicate with the host system. The boot loader
displays a banner page and then starts a seven-second countdown before booting
VxWorks. You can interrupt the boot process by pressing any key in order to set
the appropriate boot parameters.
Default boot loader images are in installDir/vxworks-6.x/target/config/bspName.
The boot loader commands and parameters are described in 3.4.1 Boot Loader Shell
Commands, p.136 and 3.5 Boot Loader Parameters, p.140. The different types of boot
loader images provided as defaults are described in 3.3 Boot Loader Image Types,
p.133.
3.3 Boot Loader Image Types
Boot loader images can be stored in ROM, flash, on disk, or in storage on a
network. Boot loader images are in ELF format. Binary file versions (.bin) are used
for disks, and the hex record file versions (.hex) are used for programming
non-volatile devices. For information about which types are available for a specific
target, see the reference documentation for the BSP in question.
The varieties of boot loader images are described below. The first of each pair of
file names listed for each image type is produced with the PROFILE_BOOTAPP
profile, and the second by the legacy bspDir/config.h method. For more
133
3
VxWorks
Kernel Programmer's Guide, 6.6
information about build methods, see 3.7 Customizing and Building Boot Loaders,
p.146.
Compressed Image
vxWorks_romCompress and bootrom files
The image is almost entirely compressed. It has a small un-compressed portion
that the processor runs immediately after power is applied or the target is
reset. This portion initializes memory and decompresses the compressed
section (stored in non-volatile storage) into RAM, and causes the processor to
switch execution to RAM. Compression of the image makes it much smaller
than other types of boot loader images, therefore it uses less non-volatile
storage. However, decompression increases the boot time.
Uncompressed Image
vxWorks_rom and bootrom_uncmp files
The image is not compressed. It copies itself to RAM and makes the processor
switch execution to RAM. Because the image is not compressed, it is larger
than the compressed image type. However, it has faster boot time because
there is no decompression operation required.
Resident in Non-Volatile Storage
vxWorks_romResident and bootrom_res files
The image copies only the data segment to RAM on startup; the text segment
stays in non-volatile storage. This means that the processor always executes
instructions out of non-volatile storage. It is therefore sometimes described as
being ROM-resident. This type of boot loader image is the one that requires the
least amount of RAM to boot load the VxWorks kernel. It is therefore useful
for boards with very little RAM, which needs to be saved for the application.
Boot loader images are located in installDir/vxworks-6.x/target/config/bspName.
Note that the default images for most BSPs are configured for a networked
development environment. For information about creating a custom boot loader,
see 3.7 Customizing and Building Boot Loaders, p.146.
134
3 Boot Loader
3.4 Boot Loader Shell
3.4 Boot Loader Shell
The boot loader shell provides commands for the following activities:
■
Changing boot loader parameters (such as the host and target IP addresses)
■
Rebooting the target system
■
Managing the boot process.
To include the boot loader shell, configure the boot loader with the
INCLUDE_BOOT_SHELL component.
!
CAUTION: Do not add any of the following components to the boot loader:
■
■
■
INCLUDE_WDB_BANNER
INCLUDE_SIMPLE_BANNER
INCLUDE_SHELL
They conflict with the boot loader shell. If you include any of them, you will
encounter configuration errors.
When the boot loader has initiated booting the system, it prints a banner. After a
countdown elapses, the boot loader loads and runs the specified image. (If the boot
loader shell is not included, the loader executes with its current parameters and
without a countdown.) To reset the boot parameters, interrupt the boot loader
during the countdown by pressing any key before the countdown period elapses.
To access the boot-loader shell prompt, power on (or reboot) the target; then stop
the boot sequence by pressing any key during the seven-second countdown. The
appearance of the boot-loader banner followed by keystroke interruption of the
boot process looks like the following:
VxWorks System Boot
Copyright 1984-2005
Wind River Systems, Inc.
CPU: PC PENTIUM3
Version: VxWorks 6.3
BSP version: 2.0/6
Creation date: Jun 08 2006, 12:08:39
135
3
VxWorks
Kernel Programmer's Guide, 6.6
Press any key to stop auto-boot...
7
[VxWorks Boot]:
To display a list of available commands, type either h or ? at the prompt, followed
by ENTER. For information about the commands, see 3.4.1 Boot Loader Shell
Commands, p.136.
3.4.1 Boot Loader Shell Commands
The VxWorks boot loader provides a set of commands that can be executed from
the boot loader shell, which are described in the following tables. For information
about the boot loader shell, see 3.4 Boot Loader Shell, p.135.
136
3 Boot Loader
3.4 Boot Loader Shell
Table 3-1
Commands Provided with INCLUDE_BOOT_SHELL
Command
Description
h
Help command. Print a list of available boot commands and
flags; the boot device in use, image path, and image file name;
and the boot devices available in the boot loader.
?
Same as h.
@
Boot (that is, load and execute the VxWorks image file) using
the current boot parameters. See 3.5 Boot Loader Parameters,
p.140.
$ [ paramString ]
Boot (that is, load and execute the VxWorks image file). If
used without a parameter string, the command is the same as
@. The parameter string can be used to set boot loader
parameter values all at once, instead of interactively. See
3.5 Boot Loader Parameters, p.140 and 3.5.3 Changing Boot
Loader Parameters Interactively, p.144).
p
Print the current boot parameter values. See 3.5 Boot Loader
Parameters, p.140.
c
Change the boot parameter values. See 3.5 Boot Loader
Parameters, p.140.
l
Load the VxWorks image file using current boot parameters,
but without executing.
g adrs
Go to (execute at) hex address adrs.
e
Display a synopsis of the last occurring VxWorks exception.
v
Display boot-loader banner page with BSP and boot loader
version information.
137
3
VxWorks
Kernel Programmer's Guide, 6.6
Table 3-2
Commands Provided with INCLUDE_BOOT_MEM_CMDS
Command
Description
d adrs[, n]
Display n words of memory starting at hex address adrs. If n
is omitted, the default is 64.
m adrs
Modify memory at location adrs (hex). The system prompts
for modifications to memory, starting at the specified address.
It prints each address, and the current 16-bit value at that
address, in turn.
You can respond in any of the following ways at the prompt:
ENTER
Pressing the ENTER key alone does not change the
address specified with adrs, but continues prompting at
the next address.
number
Entering a number sets the 16-bit contents at the
memory address to that number.
.
Entering a . (period) leaves the address unchanged, and
quits.
f adrs, nbytes, value
Fill nbytes of memory, starting at adrs with value.
t adrs1, adrs2, nbytes Copy nbytes of memory, starting at adrs1, to adrs2.
Table 3-3
Command Provided with INCLUDE_BOOT_ETH_MAC_HANDLER
Command
Description
M [dev] [unitNo] [MAC] Set and display Ethernet MAC address. For example:
M motfcc0 00:A0:1E:00:10:0A
In this case, the device is motfcc, the unit number is zero,
and the MAC address is 00A01E00100A.
138
3 Boot Loader
3.4 Boot Loader Shell
Table 3-4
Command Provided with INCLUDE_BOOT_ETH_ADR_SET
Command
Description
N [last3ByteValuesMAC] Set (last three bytes) and display Ethernet MAC address.
NOTE: The M command (see Table 3-3) is a replacement for the N command
(Table 3-4), which is maintained for backwards compatibility purposes. For
information about which of the two is supported for a given BSP, consult the BSP
reference.
The M and N command are provided by the
INCLUDE_BOOT_ETH_MAC_HANDLER and INCLUDE_BOOT_ETH_ADR_SET,
respectively. Do not use both components in the same configuration of VxWorks.
Table 3-5
Commands Provided with INCLUDE_BOOT_EDR_SUPPORT
Command
Description
P
Print the error log for the error detection and reporting
facility.
C
Clear the error log for the error detection and reporting
facility.
For information about the error detection and reporting facility, see 11. Error
Detection and Reporting.
Table 3-6
Command Provided with BOOT_FILESYSTEMS
Command
Description
devs
Display a list of all devices known to the I/O system. It
performs the same function as the kernel shell C interpreter
command of the same name.
139
3
VxWorks
Kernel Programmer's Guide, 6.6
Table 3-7
Command Provided with INCLUDE_BOOT_USB_FS_LOADER
Command
Description
usbBulkShow nodeId Displays all the logical unit numbers of the USB device
specified by nodeId (for example, a USB memory stick). It
performs the same function as the kernel shell C interpreter
command of the same name.
For information about rebooting a system that is running VxWorks, see
3.6 Rebooting VxWorks, p.145.
3.5 Boot Loader Parameters
The boot loader parameters include all the information needed to locate and load
a VxWorks image, as well as other settings used to manage the boot process. As
required by the specific boot configuration, they can include host and target IP
addresses, the full path and name of the VxWorks image to be booted, the user
name, and so on. Boot parameters can be changed interactively at runtime, or
statically configured when the boot loader is created. Interactive changes to boot
parameters are retained between reboots for most targets (in a non-volatile device
or on disk).
3.5.1 Displaying Current Boot Parameters
If the boot loader is configured with its command shell, the current set of boot
loader parameters can be displayed interactively with the p command, as follows:
[VxWorks Boot]: p
A display similar to the following appears. Note that the p command does not
actually display unassigned parameters, although this example shows them for
completeness.
boot device
unit number
processor number
host name
file name
140
:
:
:
:
:
ln
0
0
mars
c:\tmp\vxWorks
3 Boot Loader
3.5 Boot Loader Parameters
inet on ethernet (e)
: 90.0.0.50:ffffff00
inet on backplane (b) :
host inet (h)
: 90.0.0.1
gateway inet (g)
:
user (u)
: fred
ftp password (pw)(blank=use rsh) :secret
flags (f)
: 0x0
target name (tn)
: phobos
startup script (s)
:
other (o)
:
3
This example corresponds to the configuration illustrated in Figure 3-1.
Figure 3-1
Boot Configuration Example
c:\temp\vxWorks
HOST
TARGET
mars
phobos
user: fred
90.0.0.1
90.0.0.50:ffffff00
Ethernet
90.0.0.x subnet
3.5.2 Description of Boot Loader Parameters
Each of the boot loader parameters is described below, with reference to the
example shown above.
The letters in parentheses after some of the parameters are alternative names used
with the single-string interactive configuration method described in 3.5.3 Changing
Boot Loader Parameters Interactively, p.144, and the static configuration method
described in 3.7.3 Configuring Boot Loader Parameters Statically, p.147.
boot device
The type of device from which to boot. This must be one of the drivers
included in the boot loader (for example, enp for a CMC controller). Due to
limited space in boot media, only a few drivers can be included. A list of the
drivers included in the boot loader image can be displayed in the boot loader
141
VxWorks
Kernel Programmer's Guide, 6.6
shell with the devs or h command. For more information about boot devices,
see 3.7.5 Selecting a Boot Device, p.148.
unit number
The unit number of the boot device, starting at zero.
processor number
A unique numerical target identifier for systems with multiple targets on a
backplane. The backplane master must have its processor number set to zero.
For boards not connected to a backplane, a value of zero is typically used but
is not required.
host name
The name of the host machine to boot from. This is the name by which the host
is known to VxWorks; it need not be the name used by the host itself. (The host
name is mars in the example above.)
file name
The full path name of the VxWorks image to be booted (c:\myProj\vxWorks
in the example). This path name is also reported to the host when you start a
target server, so that it can locate the host-resident image of VxWorks. The
path name is limited to a 160 byte string, including the null terminator.
inet on ethernet (e)
The Internet Protocol (IP) address of a target system Ethernet interface, as well
as the subnet mask used for that interface. The address consists of the IP
address, in dot decimal format, followed by a colon, followed by the mask in
hex format (here, 90.0.0.50:ffffff00).
inet on backplane (b)
The Internet address of a target system with a backplane interface (blank in the
example).
host inet (h)
The Internet address of the host to boot from (90.0.0.1 in the example).
gateway inet (g)
The Internet address of a gateway node for the target if the host is not on the
same network as the target (blank in the example).
user (u)
The user ID used to access the host for the purpose of loading the VxWorks
image file (which is fred in the example). The user must have host permission
to read the VxWorks image file.
142
3 Boot Loader
3.5 Boot Loader Parameters
On a Windows host, the user specified with this parameter must have FTP
access to the host, and the ftp password parameter (below) must be used to
provide the associated password.
On a UNIX (Linux or Solaris) host, the user must have FTP, TFTP, or rsh
access. For rsh, the user must be granted access by adding the user ID to the
host's /etc/host.equiv file, or more typically to the user's .rhosts file
(~userName/.rhosts).
ftp password (pw)
For FTP or TFTP access, this field is used for the password for the user
identified with the user parameter (above). For rsh access it should be left
blank.
NOTE: If this parameter is not used, the boot loader attempts to load the
run-time system image using a protocol based on the UNIX rsh utility, which
is not available for Windows hosts.
flags (f)
Configuration options specified as a numeric value that is the sum of the
values of selected option bits defined below. (This field is zero in the example
because no special boot options were selected.)
= Do not enable the system controller, even if the processor number is 0.
(This option is board specific; refer to your target documentation.)
0x02 = Load all VxWorks symbolsa, instead of just globals.
0x04 = Do not auto-boot.
0x08 = Auto-boot fast (short countdown).
0x20 = Disable login security.
0x80 = Use TFTP to get boot image.
0x400 = Set system to debug mode for the error detection and reporting facility
(depending on whether you are working on kernel modules or user
applications, for more information see 11. Error Detection and Reporting.
0x01
a. Loading a very large group of symbol can cause delays of up to several minutes while
Workbench loads the symbols. For information about how to specify the size of the
symbol batch to load, see the Wind River Workbench User’s Guide.
target name (tn)
The name of the target system to be added to the host table (in the example,
phobos).
startup script (s)
If the kernel shell is included in the downloaded image, this parameter allows
you to pass to it the path and filename of a startup script to execute after the
143
3
VxWorks
Kernel Programmer's Guide, 6.6
system boots. A startup script file can contain only the shell’s C interpreter
commands. (Note that you must not add the INCLUDE_SHELL,
INCLUDE_WDB_BANNER, or INCLUDE_SIMPLE_BANNER components to a
boot loader. These components conflict with the boot loader shell. Doing so
causes project configuration errors.)
This parameter can also be used to specify process-based (RTP) applications to
run automatically at boot time, if VxWorks has been configured with the
appropriate components. See VxWorks Application Programmer’s Guide:
Applications and Processes.
other (o)
This parameter is generally unused and available for applications (blank in the
example). It can be used, for example, for specifying the default network
interface when booting from a file system device. For more information, see
3.7.4 Enabling Networking for Non-Boot Interfaces, p.148.
3.5.3 Changing Boot Loader Parameters Interactively
Boot parameters can be entered interactively from the boot loader prompt, either
individually or as a string.
NOTE: Interactively-defined boot parameters take precedence over
statically-defined parameters for targets that store those changes. Most targets
store interactively-defined boot settings in a non-volatile device (or a text file for
Pentium BSPs), and do not use the statically-defined values for subsequent
reboots.
For information about changing boot parameters statically, see 3.7.3 Configuring
Boot Loader Parameters Statically, p.147.
Changing Parameters Individually
To change parameters on an individual basis, first use the c (change) command at
the boot prompt, and then enter a new value for each parameter as it is displayed.
If a particular field already has the correct value, simply press ENTER. To clear a
field, type a period (.), then press ENTER. To go back to change the previous
parameter, type a dash (-), then press ENTER. If you want to quit before completing
all parameters (but saving your changes), press CTRL+D.
144
3 Boot Loader
3.6 Rebooting VxWorks
Changing Multiple Parameters Simultaneously
To change more than one parameter at time, use the $ boot command at the boot
prompt with a parameter string. The syntax is as follows:
$dev(0,procnum)host:/file h=# e=# b=# g=# u=usr [pw=passwd] f=# tn=targetname s=script o=other
For example:
[VxWorks Boot]:$ln(0,0)mars:c:\myProj\vxWorks e=90.0.0.50 h=90.0.0.1 u=fred pw=…
The order of the parameters with assignments (those with equal signs) is not
important. Omit any assigned fields that are irrelevant. The codes for the assigned
fields correspond to the letter codes shown in parentheses by the p command and
in 3.5.2 Description of Boot Loader Parameters, p.141.
This method can be particularly useful when booting a target from a host script.
The changes made to boot parameters are retained between reboots for most types
of targets (in a non-volatile device or on disk).
3.6 Rebooting VxWorks
When VxWorks is running, any of the following means can be used to reboot it:
■
Enter CTRL+X in the terminal window.
Note that some Windows terminal emulators do not pass CTRL+X to the
target, because of its standard Windows meaning.
■
Invoke reboot( ) from the shell.
■
Press the reset button on the target system.
■
Turn the target’s power off and on.
When you reboot VxWorks in any of these ways, the auto-boot sequence begins
again from the countdown.
145
3
VxWorks
Kernel Programmer's Guide, 6.6
3.7 Customizing and Building Boot Loaders
A boot loader can be configured with just those VxWorks components that are
required for a given system. Boot loaders may need to be customized and rebuilt
for a variety of reasons, including the following:
■
The target is not on a network.
■
The boot loader must use a device or protocol that is not included in the
default boot loader image.
■
An alternate boot method is preferable (such as booting over the Target Server
File System).
■
To statically re-define the default boot loader parameters. When the default
boot loader image is used with a system that does not have a non-volatile
device or disk, the boot loader parameters have to be re-entered manually each
time the target is rebooted. (Note that Pentium boot loaders automatically
write boot parameter changes back to disk.)
■
To create boot loaders for production systems. In addition to setting the boot
loader parameters appropriately, features that are not required (such as the
network stack) can be removed to reduce the size of the boot loader image.
3.7.1 Configuring Boot Loaders
For most BSPs, boot loaders can be configured and built with Wind River
Workbench or the command-line project tool vxprj, using the PROFILE_BOOTAPP
configuration profile.
For some BSPs, the legacy method using bspDir/config.h and bspDir/make must be
used. Note that the legacy method has been deprecated for most purposes, and
cannot be used for multiprocessor development. For information about this
method, see the VxWorks Command-Line Tools User’s Guide.
3.7.2 Boot Loader Components
The INCLUDE_BOOT_APP component provides the basic facility for loading and
executing a VxWorks image.
The PROFILE_BOOTAPP configuration profile can be used with Workbench or
vxprj to create a boot loader (including INCLUDE_BOOT_APP). This profile
includes a basic set of boot loader components, such as those for the boot loader
146
3 Boot Loader
3.7 Customizing and Building Boot Loaders
shell, drivers, file systems, and so on. Among the components that
PROFILE_BOOTAPP provides are those for loading and executing a VxWorks
image, for booting from a network with various protocols, and for booting from
various file systems.
A boot loader needs to be configured appropriately for any device or file system
from which you want to boot. Other components, which can be used to create a
boot loader for specific boot environments and with various boot management
facilities, are described throughout this chapter.
3.7.3 Configuring Boot Loader Parameters Statically
Boot loader parameters include the boot device, IP addresses of the host and target
systems, the location of VxWorks image file, and so on. For detailed information
about the parameters, see 3.5 Boot Loader Parameters, p.140. (For information about
configuring boot loader parameters dynamically, see 3.5.3 Changing Boot Loader
Parameters Interactively, p.144.)
Using Workbench, the DEFAULT_BOOT_LINE configuration parameter of the
INCLUDE_BSP_MACROS component can be used to change the default boot loader
parameters.
Using the legacy config.h method (which should only be used for BSPs that do not
support PROFILE_BOOTAPP), edit the DEFAULT_BOOT_LINE macro in
installDir/vxworks-6.x/target/config/bspName/config.h file to change the default
boot loader parameters. The DEFAULT_BOOT_LINE macro for a Pentium BSP
looks like the following:
#define DEFAULT_BOOT_LINE \
"fd=0,0(0,0)host:/fd0/vxWorks.st h=90.0.0.3 e=90.0.0.50 u=target"
For more information about configuration methods, see 3.7.1 Configuring Boot
Loaders, p.146.
NOTE: Interactively-defined boot parameters take precedence over
statically-defined parameters for targets that store those changes. Most targets
store interactively-defined boot settings in a non-volatile device (or a text file for
Pentium BSPs), and do not use the statically-defined values for subsequent
reboots.
147
3
VxWorks
Kernel Programmer's Guide, 6.6
3.7.4 Enabling Networking for Non-Boot Interfaces
The other (o) boot loader parameter can be used to specify a network interface in
addition to whatever device is specified for the boot device. For example, it can be
used when booting from a local SCSI disk to specify a network interface to be
included. The following example illustrates parameter settings for booting from a
SCSI device, and enabling the network with an on-board Ethernet device (here
with ln for LANCE Ethernet device) with the other field.
boot device
processor number
host name
file name
inet on ethernet (e)
host inet (h)
user (u)
flags (f)
target name (tn)
other
:
:
:
:
:
:
:
:
:
:
scsi=2,0
0
host
/sd0/vxWorks
147.11.1.222:ffffff00
147.11.1.3
jane
0x0
t222
ln
3.7.5 Selecting a Boot Device
The boot devices that are included in a boot loader image can be identified at
run-time with the devs or h command from the boot loader shell (see 3.4.1 Boot
Loader Shell Commands, p.136).
Boot Device Configuration
In order to boot VxWorks, however, the boot loader must be configured with the
appropriate device or devices for your target hardware and desired boot options—
they may not be provided by the default boot loader. The process of configuring
the boot loader with devices is the same as for VxWorks itself, and the topic of
device configuration is covered in 2.4.3 Device Driver Selection, p.22.
Boot Device Specification
Once a boot loader has been configured with the appropriate boot device (or
devices), it must also be instructed as to which device to use. This can be done
interactively or statically.
For information about interactive specification using the boot device parameter,
see 3.5.2 Description of Boot Loader Parameters, p.141 and 3.5.3 Changing Boot Loader
Parameters Interactively, p.144. For information about static configuration, see
3.7.3 Configuring Boot Loader Parameters Statically, p.147.
148
3 Boot Loader
3.7 Customizing and Building Boot Loaders
The boot devices that are supported by a given BSP are described in the BSP
reference. The syntax used for specifying them with the boot device boot loader
parameter is provided below.
3
ATA Device
The syntax for specifying an ATA device is as follows:
ata=controllerNum, driveNum
where controllerNum is the controller number (either 0 or 1) and driveNum is the
drive number (either 0 or 1). For example:
ata=0,0
PCMCIA Device
The syntax for specifying a PCMCIA device is as follows:
pcmcia=socket
where socket is the PCMCIA socket into which the device is plugged.
SCSI Device
The syntax for specifying a SCSI device is as follows:
scsi=id,lun
where id is the SCSI ID of the boot device, and lun is its Logical Unit Number
(LUN).
TSFS Device
The syntax for specifying a Target Server File System device is simply tsfs. No
additional boot device arguments are required. The file path and name must be
relative to the root of the host file system as defined for the target server on the
host. For information about the TSFS, see 8.9 Target Server File System: TSFS, p.518.
3.7.6 Reconfiguring Memory
The VxWorks boot loader can be customized to meet the size constraints of the
non-volatile device on a particular board, as well as the manner in which it
retrieves the VxWorks image file.
149
VxWorks
Kernel Programmer's Guide, 6.6
Persistent Memory Region
The persistent memory region is an area of RAM at the top of system memory
specifically reserved for error records and core dumps. For more information
about use of persistent memory, see 11.2.2 Configuring the Persistent Memory
Region, p.565.
If you increase the size of the persistent memory region for VxWorks beyond the
default, you must also create and install a new boot loader with the same
PM_RESERVED_MEM value.
If you do not, the boot loader (image plus heap) overlays the area of VxWorks
persistent memory that extends beyond its own when the system reboots, and any
data that may have been stored in the overlapping area will be corrupted. For a
simple illustration of this problem, see Figure 3-2.
Note that when you change the value of the PM_RESERVED_MEM for the boot
loader, you need to change the value of RAM_HIGH_ADRS if there is insufficient
room for the boot loader itself between RAM_HIGH_ADRS and sysMemTop( ). If
you do so, also be sure that there is sufficient room for the VxWorks image
between RAM_LOW_ADRS and RAM_HIGH_ADRS.
!
WARNING: Not properly configuring the boot loader (as described above) could
corrupt the persistent memory region when the system boots.
150
3 Boot Loader
3.7 Customizing and Building Boot Loaders
Figure 3-2
RAM Layout Snapshots
3
Boot Loader
Memory Layout
VxWorks
Memory Layout
LOCAL_MEM_LOCAL_ADRS
+ LOCAL_MEM_SIZE
Persistent
Memory
sysMemTop( )
Persistent
Memory
Boot Loader
(image + heap)
sysMemTop( )
RAM_HIGH_ADRS
VxWorks
(image + sys mem)
VxWorks Image
loaded here
RAM_LOW_ADRS
LOCAL_MEM_LOCAL_ADRS
Area of VxWorks persistent memory overlayed by boot loader at reboot
3.7.7 Building Boot Loaders
For boot loaders configured with Workbench or vxprj, and the
PROFILE_BOOTAPP configuration profile, building them is the same as for any
other VxWorks image project.
If you have configured a boot loader with the bspDir/config.h method (which
should only be used for BSPs that do not support PROFILE_BOOTAPP), use the
command make bootLoaderType in the
installDir/vxworks-6.x/target/config/bspName directory. For example:
151
VxWorks
Kernel Programmer's Guide, 6.6
% make bootrom
The different types of boot loader images that you can build are described in
3.3 Boot Loader Image Types, p.133.
For information about configuration methods, see 3.7.1 Configuring Boot Loaders,
p.146.
!
CAUTION: Do not build boot loaders for symmetric multiprocessor (SMP) and
asymmetric multiprocessor (AMP) configurations of VxWorks with the SMP or
AMP build option—neither with Workbench nor with vxprj. The same boot
loaders are used for uniprocessor (UP), SMP, and AMP, configurations of
VxWorks.
3.8 Installing Boot Loaders
For information about installing boot loaders in various media, see the VxWorks
BSP references.
3.9 Booting From a Network
In order to boot from a network, the boot loader must be configured with the
appropriate components for the networking protocol and devices, and boot
parameters must be set accordingly as well.
For information about boot devices, see 3.7.5 Selecting a Boot Device, p.148.
Network Protocol Components
The ability to boot over a network is provided with the
INCLUDE_BOOT_NETWORK component, which registers a network boot loader
with the basic loader facility, checks for the network device, and calls the network
boot loader for a specific protocol. Support for different protocols is as follows:
152
3 Boot Loader
3.9 Booting From a Network
INCLUDE_BOOT_FTP_LOADER
FTP boot loader.
INCLUDE_BOOT_TFTP_LOADER
TFTP boot loader.
3
INCLUDE_BOOT_RSH_LOADER
RSH boot loader.
Parameters for Booting From a Network
The parameters and settings specific to booting from a network with a give
protocol are described below.
For general about boot parameters and how to set them, see 3.5.2 Description of Boot
Loader Parameters, p.141, 3.5.3 Changing Boot Loader Parameters Interactively, p.144,
and 3.7.3 Configuring Boot Loader Parameters Statically, p.147.
FTP
The user and ftp password boot parameters must be set to match account settings
with the FTP server on the host.
TFTP
The flags boot parameter must be set to 0x80, and the user and ftp password
parameters must be set match account settings with the TFTP server on the host.
RSH
The ftp password parameter must be set to empty, that is by entering . (a period).
Updating Ethernet MAC Settings
The INCLUDE_BOOT_ETH_MAC_HANDLER provides the M boot loader shell
command, which can be used to update (and display) the Ethernet MAC address
for the target system. For more information in this regard, see 3.4.1 Boot Loader Shell
Commands, p.136.
153
VxWorks
Kernel Programmer's Guide, 6.6
3.10 Booting From a Target File System
In order to boot from a file system, the boot loader must be configured with the
appropriate components and devices, and boot parameters must be set
accordingly as well.
For information about boot devices, see 3.7.5 Selecting a Boot Device, p.148.
For information about booting from TSFS, see 3.11 Booting From the Host File System
Using TSFS, p.155.
File System Components
The ability to boot from file systems is provided with the
INCLUDE_BOOT_FILESYSTEMS component, in conjunction with a device-specific
file system loader component. The options for file system loaders are as follows:
INCLUDE_BOOT_ATA_LOADER
Support for ATA devices.
INCLUDE_BOOT_FD_LOADER
Support for floppy-disk devices.
Parameters for Booting From a Target File System
The parameters specific to booting from a target file system are described below.
For general about boot parameters and how to set them, see 3.5.2 Description of Boot
Loader Parameters, p.141, 3.5.3 Changing Boot Loader Parameters Interactively, p.144,
and 3.7.3 Configuring Boot Loader Parameters Statically, p.147.
ATA
The boot device boot loader parameter must be set to ata.
FD—Floppy Disk
The boot device boot loader parameter must be set to fd.
154
3 Boot Loader
3.11 Booting From the Host File System Using TSFS
3.11 Booting From the Host File System Using TSFS
The simplest way to boot a target from a host without a network is to use the Target
Server File System. This does not involve configuring SLIP or PPP. The TSFS can
be used to boot a target connected to the host by one or two serial lines. Configure
VxWorks with the INCLUDE_TSFS_BOOT component.
!
WARNING: The TSFS boot facility is not compatible with WDB agent network
configurations. For information about WDB, see 12.6 WDB Target Agent, p.626.
To configure a boot loader for TSFS, the boot device parameter must be tsfs, and
the file path and name must be relative to the root of the host file system defined
for the target server.
Regardless of how you specify the boot line parameters, you must reconfigure (as
described below) and rebuild the boot image.
If two serial lines connect the host and target (one for the target console and one
for WDB communications), the following configuration parameters must be set:
■
■
■
■
■
■
CONSOLE_TTY
CONSOLE_TTY 0
WDB_TTY_CHANNEL
WDB_TTY_CHANNEL 1
WDB_COMM_TYPE
WDB_COMM_TYPE WDB_COMM_SERIAL
If one serial line connects the host and target, the following configuration
parameters must be set:
■
■
■
■
■
■
CONSOLE_TTY
CONSOLE_TTY NONE
WDB_TTY_CHANNEL
WDB_TTY_CHANNEL 0
WDB_COMM_TYPE
WDB_COMM_TYPE WDB_COMM_SERIAL
With any of these TSFS configurations, you can also use the target server console
to set the boot loader parameters by including the
INCLUDE_TSFS_BOOT_VIO_CONSOLE component in VxWorks. This disables the
auto-boot mechanism, which might otherwise boot the target before the target
server could start its virtual I/O mechanism. (The auto-boot mechanism is
similarly disabled when CONSOLE_TTY is set to NONE, or when CONSOLE_TTY is
set to WDB_TTY_CHANNEL.) Using the target server console is particularly useful
155
3
VxWorks
Kernel Programmer's Guide, 6.6
for a single serial connection, as it provides an otherwise unavailable means of
changing boot loader parameters from the command line.
When you build the boot image, select bootrom.hex for the image type (see
3.7.7 Building Boot Loaders, p.151).
For more information about the TSFS, see the 8.9 Target Server File System: TSFS,
p.518.
156
4
Multitasking
4.1 Introduction 159
4.2 Tasks and Multitasking 160
4.3 Task Scheduling 166
4.4 Task Creation and Management 171
4.5 Task Error Status: errno 184
4.6 Task Exception Handling 187
4.7 Shared Code and Reentrancy 187
4.8 Intertask and Interprocess Communication 193
4.9 Public and Private Objects 194
4.10 Shared Data Structures 196
4.11 Mutual Exclusion 196
4.12 Semaphores 198
4.13 Message Queues 213
4.14 Pipes 218
4.15 VxWorks Events 219
4.16 Message Channels 226
4.17 Network Communication 226
4.18 Signals 226
157
VxWorks
Kernel Programmer's Guide, 6.6
4.19 Watchdog Timers 239
4.20 Interrupt Service Routines 241
158
4 Multitasking
4.1 Introduction
4.1 Introduction
Modern real-time systems are based on the complementary concepts of
multitasking and intertask communications. A multitasking environment allows a
real-time application to be constructed as a set of independent tasks, each with its
own thread of execution and set of system resources.
Tasks are the basic unit of scheduling in VxWorks. All tasks, whether in the kernel
or in processes, are subject to the same scheduler. VxWorks processes are not
themselves scheduled.
Intertask communication facilities allow tasks to synchronize and communicate in
order to coordinate their activity. In VxWorks, the intertask communication
facilities include semaphores, message queues, message channels, pipes,
network-transparent sockets, and signals.
For interprocess communication, VxWorks semaphores and message queues,
pipes, and events (as well as POSIX semaphores and events) can be created as
public objects to provide accessibility across memory boundaries (between the
kernel and processes, and between different processes). In addition, message
channels provide a socket-based inter-processor and inter-process
communications mechanism.
Hardware interrupt handling is a key facility in real-time systems because
interrupts are the usual mechanism to inform a system of external events. To get
the fastest possible response to interrupts, interrupt service routines (ISRs) in
VxWorks run in a special context of their own, outside any task’s context.
VxWorks includes a watchdog-timer mechanism that allows any C function to be
connected to a specified time delay. Watchdog timers are maintained as part of the
system clock ISR. For information about POSIX timers, see 5.6 POSIX Clocks and
Timers, p.259.
This chapter discusses the tasking, intertask communication, and interprocess
communication facilities that are at the heart of the VxWorks run-time
environment.
For information about POSIX support for VxWorks, see 5. POSIX Facilities.
NOTE: This chapter provides information about facilities available in the VxWorks
kernel. For information about facilities available to real-time processes, see the
corresponding chapter in the VxWorks Application Programmer’s Guide.
159
4
VxWorks
Kernel Programmer's Guide, 6.6
NOTE: This chapter provides information about multitasking facilities that are
common to both uniprocessor (UP) and symmetric multiprocessor (SMP)
configurations of VxWorks. It also provides information about those facilities that
are specific to the UP configuration. In the latter case, the alternatives available for
SMP systems are noted.
With few exceptions, the symmetric multiprocessor (SMP) and uniprocessor (UP)
configurations of VxWorks share the same API—the difference amounts to only a
few routines. Also note that some programming practices—such as implicit
synchronization techniques relying on task priority instead of explicit locking—
are not appropriate for an SMP system.
For information about SMP programming, see 15. VxWorks SMP. For information
specifically about migration, see 15.15 Migrating Code to VxWorks SMP, p.702.
4.2 Tasks and Multitasking
VxWorks tasks are the basic unit of code execution in the operating system itself,
as well as in applications that it executes as processes. In other operating systems
the term thread is used similarly. (For information about VxWorks support for
POSIX threads, see 5.10 POSIX Threads, p.264).
Multitasking provides the fundamental mechanism for an application to control
and react to multiple, discrete real-world events. The VxWorks real-time kernel
provides the basic multitasking environment. Multitasking creates the appearance
of many threads of execution running concurrently when, in fact, the kernel
interleaves their execution on the basis of a scheduling policy.
Each task has its own context, which is the CPU environment and system resources
that the task sees each time it is scheduled to run by the kernel. On a context switch,
a task’s context is saved in the task control block (TCB).
A task’s context includes:
■
a thread of execution; that is, the task’s program counter
■
the tasks’ virtual memory context (if process support is included)
■
the CPU registers and (optionally) coprocessor registers
160
4 Multitasking
4.2 Tasks and Multitasking
■
stacks for dynamic variables and function calls
■
I/O assignments for standard input, output, and error
■
a delay timer
■
a time-slice timer
■
kernel control structures
■
signal handlers
■
task private environment (for environment variables)
■
error status (errno)
■
debugging and performance monitoring values
4
If VxWorks is configured without process support (the INCLUDE_RTP
component), the context of a task does not include its virtual memory context. All
tasks can only run in a single common address space (the kernel).
However, if VxWorks is configured with process support—regardless of whether
or not any processes are active—the context of a kernel task does include its virtual
memory context, because the system has the potential to operate with other virtual
memory contexts besides the kernel. That is, the system could have tasks running
in several different virtual memory contexts (the kernel and one or more
processes).
For information about virtual memory contexts, see 6. Memory Management.
NOTE: The POSIX standard includes the concept of a thread, which is similar to a
task, but with some additional features. For details, see 5.10 POSIX Threads, p.264.
4.2.1 Task States and Transitions
The kernel maintains the current state of each task in the system. A task changes
from one state to another as a result of kernel function calls made by the
application. When created, tasks enter the suspended state. Activation is necessary
for a created task to enter the ready state. The activation phase is extremely fast,
enabling applications to pre-create tasks and activate them in a timely manner. An
alternative is the spawning primitive, which allows a task to be created and
activated with a single function. Tasks can be deleted from any state.
161
VxWorks
Kernel Programmer's Guide, 6.6
Table 4-1 describes the state symbols that you see when working with development
tools. Example 4-1 shows output from the i( ) command containing task state
information.
162
4 Multitasking
4.2 Tasks and Multitasking
Table 4-1
Task State Symbols
State Symbol
Description
READY
The task is not waiting for any resource other than the CPU.
PEND
The task is blocked due to the unavailability of some resource.
DELAY
The task is asleep for some duration.
SUSPEND
The task is unavailable for execution (but not pended or delayed).
This state is used primarily for debugging. Suspension does not
inhibit state transition, only execution. Thus, pended-suspended
tasks can still unblock and delayed-suspended tasks can still
awaken.
STOP
The task is stopped by the debugger.
DELAY + S
The task is both delayed and suspended.
PEND + S
The task is both pended and suspended.
PEND + T
The a task is pended with a timeout value.
STOP + P
Task is pended and stopped (by the debugger, error detection and
reporting facilities, or SIGSTOP signal).
STOP + S
Task is stopped by (by the debugger, error detection and reporting
facilities, or SIGSTOP signal) and suspended.
STOP + T
Task is delayed and stopped (by the debugger, error detection and
reporting facilities, or SIGSTOP signal).
PEND + S + T
The task is pended with a timeout value and suspended.
STOP +P + S
Task is pended, suspended and stopped by the debugger.
STOP + P + T
Task pended with a timeout and stopped by the debugger.
STOP +T + S
Task is suspended, delayed, and stopped by the debugger.
ST+P+S+T
Task is pended with a timeout, suspended, and stopped by the
debugger.
state + I
The task is specified by state (any state or combination of states
listed above), plus an inherited priority.
163
4
VxWorks
Kernel Programmer's Guide, 6.6
The STOP state is used by the debugger facilities when a breakpoint is hit. It is also
used by the error detection and reporting facilities when an error condition occurs
(see 11. Error Detection and Reporting).
Example 4-1
Task States in Shell Command Output
-> i
NAME
---------tIsr0
tJobTask
tExcTask
tLogTask
tNbioLog
tShell0
tWdbTask
tErfTask
tXbdServic>
tNet0
ipcom_sysl>
ipnetd
ipcom_teln>
miiBusMoni>
value = 0 =
ENTRY
TID
PRI
------------ -------- --42cb40
25b1f74
0
3732d0
25b5140
0
372850
4f033c
0
logTask
25b7754
0
373f28
25bae18
0
shellTask
2fbdcb4
1
wdbTask
2faca28
3
42e0a0
25bd0a4 10
36e4b4
25ac3d0 50
ipcomNetTask 25cdb00 50
3cba50
27fec0c 50
3e2170
2fa6d10 50
ipcom_telnet 2fa979c 50
429420
25a8010 254
0x0
164
STATUS
PC
SP
ERRNO DELAY
---------- -------- -------- ------- ----PEND
3bcf54 25b1f2c
0
0
PEND
3bcf54 25b50e8
0
0
PEND
3bcf54
4ef0f8
0
0
PEND
3bb757 25b7670
0
0
PEND
3bcf54 25bad6c
0
0
READY
3c2bdc 2fbc0d4
0
0
PEND
3bcf54 2fac974
0
0
PEND
3bd3be 25bd03c
0
0
PEND+T
3bd3be 25ac36c 3d0004
6
PEND
3bcf54 25cda88
0
0
PEND
3bd3be 27feab0
0
0
PEND
3bcf54 2fa6c98 3d0004
0
PEND
3bcf54 2fa9594
0
0
DELAY
3c162d 25a7fd0
0
93
4 Multitasking
4.2 Tasks and Multitasking
Figure 4-1 illustrates task state transitions for a deployed system—without the
STOP state associated with development activity. The routines listed are examples
of ones that would cause the associated transition. For example, a task that called
taskDelay( ) would move from the ready state to the delayed state.
Figure 4-1
4
Task State Transitions
The highest-priority ready task is executing.
pended
ready
delayed
suspended
taskInit( )
ready
ready
ready
pended
pended
delayed
delayed
suspended
suspended
suspended
pended
delayed
suspended
ready
suspended
ready
suspended
ready
pended
delayed
semTake( ) / msgQReceive( )
taskDelay( )
taskSuspend( )
semGive( ) / msgQSend( )
taskSuspend( )
expired delay
taskSuspend( )
taskResume( ) / taskActivate( )
taskResume( )
taskResume( )
165
VxWorks
Kernel Programmer's Guide, 6.6
4.3 Task Scheduling
Multitasking requires a task scheduler to allocate the CPU to ready tasks. VxWorks
provides the following scheduler options:
■
The traditional VxWorks scheduler, which provides priority-based,
preemptive scheduling, as well as a round-robin extension. See 4.3.3 VxWorks
Traditional Scheduler, p.168.
■
The VxWorks POSIX threads scheduler, which is designed (and required) for
running pthreads in processes (RTPs). See 5.12 POSIX and VxWorks Scheduling,
p.277.)
■
A custom scheduler framework, which allows you to develop your own
scheduler. See 2.10 Custom Scheduler, p.118.
4.3.1 Task Priorities
Task scheduling relies on a task’s priority. The VxWorks kernel provides 256
priority levels, numbered 0 through 255. Priority 0 is the highest and priority 255
is the lowest.
A task is assigned its priority at creation, but you can also change it
programmatically thereafter. For information about priority assignment, see
4.4.1 Task Creation and Activation, p.172 and 4.3.2 Task Scheduling Control, p.167).
Application Task Priorities
All application tasks should be in the priority range from 100 to 255.
Driver Task Priorities
In contrast to application tasks, which should be in the task priority range from 100
to 255, driver support tasks (which are associated with an ISR) can be in the range
of 51-99.
These tasks are crucial; for example, if a support task fails while copying data from
a chip, the device loses that data. Examples of driver support tasks include tNet0
(the VxWorks network daemon task), an HDLC task, and so on.
The system tNet0 has a priority of 50, so user tasks should not be assigned
priorities below that task; if they are, the network connection could die and
prevent debugging capabilities with the host tools.
166
4 Multitasking
4.3 Task Scheduling
4.3.2 Task Scheduling Control
The routines that control task scheduling are listed in Table 4-2.
Table 4-2
Task Scheduling Control Routines
4
Routine
Description
kernelTimeSlice( )
Controls round-robin scheduling. See Round-Robin
Scheduling, p.169.
taskRotate( )
Rotates tasks on the ready queue.
taskPrioritySet( )
Changes the priority of a task.
taskLock( )
Disables task rescheduling.
taskUnlock( )
Enables task rescheduling.
The taskRotate( )routine can be used as an alternative to round-robin scheduling.
It allows a program to control sharing of the CPU amongst tasks of the same
priority that are ready to run, rather than having the system do so at
predetermined equal intervals. For information about round-robin scheduling, see
Round-Robin Scheduling, p.169.
Task Priority
Tasks are assigned a priority when they are created (see 4.4.1 Task Creation and
Activation, p.172). You can change a task’s priority level while it is executing by
calling taskPrioritySet( ). The ability to change task priorities dynamically allows
applications to track precedence changes in the real world.
Preemption Locks
The scheduler can be explicitly disabled and enabled on a per-task basis in the
kernel with the routines taskLock( ) and taskUnlock( ). When a task disables the
scheduler by calling taskLock( ), no priority-based preemption can take place
while that task is running.
If the task that has disabled the scheduler with taskLock( ) explicitly blocks or
suspends, the scheduler selects the next highest-priority eligible task to execute.
167
VxWorks
Kernel Programmer's Guide, 6.6
When the preemption-locked task unblocks and begins running again, preemption
is again disabled.
NOTE: The taskLock( ) and taskUnlock( ) routines are provided for the UP
configuration of VxWorks, but not the SMP configuration. Several alternative are
available for SMP systems, including task-only spinlocks, which default to
taskLock( ) and taskUnlock( ) behavior in a UP system. For more information, see
15.6.2 Task-Only Spinlocks, p.682 and 15.15 Migrating Code to VxWorks SMP, p.702.
Note that preemption locks prevent task context switching, but do not lock out
interrupt handling.
Preemption locks can be used to achieve mutual exclusion; however, keep the
duration of preemption locking to a minimum. For more information, see
4.11 Mutual Exclusion, p.196.
A Comparison of taskLock( ) and intLock( )
When using taskLock( ), consider that it will not achieve mutual exclusion.
Generally, if interrupted by hardware, the system will eventually return to your
task. However, if you block, you lose task lockout. Thus, before you return from
the routine, taskUnlock( ) should be called.
When a task is accessing a variable or data structure that is also accessed by an ISR,
you can use intLock( ) to achieve mutual exclusion. Using intLock( ) makes the
operation atomic in a single processor environment. It is best if the operation is kept
minimal, meaning a few lines of code and no function calls. If the call is too long,
it can directly impact interrupt latency and cause the system to become far less
deterministic.
For information about interrupts, see 4.11.1 Interrupt Locks and Latency, p.197 and
4.20 Interrupt Service Routines, p.241.
4.3.3 VxWorks Traditional Scheduler
The VxWorks traditional scheduler provides priority-based preemptive
scheduling as well as the option of programmatically initiating round-robin
scheduling. The traditional scheduler may also be referred to as the original or
native scheduler.
The traditional scheduler is included in VxWorks by default with the
INCLUDE_VX_TRADITIONAL_SCHEDULER component.
168
4 Multitasking
4.3 Task Scheduling
For information about the POSIX thread scheduler and custom schedulers, see
5.12 POSIX and VxWorks Scheduling, p.277 and 2.10 Custom Scheduler, p.118,
respectively.
Priority-Based Preemptive Scheduling
4
A priority-based preemptive scheduler preempts the CPU when a task has a higher
priority than the current task running. Thus, the kernel ensures that the CPU is
always allocated to the highest priority task that is ready to run. This means that if
a task—with a higher priority than that of the current task—becomes ready to run,
the kernel immediately saves the current task’s context, and switches to the context
of the higher priority task. For example, in Figure 4-2, task t1 is preempted by
higher-priority task t2, which in turn is preempted by t3. When t3 completes, t2
continues executing. When t2 completes execution, t1 continues executing.
The disadvantage of this scheduling policy is that, when multiple tasks of equal
priority must share the processor, if a single task is never blocked, it can usurp the
processor. Thus, other equal-priority tasks are never given a chance to run.
Round-robin scheduling solves this problem.
Figure 4-2
Priority Preemption
t3
priority
HIGH
LOW
t2
t2
t1
t1
time
KEY:
= preemption
= task completion
Round-Robin Scheduling
VxWorks provides a round-robin extension to priority-based preemptive
scheduling. Round-robin scheduling accommodates instances in which there are
more than one task of a given priority that is ready to run. The round-robin
169
VxWorks
Kernel Programmer's Guide, 6.6
algorithm attempts to share the CPU amongst these tasks by using time-slicing.
Each task in a group of tasks with the same priority executes for a defined interval,
or time slice, before relinquishing the CPU to the next task in the group. No one of
them, therefore, can usurp the processor until it is blocked. See Figure 4-3 for an
illustration of this activity.
Note that while round-robin scheduling is used in some operating systems to
provide equal CPU time to all tasks (or processes), regardless of their priority, this
is not the case with VxWorks. Priority-based preemption is essentially unaffected
by the VxWorks implementation of round-robin scheduling. Any higher-priority
task that is ready to run immediately gets the CPU, regardless of whether or not
the current task is done with its slice of execution time. When the interrupted task
gets to run again, it simply continues using its unfinished execution time.
In most systems, it is not necessary to enable round-robin scheduling, the
exception being when multiple copies of the same code are to be run, such as in a
user interface task.
Note that the taskRotate( )routine can be used as an alternative to round-robin
scheduling. It is useful for situations in which you want to share the CPU amongst
tasks of the same priority that are ready to run, but to do so as a program requires,
rather than at predetermined equal intervals.
Enabling Round-Robin Scheduling
Round-robin scheduling is enabled by calling kernelTimeSlice( ), which takes a
parameter for a time slice, or interval.
Time-slice Counts and Preemption
The time-slice or interval defined with a kernelTimeSlice( ) call is the amount of
time that each task is allowed to run before relinquishing the processor to another
equal-priority task. Thus, the tasks rotate, each executing for an equal interval of
time. No task gets a second slice of time before all other tasks in the priority group
have been allowed to run.
If round-robin scheduling is enabled, and preemption is enabled for the executing
task, the system tick handler increments the task’s time-slice count. When the
specified time-slice interval is completed, the system tick handler clears the
counter and the task is placed at the tail of the list of tasks at its priority level. New
tasks joining a given priority group are placed at the tail of the group with their
run-time counter initialized to zero.
Enabling round-robin scheduling does not affect the performance of task context
switches, nor is additional memory allocated.
170
4 Multitasking
4.4 Task Creation and Management
If a task blocks or is preempted by a higher priority task during its interval, its
time-slice count is saved and then restored when the task becomes eligible for
execution. In the case of preemption, the task will resume execution once the
higher priority task completes, assuming that no other task of a higher priority is
ready to run. In the case where the task blocks, it is placed at the tail of the list of
tasks at its priority level. If preemption is disabled during round-robin scheduling,
the time-slice count of the executing task is not incremented.
Time-slice counts are accrued by the task that is executing when a system tick
occurs, regardless of whether or not the task has executed for the entire tick
interval. Due to preemption by higher priority tasks or ISRs stealing CPU time
from the task, it is possible for a task to effectively execute for either more or less
total CPU time than its allotted time slice.
Figure 4-3 shows round-robin scheduling for three tasks of the same priority: t1,
t2, and t3. Task t2 is preempted by a higher priority task t4 but resumes at the count
where it left off when t4 is finished.
Figure 4-3
Round-Robin Scheduling
t4
HIGH
priority
time slice
LOW
t1
t2
t3
t1
t2
t2
t3
time
KEY:
= preemption
= task completion
4.4 Task Creation and Management
The following sections give an overview of the basic VxWorks task routines, which
are found in the VxWorks library taskLib. These routines provide the means for
171
4
VxWorks
Kernel Programmer's Guide, 6.6
task creation and control, as well as for retrieving information about tasks. See the
VxWorks API reference for taskLib for further information.
For interactive use, you can control VxWorks tasks with the host tools or the kernel
shell; see the Wind River Workbench User’s Guide, the VxWorks Command-Line Tools
User’s Guide, and VxWorks Kernel Programmer’s Guide: Target Tools.
4.4.1 Task Creation and Activation
The routines listed in Table 4-3 are used to create tasks.
The arguments to taskSpawn( ) are the new task’s name (an ASCII string), the
task’s priority, an options word, the stack size, the main routine address, and 10
arguments to be passed to the main routine as startup parameters:
id = taskSpawn ( name, priority, options, stacksize, main, arg1, …arg10 );
Note that a task’s priority can be changed after it has been spawned; see 4.3.2 Task
Scheduling Control, p.167.
The taskSpawn( ) routine creates the new task context, which includes allocating
the stack and setting up the task environment to call the main routine (an ordinary
subroutine) with the specified arguments. The new task begins execution at the
entry to the specified routine.
Table 4-3
Task Creation Routines
Call
Description
taskSpawn( )
Spawns (creates and activates) a new task.
taskCreate( )
Creates, but not activates a new task.
taskInit( )
Initializes a new task.
taskInitExcStk( )
Initializes a task with stacks at specified addresses.
taskOpen( )
Open a task (or optionally create one, if it does not exist).
taskActivate( )
Activates an initialized task.
The taskOpen( ) routine provides a POSIX-like API for creating a task (with
optional activation) or obtaining a handle on existing task. It also provides for
creating a task as either a public or private object (see 4.4.4 Task Names and IDs,
p.177). The taskOpen( ) routine is the most general purpose task-creation routine.
172
4 Multitasking
4.4 Task Creation and Management
The taskSpawn( ) routine embodies the lower-level steps of allocation,
initialization, and activation. The initialization and activation functions are
provided by the routines taskCreate( ) and taskActivate( ); however, Wind River
recommends that you use these routines only when you need greater control over
allocation or activation.
The difference between taskInit( ) and taskInitExcStk( ) is that the taskInit( )
routine allows the specification of the execution stack address, while
taskInitExcStk( ) allows the specification of both the execution and exception
stacks.
4.4.2 Task Creation Options
When a task is spawned, you can pass in one or more option parameters, which are
listed in Table 4-4. The result is determined by performing a logical OR operation
on the specified options.
173
4
VxWorks
Kernel Programmer's Guide, 6.6
Table 4-4
Task Options
Name
Description
VX_ALTIVEC_TASK
Execute with Altivec coprocessor support.
VX_DEALLOC_EXC_STACK
Deallocates the exception stack.
VX_DEALLOC_STACK
Deallocate stack.
VX_DSP_TASK
Execute with DSP coprocessor support.
VX_FP_TASK
Executes with the floating-point coprocessor.
VX_NO_STACK_FILL
Does not fill the stack with 0xEE.
VX_NO_STACK_PROTECT
Create without stack overflow or underflow guard
zones (see 4.4.3 Task Stack, p.175).
VX_PRIVATE_ENV
Executes a task with a private environment.
VX_TASK_NOACTIVATE
Used with taskOpen( ) so that the task is not
activated.
VX_UNBREAKABLE
Disables breakpoints for the task.
Floating Point Operations
You must include the VX_FP_TASK option when creating a task that does any of
the following:
■
Performs floating-point operations.
■
Calls any function that returns a floating-point value.
■
Calls any function that takes a floating-point value as an argument.
For example:
tid = taskSpawn ("tMyTask", 90, VX_FP_TASK, 20000, myFunc, 2387, 0, 0,
0, 0, 0, 0, 0, 0, 0);
Some routines perform floating-point operations internally. The VxWorks
documentation for each of these routines clearly states the need to use the
VX_FP_TASK option.
174
4 Multitasking
4.4 Task Creation and Management
Filling Task Stacks
Note that in addition to using the VX_NO_STACK_FILL task creation option for
individual tasks, you can use the VX_GLOBAL_NO_STACK_FILL configuration
parameter (when you configure VxWorks) to disable stack filling for all tasks and
interrupts in the system.
By default, task and interrupt stacks are filled with 0xEE. Filling stacks is useful
during development for debugging with the checkStack( ) routine. It is generally
not used in deployed systems because not filling stacks provides better
performance during task creation (and at boot time for statically-initialized tasks).
After a task is spawned, you can examine or alter task options by using the
routines listed in Table 4-5. Currently, only the VX_UNBREAKABLE option can be
altered.
Table 4-5
Task Option Routines
Call
Description
taskOptionsGet( )
Examines task options.
taskOptionsSet( )
Sets task options.
4.4.3 Task Stack
The size of each task’s stack is defined when the task is created (see 4.4.1 Task
Creation and Activation, p.172).
It can be difficult, however, to know exactly how much stack space to allocate. To
help avoid stack overflow and corruption, you can initially allocated a stack that is
much larger than you expect the task to require. Then monitor the stack
periodically from the shell with checkStack( ) or ti( ). When you have determined
actual usage, adjust the stack size accordingly for testing and for the deployed
system.
In addition to experimenting with task stack size, you can also configure and test
systems with guard zone protection for task stacks (for more information, see Task
Stack Protection, p.176).
175
4
VxWorks
Kernel Programmer's Guide, 6.6
Task Stack Protection
Task stacks can be protected with guard zones and by making task stacks
non-executable.
Task Stack Guard Zones
Systems can be configured with the INCLUDE_PROTECT_TASK_STACK
component to provide guard zone protection for task stacks. If memory usage
becomes an issue, the component can be removed for final testing and the
deployed system.
An overrun guard zone prevents a task from going beyond the end of its
predefined stack size and corrupting data or other stacks. An under-run guard
zone typically prevents buffer overflows from corrupting memory that precedes
the base of the stack. The CPU generates an exception when a task attempts to
access any of the guard zones. The size of a stack is always rounded up to a
multiple of the MMU page size when either a guard zone is inserted or when the
stack is made non-executable.
Note that guard zones cannot catch instances in which a buffer that causes an
overflow is greater than the page size (although this is rare). For example, if the
guard zone is one page of 4096 bytes, and the stack is near its end, and then a buffer
of a 8000 bytes is allocated on the stack, the overflow will not be detected.
By default, kernel mode tasks do not have any task stack protection. Configuring
VxWorks with the INCLUDE_PROTECT_TASK_STACK component provides
underflow and overflow guard zones on the execution stacks, but none for the
exception stacks. Stack guard zones in the kernel are mapped to physical memory.
Note that the protection provided for user-mode tasks by configuring the system
with the INCLUDE_RTP component does not apply to kernel tasks (for information
about user-mode task stack protection, see the VxWorks Application Programmer’s
Guide: Multitasking).
Note that the INCLUDE_PROTECT_TASK_STACK component does not provide
stack protection for tasks that are created with the VX_NO_STACK_PROTECT task
option (see 4.4.2 Task Creation Options, p.173). If a task is created with this option,
no guard zones are created for that task.
The size of the guard zones are defined by the following configuration parameters:
■
TASK_KERNEL_EXEC_STACK_OVERFLOW_SIZE for kernel task execution
stack overflow size.
176
4 Multitasking
4.4 Task Creation and Management
■
TASK_KERNEL_EXEC_STACK_UNDERFLOW_SIZE for kernel task execution
stack underflow size.
The value of these parameters can be modified to increase the size of the guard
zones on a system-wide basis. The size of a guard zone is rounded up to a multiple
of the CPU MMU page size. The insertion of a guard zone can be prevented by
setting the parameter to zero.
Stack guard zones in the kernel consume RAM, as guard zones correspond to
mapped memory for which accesses are made invalid.
Non-Executable Task Stacks
VxWorks creates kernel task stacks with a non-executable attribute only if the
system is configured with the INCLUDE_TASK_STACK_NO_EXEC component, and
if the CPU supports making memory non-executable on an MMU-page basis. The
size of a stack is always rounded up to a multiple of an MMU page size when the
stack is made non-executable (as is also the case when guard zones are inserted).
4.4.4 Task Names and IDs
When a task is spawned, you can specify an ASCII string of any length to be the
task name, and a task ID is returned.
Most VxWorks task routines take a task ID as the argument specifying a task.
VxWorks uses a convention that a task ID of 0 (zero) always implies the calling
task. In the kernel, the task ID is a 4-byte handle to the task’s data structures.
The following rules and guidelines should be followed when naming tasks:
■
The names of public tasks must be unique and must begin with a forward
slash; for example /tMyTask. Note that public tasks are visible throughout the
entire system—in the kernel and any processes.
■
The names of private tasks should be unique. VxWorks does not require that
private task names be unique, but it is preferable to use unique names to avoid
confusing the user. (Note that private tasks are visible only within the entity in
which they were created—either the kernel or a process.)
To use the host development tools to their best advantage, task names should not
conflict with globally visible routine or variable names. To avoid name conflicts,
VxWorks uses a convention of prefixing any kernel task name started from the
target with the letter t, and any task name started from the host with the letter u.
In addition, the name of the initial task of a real-time process is the executable file
name (less the extension) prefixed with the letter i.
177
4
VxWorks
Kernel Programmer's Guide, 6.6
Creating a task as a public object allows other tasks from outside of its process to
send signals or events to it (with the taskKill( ) or the eventSend( ) routine,
respectively).
For more information, see 4.9 Public and Private Objects, p.194.
You do not have to explicitly name tasks. If a NULL pointer is supplied for the
name argument of taskSpawn( ), then VxWorks assigns a unique name. The name
is of the form tN, where N is a decimal integer that is incremented by one for each
unnamed task that is spawned.
The taskLib routines listed in Table 4-6 manage task IDs and names.
Table 4-6
Task Name and ID Routines
Call
Description
taskName( )
Gets the task name associated with a task ID (restricted to
the context—process or kernel—in which it is called).
taskNameToId( )
Looks up the task ID associated with a task name.
taskIdSelf( )
Gets the calling task’s ID.
taskIdVerify( )
Verifies the existence of a specified task.
178
4 Multitasking
4.4 Task Creation and Management
4.4.5 Task Information
The routines listed in Table 4-7 get information about a task by taking a snapshot
of a task’s context when the routine is called. Because the task state is dynamic, the
information may not be current unless the task is known to be dormant (that is,
suspended).
Table 4-7
Task Information Routines
Call
Description
taskIdListGet( )
Fills an array with the IDs of all active tasks.
taskInfoGet( )
Gets information about a task.
taskPriorityGet( )
Examines the priority of a task.
taskRegsGet( )
Examines a task’s registers (cannot use with current task).
taskRegsSet( )
Sets a task’s registers (cannot be used with the current task).
taskIsSuspended( ) Checks whether a task is suspended.
taskIsReady( )
Checks whether a task is ready to run.
taskIsPended( )
Checks whether a task is pended.
taskTcb( )
Gets a pointer to a task’s control block.
For information about task-specific variables and their use, see 4.7.3 Task-Specific
Variables, p.190.
179
4
VxWorks
Kernel Programmer's Guide, 6.6
4.4.6 Task Deletion and Deletion Safety
Tasks can be dynamically deleted from the system. VxWorks includes the routines
listed in Table 4-8 to delete tasks and to protect tasks from unexpected deletion.
Table 4-8
Task-Deletion Routines
Call
Description
exit( )
Terminates the calling task and frees memory (task stacks and
task control blocks only).a
taskDelete( )
Terminates a specified task and frees memory (task stacks and
task control blocks only).a The calling task may terminate itself
with this routine.
taskSafe( )
Protects the calling task from deletion.
taskUnsafe( )
Undoes a taskSafe( ), which makes calling task available for
deletion.
a. Memory that is allocated by the task during its execution is not freed when the task is
terminated.
!
WARNING: Make sure that tasks are not deleted at inappropriate times. Before an
application deletes a task, the task should release all shared resources that it holds.
Tasks implicitly call exit( ) if the entry routine specified during task creation
returns.
When a task is deleted, no other task is notified of this deletion. The routines
taskSafe( ) and taskUnsafe( ) address problems that stem from unexpected
deletion of tasks. The routine taskSafe( ) protects a task from deletion by other
tasks. This protection is often needed when a task executes in a critical region or
engages a critical resource.
For example, a task might take a semaphore for exclusive access to some data
structure. While executing inside the critical region, the task might be deleted by
another task. Because the task is unable to complete the critical region, the data
structure might be left in a corrupt or inconsistent state. Furthermore, because the
semaphore can never be released by the task, the critical resource is now
unavailable for use by any other task and is essentially frozen.
Using taskSafe( ) to protect the task that took the semaphore prevents such an
outcome. Any task that tries to delete a task protected with taskSafe( ) is blocked.
180
4 Multitasking
4.4 Task Creation and Management
When finished with its critical resource, the protected task can make itself available
for deletion by calling taskUnsafe( ), which readies any deleting task. To support
nested deletion-safe regions, a count is kept of the number of times taskSafe( ) and
taskUnsafe( ) are called. Deletion is allowed only when the count is zero, that is,
there are as many unsafes as safes. Only the calling task is protected. A task cannot
make another task safe or unsafe from deletion.
The following code fragment shows how to use taskSafe( ) and taskUnsafe( ) to
protect a critical region of code:
taskSafe ();
semTake (semId, WAIT_FOREVER);
.
.
/* critical region code */
.
semGive (semId);
taskUnsafe ();
/* Block until semaphore available */
/* Release semaphore */
Deletion safety is often coupled closely with mutual exclusion, as in this example.
For convenience and efficiency, a special kind of semaphore, the mutual-exclusion
semaphore, offers an option for deletion safety. For more information, see
4.12.3 Mutual-Exclusion Semaphores, p.205.
4.4.7 Task Execution Control
The routines listed in Table 4-9 provide direct control over a task’s execution.
Table 4-9
Task Execution Control Routines
Call
Description
taskSuspend( )
Suspends a task.
taskResume( )
Resumes a task.
taskRestart( )
Restarts a task.
taskDelay( )
Delays a task; delay units are ticks, resolution in ticks.
nanosleep( )
Delays a task; delay units are nanoseconds, resolution in ticks.
Tasks may require restarting during execution in response to some catastrophic
error. The restart mechanism, taskRestart( ), recreates a task with the original
creation arguments.
181
4
VxWorks
Kernel Programmer's Guide, 6.6
Delay operations provide a simple mechanism for a task to sleep for a fixed
duration. Task delays are often used for polling applications. For example, to delay
a task for half a second without making assumptions about the clock rate, call:
taskDelay (sysClkRateGet ( ) / 2);
The routine sysClkRateGet( ) returns the speed of the system clock in ticks per
second. Instead of taskDelay( ), you can use the POSIX routine nanosleep( ) to
specify a delay directly in time units. Only the units are different; the resolution of
both delay routines is the same, and depends on the system clock. For details, see
5.6 POSIX Clocks and Timers, p.259.
As a side effect, taskDelay( ) moves the calling task to the end of the ready queue
for tasks of the same priority. In particular, you can yield the CPU to any other
tasks of the same priority by delaying for zero clock ticks:
taskDelay (NO_WAIT);
/* allow other tasks of same priority to run */
A delay of zero duration is only possible with taskDelay( ); nanosleep( ) considers
it an error.
NOTE: ANSI and POSIX APIs are similar.
System clock resolution is typically 60Hz (60 times per second). This is a relatively
long time for one clock tick, and would be even at 100Hz or 120Hz. Thus, since
periodic delaying is effectively polling, you may want to consider using
event-driven techniques as an alternative.
4.4.8 Tasking Extensions
To allow additional task-related facilities to be added to the system, VxWorks
provides hook routines that allow additional routines to be invoked whenever a
task is created, a task context switch occurs, or a task is deleted. There are spare
fields in the task control block (TCB) available for application extension of a task’s
context
These hook routines are listed in Table 4-10; for more information, see the
VxWorks API reference for taskHookLib.
182
4 Multitasking
4.4 Task Creation and Management
Table 4-10
Task Create, Switch, and Delete Hooks
Call
Description
taskCreateHookAdd( )
Adds a routine to be called at every task create.
taskCreateHookDelete( )
Deletes a previously added task create routine.
taskSwitchHookAdd( )
Adds a routine to be called at every task switch.
taskSwitchHookDelete( )
Deletes a previously added task switch routine.
taskDeleteHookAdd( )
Adds a routine to be called at every task delete.
taskDeleteHookDelete( )
Deletes a previously added task delete routine.
When using switch hook routines, be aware of the following restrictions:
■
Do not assume any virtual memory (VM) context is current other than the
kernel context (as with ISRs).
■
Do not rely on knowledge of the current task or invoke any function that relies
on this information, for example taskIdSelf( ).
■
Do not rely on taskIdVerify (pOldTcb) to determine if a delete hook has
executed for the self-destructing task case. Instead, other state information
must be changed in the delete hook to be detected by the switch hook (for
example by setting a pointer to NULL).
Task create hook routines execute in the context of the creator task.
Task create hooks must consider the ownership of any kernel objects (such as
watchdog timers, semaphores, and so on) created in the hook routine. Since create
hook routines execute in the context of the creator task, new kernel objects will be
owned by the creator task's process. It may be necessary to assign the ownership
of these objects to the new task's process. This will prevent unexpected object
reclamation from occurring if and when the process of the creator task terminates.
When the creator task is a kernel task, the kernel will own any kernel objects that
are created. Thus there is no concern about unexpected object reclamation for this
case.
User-installed switch hooks are called within the kernel context and therefore do
not have access to all VxWorks facilities. Table 4-11 summarizes the routines that
can be called from a task switch hook; in general, any routine that does not involve
the kernel can be called.
183
4
VxWorks
Kernel Programmer's Guide, 6.6
Table 4-11
Routines Callable by Task Switch Hooks
Library
Routines
bLib
All routines
fppArchLib
fppSave( ), fppRestore( )
intLib
intContext( ), intCount( ), intVecSet( ), intVecGet( ), intLock( ),
intUnlock( )
lstLib
All routines except lstFree( )
mathALib
All are callable if fppSave( )/fppRestore( ) are used
rngLib
All routines except rngCreate( )
taskLib
taskIdVerify( ), taskIdDefault( ), taskIsReady( ),
taskIsSuspended( ), taskTcb( )
vxLib
vxTas( )
NOTE: For information about POSIX extensions, see 5. POSIX Facilities.
4.5 Task Error Status: errno
By convention, C library functions set a single global integer variable errno to an
appropriate error number whenever the function encounters an error. This
convention is specified as part of the ANSI C standard.
NOTE: This section describes the implementation and use of errno in UP
configurations of VxWorks, which is different from that in SMP configurations.For
information about errno and other global variables in VxWorks SMP, see
15.15.8 SMP CPU-Specific Variables and Uniprocessor Global Variables, p.712. For
information about migration, see 15.15 Migrating Code to VxWorks SMP, p.702.
184
4 Multitasking
4.5 Task Error Status: errno
4.5.1 Layered Definitions of errno
In VxWorks, errno is simultaneously defined in two different ways. There is, as in
ANSI C, an underlying global variable called errno, which you can display by
name using host development tools.
However, errno is also defined as a macro in errno.h; this is the definition visible
to all of VxWorks except for one function. The macro is defined as a call to a
function __errno( )that returns the address of the global variable, errno (as you
might guess, this is the single function that does not itself use the macro definition
for errno). This subterfuge yields a useful feature: because __errno( )is a function,
you can place breakpoints on it while debugging, to determine where a particular
error occurs.
Nevertheless, because the result of the macro errno is the address of the global
variable errno, C programs can set the value of errno in the standard way:
errno = someErrorNumber;
As with any other errno implementation, take care not to have a local variable of
the same name.
4.5.2 A Separate errno Value for Each Task
In VxWorks, the underlying global errno is a single predefined global variable that
can be referenced directly by application code that is linked with VxWorks (either
statically on the host or dynamically at load time).
However, for errno to be useful in the multitasking environment of VxWorks, each
task must see its own version of errno. Therefore errno is saved and restored by
the kernel as part of each task’s context every time a context switch occurs.
Similarly, interrupt service routines (ISRs) see their own versions of errno. This is
accomplished by saving and restoring errno on the interrupt stack as part of the
interrupt enter and exit code provided automatically by the kernel (see
4.20.1 Connecting Routines to Interrupts, p.242).
Thus, regardless of the VxWorks context, an error code can be stored or consulted
without direct manipulation of the global variable errno.
185
4
VxWorks
Kernel Programmer's Guide, 6.6
4.5.3 Error Return Convention
Almost all VxWorks functions follow a convention that indicates simple success or
failure of their operation by the actual return value of the function. Many functions
return only the status values OK (0) or ERROR (-1). Some functions that normally
return a nonnegative number (for example, open( ) returns a file descriptor) also
return ERROR to indicate an error. Functions that return a pointer usually return
NULL (0) to indicate an error. In most cases, a function returning such an error
indication also sets errno to the specific error code.
The global variable errno is never cleared by VxWorks routines. Thus, its value
always indicates the last error status set. When a VxWorks subroutine gets an error
indication from a call to another routine, it usually returns its own error indication
without modifying errno. Thus, the value of errno that is set in the lower-level
routine remains available as the indication of error type.
For example, the VxWorks routine intConnect( ), which connects a user routine to
a hardware interrupt, allocates memory by calling malloc( ) and builds the
interrupt driver in this allocated memory. If malloc( ) fails because insufficient
memory remains in the pool, it sets errno to a code indicating an
insufficient-memory error was encountered in the memory allocation library,
memLib. The malloc( ) routine then returns NULL to indicate the failure. The
intConnect( ) routine, receiving the NULL from malloc( ), then returns its own
error indication of ERROR. However, it does not alter errno leaving it at the
insufficient memory code set by malloc( ). For example:
if ((pNew = malloc (CHUNK_SIZE)) = = NULL)
return (ERROR);
It is recommended that you use this mechanism in your own subroutines, setting
and examining errno as a debugging technique. A string constant associated with
errno can be displayed using printErrno( ) if the errno value has a corresponding
string entered in the error-status symbol table, statSymTbl. See the VxWorks API
reference for errnoLib for details on error-status values and building statSymTbl.
4.5.4 Assignment of Error Status Values
VxWorks errno values encode the module that issues the error, in the most
significant two bytes, and uses the least significant two bytes for individual error
numbers. All VxWorks module numbers are in the range 1–500; errno values with
a module number of zero are used for source compatibility.
All other errno values (that is, positive values greater than or equal to 501<<16, and
all negative values) are available for application use.
186
4 Multitasking
4.6 Task Exception Handling
See the VxWorks API reference on errnoLib for more information about defining
and decoding errno values with this convention.
4
4.6 Task Exception Handling
Errors in program code or data can cause hardware exception conditions such as
illegal instructions, bus or address errors, divide by zero, and so forth. The
VxWorks exception handling package takes care of all such exceptions (see
11. Error Detection and Reporting).
Tasks can also attach their own handlers for certain hardware exceptions through
the signal facility. If a task has supplied a signal handler for an exception, the
default exception handling described above is not performed. A user-defined
signal handler is useful for recovering from catastrophic events. Typically,
setjmp( ) is called to define the point in the program where control will be
restored, and longjmp( ) is called in the signal handler to restore that context. Note
that longjmp( ) restores the state of the task’s signal mask.
Signals are also used for signaling software exceptions as well as hardware
exceptions. They are described in more detail in 4.18 Signals, p.226 and in the
VxWorks API reference for sigLib.
4.7 Shared Code and Reentrancy
In VxWorks, it is common for a single copy of a subroutine or subroutine library
to be invoked by many different tasks. For example, many tasks may call printf( ),
but there is only a single copy of the subroutine in the system. A single copy of
code executed by multiple tasks is called shared code. VxWorks dynamic linking
facilities make this especially easy. Shared code makes a system more efficient and
easier to maintain; see Figure 4-4.
187
VxWorks
Kernel Programmer's Guide, 6.6
Figure 4-4
Shared Code
TASKS
taskOne (void)
{
myFunc();
...
}
taskTwo (void)
{
myFunc();
...
}
SHARED CODE
myFunc (void)
{
...
}
Shared code must be reentrant. A subroutine is reentrant if a single copy of the
routine can be called from several task contexts simultaneously without conflict.
Such conflict typically occurs when a subroutine modifies global or static
variables, because there is only a single copy of the data and code. A routine’s
references to such variables can overlap and interfere in invocations from different
task contexts.
Most routines in VxWorks are reentrant. However, you should assume that any
routine someName( ) is not reentrant if there is a corresponding routine named
someName_r( ) — the latter is provided as a reentrant version of the routine. For
example, because ldiv( ) has a corresponding routine ldiv_r( ), you can assume
that ldiv( ) is not reentrant.
VxWorks I/O and driver routines are reentrant, but require careful application
design. For buffered I/O, Wind River recommends using file-pointer buffers on a
per-task basis. At the driver level, it is possible to load buffers with streams from
different tasks, due to the global file descriptor table in VxWorks.
This may or may not be desirable, depending on the nature of the application. For
example, a packet driver can mix streams from different tasks because the packet
header identifies the destination of each packet.
The majority of VxWorks routines use the following reentrancy techniques:
–
–
–
dynamic stack variables
global and static variables guarded by semaphores
task variables
Wind River recommends applying these same techniques when writing
application code that can be called from several task contexts simultaneously.
188
4 Multitasking
4.7 Shared Code and Reentrancy
NOTE: In some cases reentrant code is not preferable. A critical section should use
a binary semaphore to guard it, or use intLock( ) or intUnlock( ) if called from by
an ISR.
NOTE: Initialization routines should be callable multiple times, even if logically
they should only be called once. As a rule, routines should avoid static variables
that keep state information. Initialization routines are an exception; using a static
variable that returns the success or failure of the original initialization routine call
is appropriate.
4.7.1 Dynamic Stack Variables
Many subroutines are pure code, having no data of their own except dynamic stack
variables. They work exclusively on data provided by the caller as parameters. The
linked-list library, lstLib, is a good example of this. Its routines operate on lists and
nodes provided by the caller in each subroutine call.
Subroutines of this kind are inherently reentrant. Multiple tasks can use such
routines simultaneously, without interfering with each other, because each task
does indeed have its own stack. See Figure 4-5.
Figure 4-5
Stack Variables and Shared Code
TASKS
taskOne ( )
{
...
comFunc(1);
...
}
taskTwo ( )
{
...
comFunc(2);
...
}
TASK STACKS
COMMON SUBROUTINE
...
var = 1
...
...
var = 2
...
comFunc (arg)
{
int var = arg;
}
189
4
VxWorks
Kernel Programmer's Guide, 6.6
4.7.2 Guarded Global and Static Variables
Some libraries encapsulate access to common data. This kind of library requires
some caution because the routines are not inherently reentrant. Multiple tasks
simultaneously invoking the routines in the library might interfere with access to
common variables. Such libraries must be made explicitly reentrant by providing
a mutual-exclusion mechanism to prohibit tasks from simultaneously executing
critical sections of code. The usual mutual-exclusion mechanism is the mutex
semaphore facility provided by semMLib and described in
4.12.3 Mutual-Exclusion Semaphores, p.205.
4.7.3 Task-Specific Variables
Task-specific variables can be used to ensure that shared code is reentrant by
providing task-specific variables of the same name that are located in each task’s
stack, instead of a standard global or static variables. Each task thereby has its own
unique copy of the data item.This allows, for example, several tasks to reference a
private buffer of memory and while referring to it with the same global variable
name.
NOTE: The __thread storage class variables can be used for both UP and SMP
configurations of VxWorks, and Wind River recommends its use in both cases as
the best method of providing task-specific variables. The taskVarLib and
tlsOldLib (formerly tlsLib) facilities—for the kernel-space and user-space
respectively— are maintained primarily for backwards-compatibility, are not
compatible with VxWorks SMP, and their use is not recommended. In addition to
being incompatible with VxWorks SMP, the taskVarLib and tlsOldLib facilities
increase task context switch times. For information about migration, see
15.15 Migrating Code to VxWorks SMP, p.702.
Also note that each task has a VxWorks events register, which receives events sent
from other tasks, ISRs, semaphores, or message queues. See 4.15 VxWorks Events,
p.219 for more information about this register, and the routines used to interact
with it.
Thread-Local Variables: __thread Storage Class
Thread-local storage is a compiler facility that allows for allocation of a variable
such that there are unique instances of the variable for each thread (or task, in
VxWorks terms).
190
4 Multitasking
4.7 Shared Code and Reentrancy
The __thread storage class instructs the compiler to make the defined variable a
thread-local variable. This means one instance of the variable is created for every
task in the system. The compiler key word is used as follows:
__thread int i;
4
extern __thread struct state s;
static __thread char *p;
The __thread specifier may be used alone, with the extern or static specifiers, but
with no other storage class specifier. When used with extern or static, __thread
must appear immediately after the other storage class specifier.
The __thread specifier may be applied to any global, file-scoped static,
function-scoped static, or static data member of a class. It may not be applied to
block-scoped automatic or non-static data member.
When the address-of operator is applied to a thread-local variable, it is evaluated
at run-time and returns the address of the current task’s instance of that variable.
The address may be used by any task. When a task terminates, any pointers to
thread-local variables in that task become invalid.
No static initialization may refer to the address of a thread-local variable.
In C++, if an initializer is present for a thread-local variable, it must be a
constant-expression, as defined in 5.19.2 of the ANSI/ISO C++ standard.
taskVarLib and Task Variables
VxWorks provides a task variable facility (with taskVarLib) that allows 4-byte
variables to be added to a task’s context, so that the value of such a variable is
switched every time a task switch occurs to or from its owner task.
NOTE: Wind River does not recommend using the taskVarLib facility, which is
maintained primarily for backwards-compatibility. Use thread-local (__thread)
storage class variables instead.
191
VxWorks
Kernel Programmer's Guide, 6.6
4.7.4 Multiple Tasks with the Same Main Routine
With VxWorks, it is possible to spawn several tasks with the same main routine.
Each spawn creates a new task with its own stack and context. Each spawn can also
pass the main routine different parameters to the new task. In this case, the same
rules of reentrancy described in 4.7.3 Task-Specific Variables, p.190 apply to the
entire task.
This is useful when the same function must be performed concurrently with
different sets of parameters. For example, a routine that monitors a particular kind
of equipment might be spawned several times to monitor several different pieces
of that equipment. The arguments to the main routine could indicate which
particular piece of equipment the task is to monitor.
In Figure 4-6, multiple joints of the mechanical arm use the same code. The tasks
manipulating the joints invoke joint( ). The joint number (jointNum) is used to
indicate which joint on the arm to manipulate.
Figure 4-6
Multiple Tasks Utilizing Same Code
joint_2
joint_3
joint_1
joint
(
int jointNum
)
{
/* joint code here */
}
192
4 Multitasking
4.8 Intertask and Interprocess Communication
4.8 Intertask and Interprocess Communication
The complement to the multitasking routines described in 4.2 Tasks and
Multitasking, p.160 is the intertask communication facilities. These facilities permit
independent tasks to coordinate their actions.
VxWorks supplies a rich set of intertask and interprocess communication
mechanisms, including:
■
Shared memory, for simple sharing of data.
■
Semaphores, for basic mutual exclusion and synchronization.
■
Mutexes and condition variables for mutual exclusion and synchronization using
POSIX interfaces.
■
Message queues and pipes, for intertask message passing within a CPU.
■
VxWorks events, for communication and synchronization.
■
Message channels, for socket-based inter-processor and interprocess
communication.
■
Sockets and remote procedure calls, for network-transparent intertask
communication.
Signals, for exception handling, interprocess communication, and process
management.
■
NOTE: With few exceptions, the symmetric multiprocessor (SMP) and
uniprocessor (UP) configurations of VxWorks share the same facilities for intertask
and interprocess communications—the difference amounts to only a few routines.
This section provides information about the APIs that are common to both
configurations, as well as those APIs that are specific to the UP configuration. In
the latter case, the alternatives available for SMP systems are noted. For
information about the SMP configuration of VxWorks, see 15. VxWorks SMP; and
for information specifically about migration, see 15.15 Migrating Code to VxWorks
SMP, p.702.
In addition, the VxMP component provides for intertask communication between
multiple CPUs that share memory. See 16. Shared-Memory Objects: VxMP.
193
4
VxWorks
Kernel Programmer's Guide, 6.6
4.9 Public and Private Objects
Kernel objects such as semaphores and message queues can be created as either
private or public objects. This provides control over the scope of their
accessibility—which can be limited to a virtual memory context by defining them
as private, or extended to the entire system (the kernel and any processes) by
defining them as public. There is no difference in performance between a public
and a private object.
An object can only be defined as public or private when it is created—the
designation cannot be changed thereafter. Public objects must be named when
they are created, and the name must begin with a forward slash; for example, /foo.
Private objects do not need to be named.
For information about naming tasks in addition to that provided in this section, see
4.4.4 Task Names and IDs, p.177.
4.9.1 Creating and Naming Public and Private Objects
Public objects are always named, and the name must begin with a forward-slash.
Private objects can be named or unnamed. If they are named, the name must not
begin with a forward-slash.
Only one public object of a given class and name can be created. That is, there can
be only one public semaphore with the name /foo. But there may be a public
semaphore named /foo and a public message queue named /foo. Obviously, more
distinctive naming is preferable (such as /fooSem and /fooMQ).
The system allows creation of only one private object of a given class and name in
any given memory context; that is, in any given process or in the kernel. For
example:
■
If process A has created a private semaphore named bar, it cannot create a
second semaphore named bar.
■
However, process B could create a private semaphore named bar, as long as it
did not already own one with that same name.
Note that private tasks are an exception to this rule—duplicate names are
permitted for private tasks; see 4.4.4 Task Names and IDs, p.177.
To create a named object, the appropriate xyzOpen( ) API must be used, such as
semOpen( ). When the routine specifies a name that starts with a forward slash,
the object will be public.
194
4 Multitasking
4.9 Public and Private Objects
To delete public objects, the xyzDelete( ) API cannot be used (it can only be used
with private objects). Instead, the xyzClose( ) and xyzUnlink( ) APIs must be used
in accordance with the POSIX standard. That is, they must be unlinked from the
name space, and then the last close operation will delete the object (for example,
using the semUnlink( ) and semClose( ) APIs for a public semaphore).
Alternatively, all close operations can be performed first, and then the unlink
operation, after which the object is deleted. Note that if an object is created with the
OM_DELETE_ON_LAST_CLOSE flag, it is be deleted with the last close operation,
regardless of whether or not it was unlinked.
4.9.2 Object Ownership and Resource Reclamation
All objects are owned by the process to which the creator task belongs, or by the
kernel if the creator task is a kernel task. When ownership must be changed, for
example on a process creation hook, the objOwnerSet( ) can be used. However, its
use is restricted—the new owner must be a process or the kernel.
All objects that are owned by a process are automatically destroyed when the
process dies.
All objects that are children of another object are automatically destroyed when the
parent object is destroyed.
Processes can share public objects through an object lookup-by-name capability
(with the xyzOpen( ) set of routines). Sharing objects between processes can only
be done by name.
When a process terminates, all the private objects that it owns are deleted,
regardless of whether or not they are named. All references to public objects in the
process are closed (an xyzClose( ) operation is performed). Therefore, any public
object is deleted during resource reclamation, regardless of which process created
them, if there are no more outstanding xyzOpen( ) calls against it (that is, no other
process or the kernel has a reference to it), and the object was already unlinked or
was created with the OM_DELETE_ON_LAST_CLOSE option. The exception to this
rule is tasks, which are always reclaimed when its creator process dies.
When the creator process of a public object dies, but the object survives because it
hasn't been unlinked or because another process has a reference to it, ownership of
the object is assigned to the kernel.
The objShowAll( ) show routine can be used to display information about
ownership relations between objects.
195
4
VxWorks
Kernel Programmer's Guide, 6.6
4.10 Shared Data Structures
The most obvious way for tasks executing in the same memory space (either a
process or the kernel) to communicate is by accessing shared data structures.
Because all the tasks in a single process or in the kernel exist in a single linear
address space, sharing data structures between tasks is trivial; see Figure 4-7.
Global variables, linear buffers, ring buffers, linked lists, and pointers can be
referenced directly by code running in different contexts.
For information about using shared data regions to communicate between
processes, see VxWorks Application Programmer’s Guide: Applications and Processes.
Figure 4-7
Shared Data Structures
TASKS
task 1
MEMORY
access
sharedData
sharedData
access
task 2
sharedData
task 3
sharedData
access
4.11 Mutual Exclusion
While a shared address space simplifies exchange of data, interlocking access to
memory is crucial to avoid contention. Many methods exist for obtaining exclusive
access to resources, and vary only in the scope of the exclusion. Such methods
include disabling interrupts, disabling preemption, and resource locking with
semaphores.
For information about POSIX mutexes, see 5.11 POSIX Thread Mutexes and
Condition Variables, p.273.
196
4 Multitasking
4.11 Mutual Exclusion
4.11.1 Interrupt Locks and Latency
The most powerful method available for mutual exclusion is the disabling of
interrupts with the intLock( ) routine. Such a lock guarantees exclusive access to
the CPU:
funcA ()
{
int lock = intLock();
.
.
/* critical region of code that cannot be interrupted */
.
intUnlock (lock);
}
While this solves problems involving mutual exclusion with ISRs, it is
inappropriate as a general-purpose mutual-exclusion method for most real-time
systems, because it prevents the system from responding to external events for the
duration of these locks. Interrupt latency is unacceptable whenever an immediate
response to an external event is required. However, interrupt locking can
sometimes be necessary where mutual exclusion involves ISRs. In any situation,
keep the duration of interrupt lockouts short.
For information about ISRs, see 4.20 Interrupt Service Routines, p.241.
!
WARNING: Invoking a VxWorks system routine with interrupts locked may result
in interrupts being re-enabled for an unspecified period of time. If the called
routine blocks, or results in a higher priority task becoming eligible for execution
(READY), interrupts will be re-enabled while another task executes, or while the
kernel is idle.
NOTE: The intLock( ) routine is provided for the UP configuration of VxWorks,
but not the SMP configuration. Several alternative are available for SMP systems,
including the ISR-callable spinlock, which defaults to intLock( ) behavior in a UP
system. For more information, see 15.6.1 ISR-Callable Spinlocks, p.682 and
15.15 Migrating Code to VxWorks SMP, p.702.
4.11.2 Preemptive Locks and Latency
Disabling preemption with the taskLock( ) routine offers a somewhat less
restrictive form of mutual exclusion. While no other task is allowed to preempt the
current executing task, ISRs are able to execute:
197
4
VxWorks
Kernel Programmer's Guide, 6.6
funcA ()
{
taskLock ();
.
. /* critical region of code that cannot be interrupted */
.
taskUnlock ();
}
However, this method can lead to unacceptable real-time response. Tasks of higher
priority are unable to execute until the locking task leaves the critical region, even
though the higher-priority task is not itself involved with the critical region. While
this kind of mutual exclusion is simple, if you use it, be sure to keep the duration
short. Semaphores provide a better mechanism; see 4.12 Semaphores, p.198.
!
WARNING: The critical region code should not block. If it does, preemption could
be re-enabled.
NOTE: The taskLock( ) and taskUnlock( ) routines are provided for the UP
configuration of VxWorks, but not the SMP configuration. Several alternative are
available for SMP systems, including task-only spinlocks, which default to
taskLock( ) and taskUnlock( ) behavior in a UP system. For more information, see
15.6.2 Task-Only Spinlocks, p.682 and 15.15 Migrating Code to VxWorks SMP, p.702.
4.12 Semaphores
VxWorks semaphores are highly optimized and provide the fastest intertask
communication mechanism in VxWorks. Semaphores are the primary means for
addressing the requirements of both mutual exclusion and task synchronization,
as described below:
■
For mutual exclusion, semaphores interlock access to shared resources. They
provide mutual exclusion with finer granularity than either interrupt
disabling or preemptive locks, discussed in 4.11 Mutual Exclusion, p.196.
■
For synchronization, semaphores coordinate a task’s execution with external
events.
198
4 Multitasking
4.12 Semaphores
NOTE: Semaphores provide full memory barriers, which is of particular
significance for the SMP configuration of VxWorks. For more information, see
15.8 Memory Barriers, p.687.
VxWorks provides the following types of semaphores, which are optimized for
different types of uses:
binary
The fastest, most general-purpose semaphore. Optimized for synchronization
or mutual exclusion. For more information, see 4.12.2 Binary Semaphores, p.201.
mutual exclusion
A special binary semaphore optimized for problems inherent in mutual
exclusion: priority inversion, deletion safety, and recursion. For more
information, see 4.12.3 Mutual-Exclusion Semaphores, p.205.
counting
Like the binary semaphore, but keeps track of the number of times a
semaphore is given. Optimized for guarding multiple instances of a resource.
For more information, see 4.12.4 Counting Semaphores, p.208.
read/write
A special type of semaphore that provides mutual exclusion for tasks that need
write access to an object, and concurrent access for tasks that only need read
access to the object. This type of semaphore is particularly useful for SMP
systems. For more information, see 4.12.5 Read/Write Semaphores, p.209.
VxWorks semaphores can be created as private objects, which are accessible only
within the memory space in which they were created (kernel or process); or as
public objects, which accessible throughout the system. For more information, see
4.9 Public and Private Objects, p.194.
VxWorks not only provides the semaphores designed expressly for VxWorks, but
also POSIX semaphores, designed for portability. An alternate semaphore library
provides the POSIX-compliant semaphore interface; see 5.13 POSIX Semaphores,
p.289.
NOTE: The semaphores described here are for use with UP and SMP
configurations of VxWorks. The optional product VxMP provides semaphores
that can be used in an asymmetric multiprocessor (AMP) system, in the VxWorks
kernel (but not in UP or SMP systems). For more information, see
16. Shared-Memory Objects: VxMP.
199
4
VxWorks
Kernel Programmer's Guide, 6.6
4.12.1 Semaphore Control
In most cases, VxWorks provides a single, uniform interface for semaphore
control—instead of defining a full set of semaphore control routines specific to
each type of semaphore.
The exceptions are the creation routines, which are specific to each semaphore
type; and the give and take routines for read/write semaphores, which support
read and write modes for each operation.
Table 4-12 lists the semaphore control routines.
Table 4-12
Semaphore Control Routines
Call
Description
semBCreate( )
Allocates and initializes a binary semaphore.
semMCreate( )
Allocates and initializes a mutual-exclusion semaphore.
semCCreate( )
Allocates and initializes a counting semaphore.
semRWCreate( ) Allocates and initializes a read/write semaphore.
semDelete( )
Terminates and frees a semaphore.
semTake( )
Takes a binary, mutual-exclusion, or counting semaphore.,
or a read/write semaphore in write mode.
semRTake( )
Takes a read/write semaphore in read mode.
semWTake( )
Takes a read/write semaphore in write mode.
semGive( )
Gives a binary, mutual -exclusion, or counting semaphore.
semRWGive( )
Gives a read/write semaphore.
semFlush( )
Unblocks all tasks that are waiting for a semaphore.
semExchange( )
Provides for an atomic give and exchange of semaphores in
SMP systems.
The creation routines return a semaphore ID that serves as a handle on the
semaphore during subsequent use by the other semaphore-control routines. When
a semaphore is created, the queue type is specified. Tasks pending on a semaphore
can be queued in priority order (SEM_Q_PRIORITY) or in first-in first-out order
(SEM_Q_FIFO).
200
4 Multitasking
4.12 Semaphores
!
WARNING: The semDelete( ) call terminates a semaphore and deallocates all
associated memory. Take care when deleting semaphores, particularly those used
for mutual exclusion, to avoid deleting a semaphore that another task still requires.
Do not delete a semaphore unless the same task first succeeds in taking it.
4
Static Instantiation of Semaphores
The semaphore creation routines listed in Table 4-12 perform a dynamic, two-step
operation, in which memory is allocated for the semaphore object at runtime, and
then the object is initialized. Semaphores (and other VxWorks objects) can also be
statically instantiated—which means that their memory is allocated for the object
at compile time—and the object is then initialized at runtime with an initialization
routine.
For information about static instantiation, see 2.6.4 Static Instantiation of Kernel
Objects, p.56. For information about semaphore initialization routines, see the
VxWorks API references.
4.12.2 Binary Semaphores
The general-purpose binary semaphore is capable of addressing the requirements
of both forms of task coordination: mutual exclusion and synchronization. The
binary semaphore has the least overhead associated with it, making it particularly
applicable to high-performance requirements. The mutual-exclusion semaphore
described in 4.12.3 Mutual-Exclusion Semaphores, p.205 is also a binary semaphore,
but it has been optimized to address problems inherent to mutual exclusion.
Alternatively, the binary semaphore can be used for mutual exclusion if the
advanced features of the mutual-exclusion semaphore are deemed unnecessary.
A binary semaphore can be viewed as a flag that is available (full) or unavailable
(empty). When a task takes a binary semaphore, with semTake( ), the outcome
depends on whether the semaphore is available (full) or unavailable (empty) at the
time of the call; see Figure 4-8. If the semaphore is available (full), the semaphore
becomes unavailable (empty) and the task continues executing immediately. If the
semaphore is unavailable (empty), the task is put on a queue of blocked tasks and
enters a state of pending on the availability of the semaphore.
201
VxWorks
Kernel Programmer's Guide, 6.6
Figure 4-8
Taking a Semaphore
no
no
semaphore
available?
NO_WAIT
timeout =
yes
yes
task continues;
semaphore
taken
task is
pended for
timeout
value
task continues;
semaphore
not taken
When a task gives a binary semaphore, using semGive( ), the outcome also
depends on whether the semaphore is available (full) or unavailable (empty) at the
time of the call; see Figure 4-9. If the semaphore is already available (full), giving
the semaphore has no effect at all. If the semaphore is unavailable (empty) and no
task is waiting to take it, then the semaphore becomes available (full). If the
semaphore is unavailable (empty) and one or more tasks are pending on its
availability, then the first task in the queue of blocked tasks is unblocked, and the
semaphore is left unavailable (empty).
Figure 4-9
Giving a Semaphore
no
semaphore
available?
yes
task continues;
semaphore
remains
unchanged
202
no
tasks
pended?
yes
task at front of
queue made ready;
semaphore remains
unavailable
task continues,
semaphore
made available
4 Multitasking
4.12 Semaphores
Mutual Exclusion
Binary semaphores interlock access to a shared resource efficiently. Unlike
disabling interrupts or preemptive locks, binary semaphores limit the scope of the
mutual exclusion to only the associated resource. In this technique, a semaphore is
created to guard the resource. Initially the semaphore is available (full).
/* includes */
#include
#include
SEM_ID semMutex;
/* Create a binary semaphore that is initially full. Tasks *
* blocked on semaphore wait in priority order.
*/
semMutex = semBCreate (SEM_Q_PRIORITY, SEM_FULL);
When a task wants to access the resource, it must first take that semaphore. As long
as the task keeps the semaphore, all other tasks seeking access to the resource are
blocked from execution. When the task is finished with the resource, it gives back
the semaphore, allowing another task to use the resource.
Thus, all accesses to a resource requiring mutual exclusion are bracketed with
semTake( ) and semGive( ) pairs:
semTake (semMutex, WAIT_FOREVER);
.
. /* critical region, only accessible by a single task at a time */
.
semGive (semMutex);
Synchronization
When used for task synchronization, a semaphore can represent a condition or
event that a task is waiting for. Initially, the semaphore is unavailable (empty). A
task or ISR signals the occurrence of the event by giving the semaphore. Another
task waits for the semaphore by calling semTake( ). The waiting task blocks until
the event occurs and the semaphore is given.
(See 4.20 Interrupt Service Routines, p.241 for a complete discussion of ISRs)
Note the difference in sequence between semaphores used for mutual exclusion
and those used for synchronization. For mutual exclusion, the semaphore is
initially full, and each task first takes, then gives back the semaphore. For
synchronization, the semaphore is initially empty, and one task waits to take the
semaphore given by another task.
203
4
VxWorks
Kernel Programmer's Guide, 6.6
In Example 4-2, the init( ) routine creates the binary semaphore, attaches an ISR to
an event, and spawns a task to process the event. The routine task1( ) runs until it
calls semTake( ). It remains blocked at that point until an event causes the ISR to
call semGive( ). When the ISR completes, task1( ) executes to process the event.
There is an advantage of handling event processing within the context of a
dedicated task: less processing takes place at interrupt level, thereby reducing
interrupt latency. This model of event processing is recommended for real-time
applications.
Example 4-2
Using Semaphores for Task Synchronization
/* This example shows the use of semaphores for task synchronization. */
/* includes */
#include
#include
#include /* replace arch with architecture type */
SEM_ID syncSem;
/* ID of sync semaphore */
init (
int someIntNum
)
{
/* connect interrupt service routine */
intConnect (INUM_TO_IVEC (someIntNum), eventInterruptSvcRout, 0);
/* create semaphore */
syncSem = semBCreate (SEM_Q_FIFO, SEM_EMPTY);
/* spawn task used for synchronization. */
taskSpawn ("sample", 100, 0, 20000, task1, 0,0,0,0,0,0,0,0,0,0);
}
task1 (void)
{
...
semTake (syncSem, WAIT_FOREVER); /* wait for event to occur */
printf ("task 1 got the semaphore\n");
...
/* process event */
}
eventInterruptSvcRout (void)
{
...
semGive (syncSem);
/* let task 1 process event */
...
}
Broadcast synchronization allows all processes that are blocked on the same
semaphore to be unblocked atomically. Correct application behavior often
requires a set of tasks to process an event before any task of the set has the
204
4 Multitasking
4.12 Semaphores
opportunity to process further events. The routine semFlush( ) addresses this class
of synchronization problem by unblocking all tasks pended on a semaphore.
4.12.3 Mutual-Exclusion Semaphores
4
The mutual-exclusion semaphore is a specialized binary semaphore designed to
address issues inherent in mutual exclusion, including priority inversion, deletion
safety, and recursive access to resources.
The fundamental behavior of the mutual-exclusion semaphore is identical to the
binary semaphore, with the following exceptions:
■
■
■
■
It can be used only for mutual exclusion.
It can be given only by the task that took it.
It cannot be given from an ISR.
The semFlush( ) operation is illegal.
Priority Inversion
Figure 4-10 illustrates a situation called priority inversion.
Priority Inversion
t1
t1
HIGH
priority
Figure 4-10
LOW
t2
t3
t3
t3
time
KEY:
= take semaphore
= preemption
= give semaphore
= priority inheritance/release
= own semaphore
= block
205
VxWorks
Kernel Programmer's Guide, 6.6
Priority inversion arises when a higher-priority task is forced to wait an indefinite
period of time for a lower-priority task to complete. Consider the scenario in
Figure 4-10: t1, t2, and t3 are tasks of high, medium, and low priority, respectively.
t3 has acquired some resource by taking its associated binary guard semaphore.
When t1 preempts t3 and contends for the resource by taking the same semaphore,
it becomes blocked. If we could be assured that t1 would be blocked no longer than
the time it normally takes t3 to finish with the resource, there would be no problem
because the resource cannot be preempted. However, the low-priority task is
vulnerable to preemption by medium-priority tasks (like t2), which could inhibit
t3 from relinquishing the resource. This condition could persist, blocking t1 for an
indefinite period of time.
The mutual-exclusion semaphore has the option SEM_INVERSION_SAFE, which
enables a priority-inheritance policy. The priority-inheritance policy assures that a
task that holds a resource executes at the priority of the highest-priority task
blocked on that resource. Once the task priority has been elevated, it remains at the
higher level until all mutual-exclusion semaphores that have contributed to the
tasks elevated priority are released. Hence, the inheriting task is protected from
preemption by any intermediate-priority tasks. This option must be used in
conjunction with a priority queue (SEM_Q_PRIORITY).
Priority Inheritance
t1
HIGH
t3
t1
priority
Figure 4-11
LOW
t2
t3
time
KEY:
206
= take semaphore
= preemption
= give semaphore
= priority inheritance/release
= own semaphore
= block
4 Multitasking
4.12 Semaphores
In Figure 4-11, priority inheritance solves the problem of priority inversion by
elevating the priority of t3 to the priority of t1 during the time t1 is blocked on the
semaphore. This protects t3, and indirectly t1, from preemption by t2.
The following example creates a mutual-exclusion semaphore that uses the
priority inheritance policy:
semId = semMCreate (SEM_Q_PRIORITY | SEM_INVERSION_SAFE);
Deletion Safety
Another problem of mutual exclusion involves task deletion. Within a critical
region guarded by semaphores, it is often desirable to protect the executing task
from unexpected deletion. Deleting a task executing in a critical region can be
catastrophic. The resource might be left in a corrupted state and the semaphore
guarding the resource left unavailable, effectively preventing all access to the
resource.
The primitives taskSafe( ) and taskUnsafe( ) provide one solution to task deletion.
However, the mutual-exclusion semaphore offers the option SEM_DELETE_SAFE,
which enables an implicit taskSafe( ) with each semTake( ), and a taskUnsafe( )
with each semGive( ). In this way, a task can be protected from deletion while it
has the semaphore. This option is more efficient than the primitives taskSafe( )
and taskUnsafe( ), as the resulting code requires fewer entrances to the kernel.
semId = semMCreate (SEM_Q_FIFO | SEM_DELETE_SAFE);
Recursive Resource Access
Mutual-exclusion semaphores can be taken recursively. This means that the
semaphore can be taken more than once by the task that holds it before finally
being released. Recursion is useful for a set of routines that must call each other but
that also require mutually exclusive access to a resource. This is possible because
the system keeps track of which task currently holds the mutual-exclusion
semaphore.
Before being released, a mutual-exclusion semaphore taken recursively must be
given the same number of times it is taken. This is tracked by a count that
increments with each semTake( ) and decrements with each semGive( ).
207
4
VxWorks
Kernel Programmer's Guide, 6.6
Example 4-3
Recursive Use of a Mutual-Exclusion Semaphore
/* Function A requires access to a resource which it acquires by taking
* mySem;
* Function A may also need to call function B, which also requires mySem:
*/
/* includes */
#include
#include
SEM_ID mySem;
/* Create a mutual-exclusion semaphore. */
init ()
{
mySem = semMCreate (SEM_Q_PRIORITY);
}
funcA ()
{
semTake (mySem, WAIT_FOREVER);
printf ("funcA: Got mutual-exclusion semaphore\n");
...
funcB ();
...
semGive (mySem);
printf ("funcA: Released mutual-exclusion semaphore\n");
}
funcB ()
{
semTake (mySem, WAIT_FOREVER);
printf ("funcB: Got mutual-exclusion semaphore\n");
...
semGive (mySem);
printf ("funcB: Releases mutual-exclusion semaphore\n");
}
4.12.4 Counting Semaphores
Counting semaphores are another means to implement task synchronization and
mutual exclusion. The counting semaphore works like the binary semaphore
except that it keeps track of the number of times a semaphore is given. Every time
a semaphore is given, the count is incremented; every time a semaphore is taken,
the count is decremented. When the count reaches zero, a task that tries to take the
semaphore is blocked. As with the binary semaphore, if a semaphore is given and
a task is blocked, it becomes unblocked. However, unlike the binary semaphore, if
a semaphore is given and no tasks are blocked, then the count is incremented. This
means that a semaphore that is given twice can be taken twice without blocking.
208
4 Multitasking
4.12 Semaphores
Table 4-13 shows an example time sequence of tasks taking and giving a counting
semaphore that was initialized to a count of 3.
Table 4-13
Counting Semaphore Example
Semaphore Call
Count
after Call
semCCreate( )
3
Semaphore initialized with an initial count of 3.
semTake( )
2
Semaphore taken.
semTake( )
1
Semaphore taken.
semTake( )
0
Semaphore taken.
semTake( )
0
Task blocks waiting for semaphore to be available.
semGive( )
0
Task waiting is given semaphore.
semGive( )
1
No task waiting for semaphore; count incremented.
4
Resulting Behavior
Counting semaphores are useful for guarding multiple copies of resources. For
example, the use of five tape drives might be coordinated using a counting
semaphore with an initial count of 5, or a ring buffer with 256 entries might be
implemented using a counting semaphore with an initial count of 256. The initial
count is specified as an argument to the semCCreate( ) routine.
4.12.5 Read/Write Semaphores
Read/write semaphores provide enhanced performance for applications that can
effectively make use of differentiation between read access to a resource, and write
access to a resource. A read/write semaphore can be taken in either read mode or
write mode. They are particularly suited to SMP systems (for information about
the SMP configuration of VxWorks, see 15. VxWorks SMP).
A task holding a read/write semaphore in write mode has exclusive access to a
resource. On the other hand, a task holding a read/write semaphore in read mode
does not have exclusive access. More than one task can take a read/write
semaphore in read mode, and gain access to the same resource.
Because it is exclusive, write-mode permits only serial access to a resource, while
while read-mode allows shared or concurrent access. In a multiprocessor system,
more than one task (running in different CPUs) can have read-mode access to a
209
VxWorks
Kernel Programmer's Guide, 6.6
resource in a truly concurrent manner. In a uniprocessor system, however, access
is shared but the concurrency is virtual. More than one task can have read-mode
access to a resource at the same time, but since the tasks do not run simultaneously,
access is effectively multiplexed.
All tasks that hold a read/write semaphore in read mode must give it up before
any task can take it in write mode.
Specification of Read or Write Mode
A read/write semaphore differs from other types of semaphore in that the access
mode must be specified when the semaphore is taken. The mode determines
whether the access is exclusive (write mode), or if concurrent access is allowed
(read mode). Different APIs correspond to the different modes of access, as
follows:
■
■
semRTake( ) for read (exclusive) mode
semWTake( ) for write (concurrent) mode
You can also use semTake( ) on a read/write semaphore, but the behavior is the
same as semWTake( ). And you can use semGive( ) on a read/write semaphore as
long as the task owns it is in the same mode.
For more information about read/write semaphore APIs, see Table 4-12 and the
VxWorks API references.
When a task takes a read/write semaphore in write mode, the behavior is identical
to that of a mutex semaphore. The task owns the semaphore exclusively. An
attempt to give a semaphore held by one task in this mode by task results in a
return value of ERROR.
When a task takes a read/write semaphore in read mode, the behavior is different
from other semaphores. It does not provide exclusive access to a resource (does not
protect critical sections), and the semaphore may be concurrently held in read
mode by more than one task.
The maximum number of tasks that can take a read/write semaphore in read
mode can be specified when the semaphore is created with the create routine call.
The system maximum for all read/write semaphores can also be set with
SEM_RW_MAX_CONCURRENT_READERS component parameter. By default it is
set to 32.
If the number of tasks is not specified when the create routine is called, the system
default is used.
210
4 Multitasking
4.12 Semaphores
Read/write semaphores can be taken recursively in both read and write mode.
Optionally, priority inheritance and deletion safety are available for each mode.
Precedence for Write Access Operations
4
When a read/write semaphore becomes available, precedence is given to pended
tasks that require write access, regardless of their task priority relative to pended
tasks that require read access. That is, the highest priority task attempting a
semWTake( ) operation gets the semaphore, even if there are higher priority tasks
attempting a semRTake( ). Precedence for write access helps to ensure that the
protected resource is kept current because there is no delay due to read operations
occurring before a pending write operation can take place.
Note, however, that all read-mode takes must be given before a read/write
semaphore can be taken in write mode.
Read/Write Semaphores and System Performance
The performance of systems that implement read/write semaphores for their
intended use should be enhanced, particularly so in SMP systems. However, due
to the additional bookkeeping overhead involved in tracking multiple read-mode
owners, performance is likely to be adversely affected in those cases where the
feature does fit a clear design goal. In particular, interrupt latency in a
uniprocessor system and kernel latency in a multiprocessor system may be
adversely affected.
4.12.6 Special Semaphore Options
The uniform VxWorks semaphore interface includes three special options:
timeouts, queues, and use with VxWorks events. These options are not available
for either the read/write semaphores described in 4.12.5 Read/Write Semaphores,
p.209, or the POSIX-compliant semaphores described in 5.13 POSIX Semaphores,
p.289.
211
VxWorks
Kernel Programmer's Guide, 6.6
Timeouts
As an alternative to blocking until a semaphore becomes available, semaphore
take operations can be restricted to a specified period of time. If the semaphore is
not taken within that period, the take operation fails.
This behavior is controlled by a parameter to semTake( ) and the take routines for
read/write semaphores that specifies the amount of time in ticks that the task is
willing to wait in the pended state. If the task succeeds in taking the semaphore
within the allotted time, the take routine returns OK. The errno set when a take
routine returns ERROR due to timing out before successfully taking the semaphore
depends upon the timeout value passed.
A semTake( ) with NO_WAIT (0), which means do not wait at all, sets errno to
S_objLib_OBJ_UNAVAILABLE. A semTake( ) with a positive timeout value returns
S_objLib_OBJ_TIMEOUT. A timeout value of WAIT_FOREVER (-1) means wait
indefinitely.
Queues
VxWorks semaphores include the ability to select the queuing mechanism
employed for tasks blocked on a semaphore. They can be queued based on either
of two criteria: first-in first-out (FIFO) order, or priority order; see Figure 4-12.
212
4 Multitasking
4.13 Message Queues
Figure 4-12
Task Queue Types
FIFO QUEUE
PRIORITY QUEUE
4
TCB
TCB
TCB
TCB
200
120
TCB
TCB
90
100
80
TCB
110
140
TCB
priority
110
Priority ordering better preserves the intended priority structure of the system at
the expense of some overhead in take operations because of sorting the tasks by
priority. A FIFO queue requires no priority sorting overhead and leads to
constant-time performance. The selection of queue type is specified during
semaphore creation with the semaphore creation routine. Semaphores using the
priority inheritance option (SEM_INVERSION_SAFE) must select priority-order
queuing.
4.12.7 Semaphores and VxWorks Events
Semaphores can send VxWorks events to a specified task when they becomes free.
For more information, see 4.15 VxWorks Events, p.219.
4.13 Message Queues
Modern real-time applications are constructed as a set of independent but
cooperating tasks. While semaphores provide a high-speed mechanism for the
synchronization and interlocking of tasks, often a higher-level mechanism is
213
VxWorks
Kernel Programmer's Guide, 6.6
necessary to allow cooperating tasks to communicate with each other. In VxWorks,
the primary intertask communication mechanism within a single CPU is message
queues.
For information about socket-based message communication across memory
spaces (kernel and processes), and between multiple nodes, see 4.16 Message
Channels, p.226.
Message queues allow a variable number of messages, each of variable length, to
be queued. Tasks and ISRs can send messages to a message queue, and tasks can
receive messages from a message queue.
Figure 4-13
Full Duplex Communication Using Message Queues
message queue 1
message
task 1
task 2
message
message queue 2
Multiple tasks can send to and receive from the same message queue. Full-duplex
communication between two tasks generally requires two message queues, one for
each direction; see Figure 4-13.
VxWorks message queues can be created as private objects, which accessible only
within the memory space in which they were created (process or kernel); or as
public objects, which accessible throughout the system. For more information, see
4.9 Public and Private Objects, p.194.
There are two message-queue subroutine libraries in VxWorks. The first of these,
msgQLib, provides VxWorks message queues, designed expressly for VxWorks;
the second, mqPxLib, is compliant with the POSIX standard (1003.1b) for real-time
extensions. See 5.13.1 Comparison of POSIX and VxWorks Semaphores, p.290 for a
discussion of the differences between the two message-queue designs.
214
4 Multitasking
4.13 Message Queues
4.13.1 VxWorks Message Queue Routines
VxWorks message queues are created, used, and deleted with the routines shown
in Table 4-14. This library provides messages that are queued in FIFO order, with
a single exception: there are two priority levels, and messages marked as high
priority are attached to the head of the queue.
Table 4-14
VxWorks Message Queue Control
Call
Description
msgQCreate( )
Allocates and initializes a message queue.
msgQDelete( )
Terminates and frees a message queue.
msgQSend( )
Sends a message to a message queue.
msgQReceive( )
Receives a message from a message queue.
A message queue is created with msgQCreate( ). Its parameters specify the
maximum number of messages that can be queued in the message queue and the
maximum length in bytes of each message. Enough buffer space is allocated for the
specified number and length of messages.
A task or ISR sends a message to a message queue with msgQSend( ). If no tasks
are waiting for messages on that queue, the message is added to the queue’s buffer
of messages. If any tasks are already waiting for a message from that message
queue, the message is immediately delivered to the first waiting task.
A task receives a message from a message queue with msgQReceive( ). If
messages are already available in the message queue’s buffer, the first message is
immediately dequeued and returned to the caller. If no messages are available,
then the calling task blocks and is added to a queue of tasks waiting for messages.
This queue of waiting tasks can be ordered either by task priority or FIFO, as
specified in an option parameter when the queue is created.
Timeouts
Both msgQSend( ) and msgQReceive( ) take timeout parameters. When sending a
message, the timeout specifies how many ticks to wait for buffer space to become
available, if no space is available to queue the message. When receiving a message,
the timeout specifies how many ticks to wait for a message to become available, if
no message is immediately available. As with semaphores, the value of the timeout
parameter can have the special values of NO_WAIT (0), meaning always return
immediately, or WAIT_FOREVER (-1), meaning never time out the routine.
215
4
VxWorks
Kernel Programmer's Guide, 6.6
Urgent Messages
The msgQSend( ) function allows specification of the priority of the message as
either normal (MSG_PRI_NORMAL) or urgent (MSG_PRI_URGENT). Normal
priority messages are added to the tail of the list of queued messages, while urgent
priority messages are added to the head of the list.
Example 4-4
VxWorks Message Queues
/* In this example, task t1 creates the message queue and sends a message
* to task t2. Task t2 receives the message from the queue and simply
* displays the message.
*/
/* includes */
#include
#include
/* defines */
#define MAX_MSGS (10)
#define MAX_MSG_LEN (100)
MSG_Q_ID myMsgQId;
task2 (void)
{
char msgBuf[MAX_MSG_LEN];
/* get message from queue; if necessary wait until msg is available */
if (msgQReceive(myMsgQId, msgBuf, MAX_MSG_LEN, WAIT_FOREVER) == ERROR)
return (ERROR);
/* display message */
printf ("Message from task 1:\n%s\n", msgBuf);
}
#define MESSAGE "Greetings from Task 1"
task1 (void)
{
/* create message queue */
if ((myMsgQId = msgQCreate (MAX_MSGS, MAX_MSG_LEN, MSG_Q_PRIORITY))
== NULL)
return (ERROR);
/* send a normal priority message, blocking if queue is full */
if (msgQSend (myMsgQId, MESSAGE, sizeof (MESSAGE), WAIT_FOREVER,
MSG_PRI_NORMAL) == ERROR)
return (ERROR);
}
216
4 Multitasking
4.13 Message Queues
Queuing
VxWorks message queues include the ability to select the queuing mechanism
employed for tasks blocked on a message queue. The MSG_Q_FIFO and
MSG_Q_PRIORITY options are provided to specify (to the msgQCreate( ) and
msgQOpen( ) routines) the queuing mechanism that should be used for tasks that
pend on msgQSend( ) and msgQReceive( ).
4.13.2 Displaying Message Queue Attributes
The VxWorks show( ) command produces a display of the key message queue
attributes, for either kind of message queue. For example, if myMsgQId is a
VxWorks message queue, the output is sent to the standard output device, and
looks like the following from the shell (using the C interpreter):
-> show myMsgQId
Message Queue Id
Task Queuing
Message Byte Len
Messages Max
Messages Queued
Receivers Blocked
Send timeouts
Receive timeouts
:
:
:
:
:
:
:
0x3adaf0
FIFO
4
30
14
0
0
: 0
4.13.3 Servers and Clients with Message Queues
Real-time systems are often structured using a client-server model of tasks. In this
model, server tasks accept requests from client tasks to perform some service, and
usually return a reply. The requests and replies are usually made in the form of
intertask messages. In VxWorks, message queues or pipes (see 4.14 Pipes, p.218)
are a natural way to implement this functionality.
For example, client-server communications might be implemented as shown in
Figure 4-14. Each server task creates a message queue to receive request messages
from clients. Each client task creates a message queue to receive reply messages
from servers. Each request message includes a field containing the msgQId of the
client’s reply message queue. A server task’s main loop consists of reading request
messages from its request message queue, performing the request, and sending a
reply to the client’s reply message queue.
217
4
VxWorks
Kernel Programmer's Guide, 6.6
Figure 4-14
Client-Server Communications Using Message Queues
reply queue 1
message
client 1
request queue
server task
message
client 2
reply queue 2
message
The same architecture can be achieved with pipes instead of message queues, or by
other means that are tailored to the needs of the particular application.
4.13.4 Message Queues and VxWorks Events
Message queues can send VxWorks events to a specified task when a message
arrives on the queue and no task is waiting on it. For more information, see
4.15 VxWorks Events, p.219.
4.14 Pipes
Pipes provide an alternative interface to the message queue facility that goes
through the VxWorks I/O system. Pipes are virtual I/O devices managed by the
218
4 Multitasking
4.15 VxWorks Events
driver pipeDrv. The routine pipeDevCreate( ) creates a pipe device and the
underlying message queue associated with that pipe. The call specifies the name
of the created pipe, the maximum number of messages that can be queued to it,
and the maximum length of each message:
status = pipeDevCreate ("/pipe/name", max_msgs, max_length);
4
The created pipe is a normally named I/O device. Tasks can use the standard I/O
routines to open, read, and write pipes, and invoke ioctl routines. As they do with
other I/O devices, tasks block when they read from an empty pipe until data is
available, and block when they write to a full pipe until there is space available.
Like message queues, ISRs can write to a pipe, but cannot read from a pipe.
As I/O devices, pipes provide one important feature that message queues
cannot—the ability to be used with select( ). This routine allows a task to wait for
data to be available on any of a set of I/O devices. The select( ) routine also works
with other asynchronous I/O devices including network sockets and serial
devices. Thus, by using select( ), a task can wait for data on a combination of
several pipes, sockets, and serial devices; see 7.4.9 Pending on Multiple File
Descriptors with select( ), p.374.
Pipes allow you to implement a client-server model of intertask communications;
see 4.13.3 Servers and Clients with Message Queues, p.217.
4.15 VxWorks Events
VxWorks events provide a means of communication and synchronization between
tasks and other tasks, interrupt service routines (ISRs) and tasks, semaphores and
tasks, and message queues and tasks.1
Events can be used as a lighter-weight alternative to binary semaphores for
task-to-task and ISR-to-task synchronization (because no object must be created).
They can also be used to notify a task that a semaphore has become available, or
that a message has arrived on a message queue.
The events facility provides a mechanism for coordinating the activity of a task
using up to thirty-two events that can be sent to it by other tasks, ISRs, semaphores,
1. VxWorks events are based on pSOS operating system events. VxWorks introduced functionality similar to pSOS events (but with enhancements) with the VxWorks 5.5 release.
219
VxWorks
Kernel Programmer's Guide, 6.6
and message queues. A task can wait on multiple events from multiple sources.
Events thereby provide a means for coordination of complex matrix of activity
without allocation of additional system resources.
Each task has 32 event flags, bit-wise encoded in a 32-bit word (bits 25 to 32 are
reserved for Wind River use). These flags are stored in the task’s event register. Note
that an event flag itself has no intrinsic meaning. The significance of each of the 32
event flags depends entirely on how any given task is coded to respond to their
being set. There is no mechanism for recording how many times any given event
has been received by a task. Once a flag has been set, its being set again by the same
or a different sender is essentially an invisible operation.
Events are similar to signals in that they are sent to a task asynchronously; but
differ in that receipt is synchronous. That is, the receiving task must call a routine
to receive at will, and can choose to pend while waiting for events to arrive. Unlike
signals, therefore, events do not require a handler.
For a code example of how events can be used, see the eventLib API reference.
NOTE: VxWorks events, which are also simply referred to as events in this section,
should not be confused with System Viewer events.
Configuring VxWorks for Events
To provide events facilities, VxWorks must be configured with the
INCLUDE_VXEVENTS component.
4.15.1 Preparing a Task to Receive Events
A task can pend on one or more events, or simply check on which events have been
received, with a call to eventReceive( ). The routine specifies which events to wait
for, and provides options for waiting for one or all of those events. It also provides
various options for how to manage unsolicited events.
In order for a task to receive events from a semaphore or a message queue,
however, it must first register with the specific object, using semEvStart( ) for a
semaphore or msgQEvStart( ) for a message queue. Only one task can be
registered with any given semaphore or message queue at a time.
The semEvStart( ) routine identifies the semaphore and the events that it should
send to the task when the semaphore is free. It also provides a set of options to
specify whether the events are sent only the first time the semaphore is free, or
each time; whether to send events if the semaphore is free at the time of
220
4 Multitasking
4.15 VxWorks Events
registration; and whether a subsequent semEvStart( ) call from another task is
allowed to take effect (and to unregister the previously registered task).
Once a task has registered with a semaphore, every time the semaphore is released
with semGive( ), and as long as no other tasks are pending on it, the semaphore
sends events to the registered task.
To request that the semaphore stop sending events to it, the registered task calls
semEvStop( ).
Registration with a message queue is similar to registration with a semaphore. The
msgQEvStart( ) routine identifies the message queue and the events that it should
send to the task when a message arrives and no tasks are pending on it. It provides
a set of options to specify whether the events are sent only the first time a message
is available, or each time; whether a subsequent call to msgQEvStart( ) from
another task is allowed to take effect (and to unregister the previously registered
task).
Once a task has registered with a message queue, every time the message queue
receives a message and there are no tasks pending on it, the message queue sends
events to the registered task.
To request that the message queue stop sending events to it, the registered task
calls msgQEvStop( ).
4.15.2 Sending Events to a Task
Tasks and ISRs can send specific events to a task using eventSend( ), whether or
not the receiving task is prepared to make use of them.
Semaphores and message queues send events automatically to tasks that have
registered for notification with semEvStart( ) or msgQEvStart( ), respectively.
These objects send events when they are free. The conditions under which objects
are free are as follows:
Mutex Semaphore
A mutex semaphore is considered free when it no longer has an owner and no
task is pending on it. For example, following a call to semGive( ), the
semaphore will not send events if another task is pending on a semTake( ) for
the same semaphore.
Binary Semaphore
A binary semaphore is considered free when no task owns it and no task is
waiting for it.
221
4
VxWorks
Kernel Programmer's Guide, 6.6
Counting Semaphore
A counting semaphore is considered free when its count is nonzero and no
task is pending on it. Events cannot, therefore, be used as a mechanism to
compute the number of times a semaphore is released or given.
Message Queue
A message queue is considered free when a message is present in the queue
and no task is pending for the arrival of a message in that queue. Events
cannot, therefore, be used as a mechanism to compute the number of messages
sent to a message queue.
Note that just because an object has been released does not mean that it is free. For
example, if a semaphore is given, it is released; but it is not free if another task is
waiting for it at the time it is released. When two or more tasks are constantly
exchanging ownership of an object, it is therefore possible that the object never
becomes free, and never sends events.
Also note that when a semaphore or message queue sends events to a task to
indicate that it is free, it does not mean that the object is in any way reserved for the
task. A task waiting for events from an object unpends when the resource becomes
free, but the object may be taken in the interval between notification and
unpending. The object could be taken by a higher priority task if the task receiving
the event was pended in eventReceive( ). Or a lower priority task might steal the
object: if the task receiving the event was pended in some routine other than
eventReceive( ), a low priority task could execute and (for example) perform a
semTake( ) after the event is sent, but before the receiving task unpends from the
blocking call. There is, therefore, no guarantee that the resource will still be
available when the task subsequently attempts to take ownership of it.
!
WARNING: Because events cannot be reserved for an application in any way, care
should be taken to ensure that events are used uniquely and unambiguously. Note
that events 25 to 32 (VXEV25 to VXEV32) are reserved for Wind River’s use, and
should not be used by customers. Third parties should be sure to document their
use of events so that their customers do not use the same ones for their
applications.
Events and Object Deletion
If a semaphore or message queue is deleted while a task is waiting for events from
it, the task is automatically unpended by the semDelete( ) or msgQDelete( )
implementation. This prevents the task from pending indefinitely while waiting
for events from an object that has been deleted. The pending task then returns to
222
4 Multitasking
4.15 VxWorks Events
the ready state (just as if it were pending on the semaphore itself) and receives an
ERROR return value from the eventReceive( ) call that caused it to pend initially.
If, however, the object is deleted between a tasks’ registration call and its
eventReceive( ) call, the task pends anyway. For example, if a semaphore is
deleted while the task is between the semEvStart( ) and eventReceive( ) calls, the
task pends in eventReceive( ), but the event is never sent. It is important, therefore,
to use a timeout other than WAIT_FOREVER when object deletion is expected.
Events and Task Deletion
If a task is deleted before a semaphore or message queue sends events to it, the
events can still be sent, but are obviously not received. By default, VxWorks
handles this event-delivery failure silently.
It can, however, be useful for an application that created an object to be informed
when events were not received by the (now absent) task that registered for them.
In this case, semaphores and message queues can be created with an option that
causes an error to be returned if event delivery fails (the
SEM_EVENTSEND_ERROR_NOTIFY and MSG_Q_EVENTSEND_ERROR_NOTIFY
options, respectively). The semGive( ) or msgQSend( ) call then returns ERROR
when the object becomes free.
The error does not mean the semaphore was not given or that the message was not
properly delivered. It simply means the resource could not send events to the
registered task. Note that a failure to send a message or give a semaphore takes
precedence over an events failure.
4.15.3 Accessing Event Flags
When events are sent to a task, they are stored in the task’s events register (see
4.15.5 Task Events Register, p.224), which is not directly accessible to the task itself.
When the events specified with an eventReceive( ) call have been received and the
task unpends, the contents of the events register is copied to a variable that is
accessible to the task.
When eventReceive( ) is used with the EVENTS_WAIT_ANY option—which
means that the task unpends for the first of any of the specified events that it
receives—the contents of the events variable can be checked to determine which
event caused the task to unpend.
The eventReceive( ) routine also provides an option that allows for checking
which events have been received prior to the full set being received.
223
4
VxWorks
Kernel Programmer's Guide, 6.6
4.15.4 Events Routines
The routines used for working with events are listed in Table 4-15.
Table 4-15
Events Routines
Routine
Description
eventSend( )
Sends specified events to a task.
eventReceive( )
Pends a task until the specified events have been received. Can
also be used to check what events have been received in the
interim.
eventClear( )
Clears the calling task’s event register.
semEvStart( )
Registers a task to be notified of semaphore availability.
semEvStop( )
Unregisters a task that had previously registered for
notification of semaphore availability.
msgQEvStart( )
Registers a task to be notified of message arrival on a message
queue when no recipients are pending.
msgQEvStop( )
Unregisters a task that had previously registered for
notification of message arrival on a message queue.
For more information about these routines, see the VxWorks API references for
eventLib, semEvLib, and msgQEvLib.
4.15.5 Task Events Register
Each task has its own task events register. The task events register is a 32-bit field
used to store the events that the task receives from other tasks (or itself), ISRs,
semaphores, and message queues.
Events 25 to 32 (VXEV25 or 0x01000000 to VXEV32 or 0x80000000) are reserved for
Wind River use only, and should not be used by customers.
As noted above (4.15.3 Accessing Event Flags, p.223), a task cannot access the
contents of its events registry directly.
Table 4-16 describes the routines that affect the contents of the events register.
224
4 Multitasking
4.15 VxWorks Events
Table 4-16
Routines That Modify the Task Events Register
Routine
Effect on the Task Events Register
eventReceive( ) Clears or leaves the contents of the task’s events register intact,
depending on the options selected.
eventClear( )
Clears the contents of the task’s events register.
eventSend( )
Writes events to a tasks’s events register.
semGive( )
Writes events to the tasks’s events register, if the task is
registered with the semaphore.
msgQSend( )
Writes events to a task’s events register, if the task is registered
with the message queue.
4.15.6 Show Routines and Events
For the purpose of debugging systems that make use of events, the taskShow,
semShow, and msgQShow libraries display event information.
The taskShow library displays the following information:
■
■
■
the contents of the event register
the desired events
the options specified when eventReceive( ) was called
The semShow and msgQShow libraries display the following information:
■
■
■
the task registered to receive events
the events the resource is meant to send to that task
the options passed to semEvStart( ) or msgQEvStart( )
225
4
VxWorks
Kernel Programmer's Guide, 6.6
4.16 Message Channels
Message channels are a socket-based facility that provides for inter-task
communication within a memory boundary, between memory boundaries (kernel
and processes), between nodes (processors) in a multi-node cluster, and between
between multiple clusters.
In addition to providing a superior alternative to TCP for multi-node
intercommunication, message channels provide a useful alternative to message
queues for exchanging data between two tasks on a single node. Message channels
can be used in both kernel and user (RTP) space.
For a comparison of message channels and message queues, see 18.8 Comparison of
Message Channels and Message Queues, p.806. For detailed information about
message channels, see 18. Message Channels.
4.17 Network Communication
To communicate with a peer on a remote networked system, you can use an
Internet domain socket or RPC. For information on working with Internet domain
sockets or RPC under VxWorks, see the Wind River Network Stack for VxWorks 6
Programmer’s Guide. For information about TIPC networking, see Wind River TIPC
for VxWorks 6 Programmer's Guide.
4.18 Signals
Signals are an operating system facility designed for handling exceptional
conditions and asynchronously altering the flow of control. In many respects
signals are the software equivalent to hardware interrupts. Signals generated by
the operating system include those produced in response to bus errors and floating
point exceptions. The signal facility also provides APIs that can be used to generate
and manage signals programmatically.
In applications, signals are most appropriate for error and exception handling, and
not for a general-purpose inter-task communication. Common uses include using
226
4 Multitasking
4.18 Signals
signals to kill processes and tasks, to send signal events when a timer has fired or
message has arrived at a message queue, and so on.
In accordance with POSIX, VxWorks supports 63 signals, each of which has a
unique number and default action (defined in signal.h). The value 0 is reserved for
use as the NULL signal.
Signals can be raised (sent) from tasks to tasks or to processes. Signals can be either
caught (received) or ignored by the receiving task or process. Whether signals are
caught or ignored generally depends on the setting of a signal mask. In the kernel,
signal masks are specific to tasks, and if no task is set up to receive a specific signal,
it is ignored. In user space, signal masks are specific to processes; and some signals,
such as SIGKILL and SIGSTOP, cannot be ignored.
To manage responses to signals, you can create and register signal handling
routines that allow a task to respond to a specific signal in whatever way is useful
for your application.
A kernel task or interrupt service routine (ISR) can raise a signal for a specific task
or process. In the kernel, signal generation and delivery runs in the context of the
task or ISR that generates the signal. In accordance with the POSIX standard, a
signal sent to a process is handled by the first available task that has been set up to
handle the signal in the process.
Each kernel task has a signal mask associated with it. The signal mask determines
which signals the task accepts. By default, the signal mask is initialized with all
signals unblocked (there is no inheritance of mask settings in the kernel). The mask
can be changed with sigprocmask( ).
Signal handlers in the kernel can be registered for a specific task. A signal handler
executes in the receiving task’s context and makes use of that task’s execution
stack. The signal handler is invoked even if the task is blocked (suspended or
pended).
VxWorks provides a software signal facility that includes POSIX routines, UNIX
BSD-compatible routines, and native VxWorks routines. The POSIX-compliant
signal interfaces include both the basic signaling interface specified in the POSIX
standard 1003.1, and the queued-signals extension from POSIX 1003.1b.
Additional, non-POSIX APIs provide support for signals between kernel and user
applications. These non-POSIX APIs are: taskSigqueue( ), rtpSigqueue( ),
rtpTaskSigqueue( ), taskKill( ), rtpKill( ), rtpTaskKill( ), and taskRaise( ).
In the VxWorks kernel—for backward compatibility with prior versions of
VxWorks—the POSIX API that would take a process identifier as one of their
parameters, take a task identifier instead.
227
4
VxWorks
Kernel Programmer's Guide, 6.6
NOTE: Wind River recommends that you do not use both POSIX APIs and
VxWorks APIs in the same application. Doing so may make a POSIX application
non-conformant.
NOTE: POSIX signals are handled differently in the kernel and in real-time
processes. In the kernel the target of a signal is always a task; but in user space, the
target of a signal may be either a specific task or an entire process.
NOTE: The VxWorks implementation of sigLib does not impose any special
restrictions on operations on SIGKILL, SIGCONT, and SIGSTOP signals such as
those imposed by UNIX. For example, the UNIX implementation of signal( )
cannot be called on SIGKILL and SIGSTOP.
For information about using signals in processes, see VxWorks Application
Programmer’s Guide: Multitasking.
In addition to signals, VxWorks also provides another type of event notification
with the VxWorks events facility. While signal events are completely
asynchronous, VxWorks events are sent asynchronously, but received
synchronously, and do not require a signal handler. For more information, see
4.15 VxWorks Events, p.219.
4.18.1 Configuring VxWorks for Signals
By default, VxWorks includes the basic signal facility component
INCLUDE_SIGNALS. This component automatically initializes signals with
sigInit( ).
To use the signal event facility, configure VxWorks with the INCLUDE_SIGEVENT
component. Note that SIGEV_THREAD option is only supported in processes, and
that it requires that VxWorks also be configured with the
INCLUDE_SIGEVENTS_THREAD component and full POSIX thread support (the
BUNDLE_RTP_POSIX_PSE52 bundle includes everything required for this option).
To include POSIX queued signals in the system, configure VxWorks with the
INCLUDE_POSIX_SIGNALS component. This component automatically initializes
POSIX queued signals with sigqueueInit( ). The sigqueueInit( ) routine allocates
buffers for use by sigqueue( ), which requires a buffer for each currently queued
signal. A call to sigqueue( ) fails if no buffer is available.
228
4 Multitasking
4.18 Signals
The maximum number of queued signals in the kernel is set with the configuration
parameter NUM_SIGNAL_QUEUES. The default value is 16.
4.18.2 Basic Signal Routines
4
Signals are in many ways analogous to hardware interrupts. The basic signal
facility provides a set of 63 distinct signals. A signal handler binds to a particular
signal with sigvec( ) or sigaction( ) in much the same way that an ISR is connected
to an interrupt vector with intConnect( ). A signal can be asserted by calling kill( )
or sigqueue( ). This is similar to the occurrence of an interrupt. The sigprocmask( )
routine let signals be selectively inhibited. Certain signals are associated with
hardware exceptions. For example, bus errors, illegal instructions, and
floating-point exceptions raise specific signals.
For a list and description of basic POSIX and BSD signal routines provided by
VxWorks in the kernel, see Table 4-17.
Table 4-17
Basic Signal Calls
POSIX 1003.1b
Compatible
Routine
UNIX BSD
Compatible
Routine
Description
signal( )
signal( )
Specifies the handler associated with a signal.
raise( )
N/A
Sends a signal to yourself.
sigaction( )
sigvec( )
Examines or sets the signal handler for a signal.
sigsuspend( )
pause( )
Suspends a task until a signal is delivered.
sigpending( )
N/A
Retrieves a set of pending signals blocked from
delivery.
sigemptyset( ) N/A
sigfillset( )
sigaddset( )
sigdelset( )
sigismember( )
Manipulates a signal mask.
sigprocmask( ) N/A
Sets the mask of blocked signals.
sigprocmask( ) N/A
Adds to a set of blocked signals.
229
VxWorks
Kernel Programmer's Guide, 6.6
VxWorks also provides a POSIX and BSD-like kill( ) routine, which sends a signal
to a task.
VxWorks also provides additional routines that serve as aliases for POSIX
routines, such as rtpKill( ), that provide for sending signals from the kernel to
processes.
For more information about signal routines, see the VxWorks API reference for
sigLib and rtpSigLib.
4.18.3 Queued Signal Routines
The sigqueue( ) family of routines provides an alternative to the kill( ) family of
routines for sending signals. The important differences between the two are the
following:
■
The sigqueue( ) routine includes an application-specified value that is sent as
part of the signal. This value supplies whatever context is appropriate for the
signal handler. This value is of type sigval (defined in signal.h); the signal
handler finds it in the si_value field of one of its arguments, a structure
siginfo_t.
■
The sigqueue( ) routine enables the queueing of multiple signals for any task.
The kill( ) routine, by contrast, delivers only a single signal, even if multiple
signals arrive before the handler runs.
VxWorks includes signals reserved for application use, numbered consecutively
from SIGRTMIN to SIGRTMAX. The number of signals reserved is governed by the
RTSIG_MAX macro (with a value of 16), which defined in the POSIX 1003.1
standard. The signal values themselves are not specified by POSIX. For portability,
specify these signals as offsets from SIGRTMIN (for example, use SIGRTMIN+2 to
refer to the third reserved signal number). All signals delivered with sigqueue( )
are queued by numeric order, with lower-numbered signals queuing ahead of
higher-numbered signals.
POSIX 1003.1 also introduced an alternative means of receiving signals. The
routine sigwaitinfo( ) differs from sigsuspend( ) or pause( ) in that it allows your
application to respond to a signal without going through the mechanism of a
registered signal handler: when a signal is available, sigwaitinfo( ) returns the
value of that signal as a result, and does not invoke a signal handler even if one is
registered. The routine sigtimedwait( ) is similar, except that it can time out.
The basic queued signal routines are described in Table 4-18. For detailed
information on signals, see the API reference for sigLib.
230
4 Multitasking
4.18 Signals
Table 4-18
POSIX 1003.1b Queued Signal Routines
Routine
Description
sigqueue( )
Sends a queued signal to a task.
sigwaitinfo( )
Waits for a signal.
sigtimedwait( )
Waits for a signal with a timeout.
4
Additional non-POSIX VxWorks queued signal routines are described in
Table 4-19. These routines are provided for assisting in porting VxWorks 5.x kernel
applications to processes. The POSIX routines described in Table 4-18 should be
used for developing new applications that execute as real-time processes.
Note that a parallel set of non-POSIX APIs are provided for the kill( ) family of
POSIX routines—taskKill( ), rtpKill( ), and rtpTaskKill( ).
Table 4-19
Non-POSIX Queued Signal Routines
Routine
Description
taskSigqueue( )
Sends a queued signal from a task in a process to another task
in the same process, to a public task in another process, or
from a kernel task to a process task.
rtpSigqueue( )
Sends a queued signal from a kernel task to a process or from
a process to another process.
rtpTaskSigqueue( ) Sends a queued signal from a kernel task to a specified task
in a process (kernel-space only).
231
VxWorks
Kernel Programmer's Guide, 6.6
Example 4-5
Queued Signals
#include
#include
#include
#include
#ifdef _WRS_KERNEL
#include
#include
#include
#endif
typedef void (*FPTR) (int);
void sigMasterHandler
(
int
sig,
/* caught signal */
#ifdef _WRS_KERNEL
int
code,
#else
siginfo_t * pInfo,
/* signal info */
#endif
struct sigcontext *pContext /* unused */
);
/****************************************************************************
*
* main - entry point for the queued signal demo
*
* This routine acts the task entry point in the case of the demo spawned as a
* kernel task. It also can act as a RTP entry point in the case of RTP based
* demo.
*/
STATUS main (void)
{
sigset_t sig = sigmask (SIGUSR1);
union sigval sval;
struct sigaction in;
sigprocmask (SIG_UNBLOCK, &sig, NULL);
in.sa_handler = (FPTR) sigMasterHandler;
in.sa_flags = 0;
(void) sigemptyset (&in.sa_mask);
if (sigaction (SIGUSR1, &in, NULL) != OK)
{
printf ("Unable to set up handler for task (0x%x)\n", taskIdCurrent);
return (ERROR);
}
printf ("Task 0x%x installed signal handler for signal # %d.\
Ready for signal.\n", taskIdCurrent, SIGUSR1);
232
4 Multitasking
4.18 Signals
for (;;);
}
/****************************************************************************
*
* sigMasterHandler - signal handler
*
* This routine is the signal handler for the SIGUSR1 signal
*/
void sigMasterHandler
(
int
sig,
/* caught signal */
#ifdef _WRS_KERNEL
int
code,
#else
siginfo_t * pInfo ,
/* signal info */
#endif
struct sigcontext *pContext /* unused */
)
{
printf ("Task 0x%x got signal # %d signal value %d \n",
taskIdCurrent, sig,
#ifdef _WRS_KERNEL
code
#else
pInfo->si_value.sival_int
#endif
);
}
/****************************************************************************
*
* sig - helper routine to send a queued signal
*
* This routine can send a queued signal to a kernel task or RTP task or RTP.
* is the ID of the receiver entity. is the value to be sent
* along with the signal. The signal number being sent is SIGUSR1.
*/
#ifdef _WRS_KERNEL
STATUS sig
(
int id,
int val
)
{
union sigval
valueCode;
valueCode.sival_int = val;
233
4
VxWorks
Kernel Programmer's Guide, 6.6
if (TASK_ID_VERIFY (id) == OK)
{
if (IS_KERNEL_TASK (id))
{
if (sigqueue (id, SIGUSR1, valueCode) == ERROR)
{
printf ("Unable to send SIGUSR1 signal to 0xx%x, errno = 0x%x\n",
id, errnoGet());
return ERROR;
}
}
else
{
rtpTaskSigqueue ((WIND_TCB *)id, SIGUSR1, valueCode);
}
}
else if (OBJ_VERIFY ((RTP_ID)id, rtpClassId) != ERROR)
{
rtpSigqueue ((RTP_ID)id, SIGUSR1, valueCode);
}
else
{
return (ERROR);
}
return (OK);
}
#endif
234
4 Multitasking
4.18 Signals
The code provided in this example can be used to do any of the following:
■
■
■
Send a queued signal to a kernel task.
Send a queued signal to a task in a process (RTP).
Send a queued signal to a process.
The sig( ) routine provided in this code is a helper routine used to send a queued
signal.
To use the code as a kernel application, VxWorks must be configured with
BUNDLE_NET_SHELL and BUNDLE_POSIX.
To send a queued signal to a kernel task:
1.
Build the code with the VxWorks image or as a downloadable kernel module.
Boot the system. If the code is not linked to the system image, load the module;
for example, from the kernel shell:
-> ld < signal_ex.o
value = 8382064 = 0x7fe670
2.
Spawn a task with main( ) as the entry point. For example:
-> sp main
Task spawned: id = 0x7fd620, name = t1
value = 8377888 = 0x7fd620
-> Task 0x7fd620 installed signal handler for signal # 30.
signal.
sp main
3.
Ready for
Send a queued signal to the spawned kernel task. From the kernel shell use the
command:
sig kernelTaskId, signalValue
For example:
-> sig 0x7fd620, 20
value = 0 = 0x0
-> Task 0x7fd620 got signal # 30
sig 0x7fd620, 20
value = 0 = 0x0
-> Task 0x7fd620 got signal # 30
signal value 20
signal value 20
For information on using the code in a process (as an RTP application), see the
VxWorks Application Programmer’s Guide: Multitasking.
235
4
VxWorks
Kernel Programmer's Guide, 6.6
4.18.4 Signal Events
The signal event facility allows a pthread or task to receive notification that a
particular event has occurred (such as the arrival of a message at a message queue,
or the firing of a timer) by way of a signal.
The following routines can be used to register for signal notification of their
respective event activities: mq_notify( ), timer_create( ), timer_open( ),
aio_read( ), aio_write( ) and lio_listio( ).
The POSIX 1003.1-2001 standard defines three signal event notification types:
SIGEV_NONE
Indicates that no notification is required when the event occurs. This is useful
for applications that use asynchronous I/O with polling for completion.
SIGEV_SIGNAL
Indicates that a signal is generated when the event occurs.
SIGEV_THREAD
Provides for callback functions for asynchronous notifications done by a
function call within the context of a new thread. This provides a
multi-threaded process with a more natural means of notification than signals.
VxWorks supports this option in user space (processes), but not in the kernel.
The notification type is specified using the sigevent structure, which is defined in
installDir/vxworks-6.x/target/h/sigeventCommon.h. A pointer the structure is
used in the call to register for signal notification; for example, with mq_notify( ).
To use the signal event facility, configure VxWorks with the INCLUDE_SIGEVENT
component.
4.18.5 Signal Handlers
Signals are more appropriate for error and exception handling than as a
general-purpose intertask communication mechanism. And normally, signal
handlers should be treated like ISRs: no routine should be called from a signal
handler that might cause the handler to block. Because signals are asynchronous,
it is difficult to predict which resources might be unavailable when a particular
signal is raised.
To be perfectly safe, call only those routines listed in Table 4-20. Deviate from this
practice only if you are certain that your signal handler cannot create a deadlock
situation.
236
4 Multitasking
4.18 Signals
In addition, you should be particularly careful when using C++ for a signal
handler or when invoking a C++ method from a signal handler written in C or
assembly. Some of the issues involved in using C++ include the following:
■
Table 4-20
The VxWorks intConnect( ) and signal( ) routines require the address of the
function to execute when the interrupt or signal occurs, but the address of a
non-static member function cannot be used, so static member functions must
be implement.
■
Objects cannot be instantiated or deleted in signal handling code.
■
C++ code used to execute in a signal handler should restrict itself to Embedded
C++. No exceptions nor run-time type identification (RTTI) should be used.
Routines Callable by Signal Handlers
Library
Routines
bLib
All routines
errnoLib
errnoGet( ), errnoSet( )
eventLib
eventSend( )
fppArchLib
fppSave( ), fppRestore( )
intLib
intContext( ), intCount( ), intVecSet( ), intVecGet( )
intArchLib
intLock( ), intUnlock( )
logLib
logMsg( )
lstLib
All routines except lstFree( )
mathALib
All routines, if fppSave( )/fppRestore( ) are used
msgQLib
msgQSend( )
rngLib
All routines except rngCreate( ) and rngDelete( )
semLib
semGive( ) except mutual-exclusion semaphores, semFlush( )
sigLib
kill( )
taskLib
taskSuspend( ), taskResume( ), taskPrioritySet( ),
taskPriorityGet( ), taskIdVerify( ), taskIdDefault( ),
taskIsReady( ), taskIsSuspended( ), taskTcb( )
237
4
VxWorks
Kernel Programmer's Guide, 6.6
Table 4-20
Routines Callable by Signal Handlers (cont’d)
Library
Routines
tickLib
tickAnnounce( ), tickSet( ), tickGet( )
tyLib
tyIRd( ), tyITx( )
vxLib
vxTas( ), vxMemProbe( )
wdLib
wdStart( ), wdCancel( )
Most signals are delivered asynchronously to the execution of a program.
Therefore programs must be written to account for the unexpected occurrence of
signals, and handle them gracefully. Unlike ISR's, signal handlers execute in the
context of the interrupted task.
VxWorks does not distinguish between normal task execution and a signal
context, as it distinguishes between a task context and an ISR. Therefore the system
has no way of distinguishing between a task execution context and a task
executing a signal handler. To the system, they are the same.
When you write signal handlers make sure that they:
■
Release resources prior to exiting:
–
–
–
Free any allocated memory.
Close any open files.
Release any mutual exclusion resources such as semaphores.
■
Leave any modified data structures in a sane state.
■
Notify the kernel with an appropriate error return value.
Mutual exclusion between signal handlers and tasks must be managed with care.
In general, users should avoid the following activity in signal handlers:
■
Taking mutual exclusion (such as semaphores) resources that can also be taken
by any other element of the application code. This can lead to deadlock.
■
Modifying any shared data memory that may have been in the process of
modification by any other element of the application code when the signal was
delivered. This compromises mutual exclusion and leads to data corruption.
■
Using longjmp( ) to change the flow of task execution. If longjmp( ) is used in
a signal handler to re-initialize a running task, you must ensure that the signal
is not sent to the task while the task is holding a critical resource (such as a
kernel mutex). For example, if a signal is sent to a task that is executing
238
4 Multitasking
4.19 Watchdog Timers
malloc( ), the signal handler that calls longjmp( ) could leave the kernel in an
inconsistent state.
These scenarios are very difficult to debug, and should be avoided. One safe way
to synchronize other elements of the application code and a signal handler is to set
up dedicated flags and data structures that are set from signal handlers and read
from the other elements. This ensures a consistency in usage of the data structure.
In addition, the other elements of the application code must check for the
occurrence of signals at any time by periodically checking to see if the
synchronizing data structure or flag has been modified in the background by a
signal handler, and then acting accordingly. The use of the volatile keyword is
useful for memory locations that are accessed from both a signal handler and other
elements of the application.
Taking a mutex semaphore in a signal handler is an especially bad idea. Mutex
semaphores can be taken recursively. A signal handler can therefore easily
re-acquire a mutex that was taken by any other element of the application. Since
the signal handler is an asynchronously executing entity, it has thereby broken the
mutual exclusion that the mutex was supposed to provide.
Taking a binary semaphore in a signal handler is an equally bad idea. If any other
element has already taken it, the signal handler will cause the task to block on
itself. This is a deadlock from which no recovery is possible. Counting semaphores,
if available, suffer from the same issue as mutexes, and if unavailable, are
equivalent to the binary semaphore situation that causes an unrecoverable
deadlock.
On a general note, the signal facility should be used only for notifying/handling
exceptional or error conditions. Usage of signals as a general purpose IPC
mechanism or in the data flow path of an application can cause some of the pitfalls
described above.
4.19 Watchdog Timers
VxWorks includes a watchdog-timer mechanism that allows any C function to be
connected to a specified time delay. Watchdog timers are maintained as part of the
system clock ISR. Functions invoked by watchdog timers execute as interrupt
service code at the interrupt level of the system clock. Restrictions on ISRs apply to
239
4
VxWorks
Kernel Programmer's Guide, 6.6
routines connected to watchdog timers. The functions in Table 4-21 are provided
by the wdLib library.
Table 4-21
Watchdog Timer Calls
Call
Description
wdCreate( )
Allocates and initializes a watchdog timer.
wdDelete( )
Terminates and deallocates a watchdog timer.
wdStart( )
Starts a watchdog timer.
wdCancel( )
Cancels a currently counting watchdog timer.
A watchdog timer is first created by calling wdCreate( ). Then the timer can be
started by calling wdStart( ), which takes as arguments the number of ticks to
delay, the C function to call, and an argument to be passed to that function. After
the specified number of ticks have elapsed, the function is called with the specified
argument. The watchdog timer can be canceled any time before the delay has
elapsed by calling wdCancel( ).
Example 4-6
Watchdog Timers
/* Creates a watchdog timer and sets it to go off in 3 seconds.*/
/* includes */
#include
#include
#include
/* defines */
#define SECONDS (3)
WDOG_ID myWatchDogId;
task (void)
{
/* Create watchdog */
if ((myWatchDogId = wdCreate( )) == NULL)
return (ERROR);
/* Set timer to go off in SECONDS - printing a message to stdout */
if (wdStart (myWatchDogId, sysClkRateGet( ) * SECONDS, logMsg,
"Watchdog timer just expired\n") == ERROR)
return (ERROR);
/* ... */
}
For information about POSIX timers, see 5.6 POSIX Clocks and Timers, p.259.
240
4 Multitasking
4.20 Interrupt Service Routines
4.20 Interrupt Service Routines
Hardware interrupt handling is of key significance in real-time systems, because it
is usually through interrupts that the system is informed of external events. For the
fastest possible response to interrupts, VxWorks runs interrupt service routines
(ISRs) in a special context outside of any task’s context. Thus, interrupt handling
involves no task context switch. Table 4-22 lists the interrupt routines provided in
intLib and intArchLib.
Table 4-22
Interrupt Routines
Call
Description
intConnect( )
Connects a C routine to an interrupt vector.
intContext( )
Returns TRUE if called from interrupt level.
intCount( )
Gets the current interrupt nesting depth.
intLevelSet( )
Sets the processor interrupt mask level.
intLock( )
Disables interrupts.
intUnlock( )
Re-enables interrupts.
intVecBaseSet( )
Sets the vector base address.
intVecBaseGet( ) Gets the vector base address.
intVecSet( )
Sets an exception vector.
intVecGet( )
Gets an exception vector.
For information about interrupt locks and latency, see 4.11.1 Interrupt Locks and
Latency, p.197.
NOTE: The intLock( ) and intUnlock( ) routines are provided for the UP
configuration of VxWorks, but not the SMP configuration. Several alternative are
available for SMP systems, including the ISR-callable spinlock, which default to
intLock( ) and intUnlock( ) behavior in a UP system. For more information, see
15.6.1 ISR-Callable Spinlocks, p.682 and 15.15 Migrating Code to VxWorks SMP,
p.702.
241
4
VxWorks
Kernel Programmer's Guide, 6.6
4.20.1 Connecting Routines to Interrupts
You can use system hardware interrupts other than those used by VxWorks.
VxWorks provides the routine intConnect( ), which allows C functions to be
connected to any interrupt. The arguments to this routine are the byte offset of the
interrupt vector to connect to, the address of the C function to be connected, and
an argument to pass to the function. When an interrupt occurs with a vector
established in this way, the connected C function is called at interrupt level with
the specified argument. When the interrupt handling is finished, the connected
function returns. A routine connected to an interrupt in this way is called an
interrupt service routine (ISR).
Interrupts cannot actually vector directly to C functions. Instead, intConnect( )
builds a small amount of code that saves the necessary registers, sets up a stack
entry (either on a special interrupt stack, or on the current task’s stack) with the
argument to be passed, and calls the connected function. On return from the
function it restores the registers and stack, and exits the interrupt; see Figure 4-15.
Figure 4-15
Routine Built by intConnect( )
Wrapper built by intConnect( )
save registers
set up stack
invoke routine
restore registers and stack
exit
Interrupt Service Routine
myISR
(
int val;
)
(
/* deal with hardware*/
...
)
intConnect (INUM_TO_IVEC (someIntNum), myISR, someVal);
For target boards with VME backplanes, the BSP provides two standard routines
for controlling VME bus interrupts, sysIntEnable( ) and sysIntDisable( ).
4.20.2 Interrupt Stack
All ISRs use the same interrupt stack. This stack is allocated and initialized by the
system at startup according to specified configuration parameters. It must be large
enough to handle the worst possible combination of nested interrupts.
242
4 Multitasking
4.20 Interrupt Service Routines
!
CAUTION: Some architectures do not permit using a separate interrupt stack, and
ISRs use the stack of the interrupted task. With such architectures, make sure to
create tasks with enough stack space to handle the worst possible combination of
nested interrupts and the worst possible combination of ordinary nested calls. See
the VxWorks reference for your BSP to determine whether your architecture
supports a separate interrupt stack. If it does not, also see Task Stack Protection,
p.176.
Use the checkStack( ) facility during development to see how close your tasks and
ISRs have come to exhausting the available stack space.
In addition to experimenting with stack size, you can also configure and test
systems with guard zone protection for interrupt stacks (for more information, see
Interrupt Stack Protection, p.243).
Filling Interrupt Stacks
By default, interrupt (and task) stacks are filled with 0xEE. Filling stacks is useful
during development for debugging with the checkStack( ) routine. It is generally
not used in deployed systems because not filling stacks provides better
performance. You can use the VX_GLOBAL_NO_STACK_FILL configuration
parameter (when you configure VxWorks) to disable stack filling for all interrupts
(and tasks) in the system.
Interrupt Stack Protection
Systems can be configured with the INCLUDE_PROTECT_INTERRUPT_STACK
component to provide guard zone protection at the start and end of the interrupt
stack. If the size of the system becomes an issue, the component can be removed
for final testing and the deployed system.
An overrun guard zone prevents a task from going beyond the end of its
predefined stack size and ensures that the integrity of the system is not
compromised. An under-run guard zone typically prevents buffer overflows from
corrupting memory above the stack. The CPU generates an exception when a task
attempts to access any of the guard zones. The size of a stack is always rounded up
to a multiple the MMU page size when a guard zone is inserted.
243
4
VxWorks
Kernel Programmer's Guide, 6.6
The sizes of the guard zones are defined by the following configuration
parameters:
■
INTERRUPT_STACK_OVERFLOW_SIZE for interrupt stack overflow size.
■
INTERRUPT_STACK_UNDERFLOW_SIZE for interrupt stack underflow size.
The value of these parameters can be modified to increase the size of the guard
zone. The size of a guard zone is rounded up to the CPU MMU page size. The
insertion of a guard zone can be prevented by setting the parameter to zero.
4.20.3 Writing and Debugging ISRs
There are some restrictions on the routines you can call from an ISR. For example,
you cannot use routines like printf( ), malloc( ), and semTake( ) in your ISR. You
can, however, use semGive( ), logMsg( ), msgQSend( ), and bcopy( ). For more
information, see 4.20.5 Special Limitations of ISRs, p.245.
Two basic techniques for debugging an ISR are to use logMsg( ) and to have global
variables that are incremented each time though the ISR. You can then print the
globals from the shell. You can also use global variables from the shell to turn on
different condition flows through the ISR.
4.20.4 ISRs and the Kernel Work Queue
The VxWorks kernel reduces interrupt latency to a minimum by protecting
portions of its critical sections using a work deferral mechanism as opposed to
locking interrupts. Work deferral consists of storing kernel work requests
performed by ISRs that interrupt the kernel while it is in one of its critical sections.
For example, an ISR that performs a semGive( ) after having interrupted the kernel
in a critical section would cause work to be stored in the work queue. This work is
processed after the ISR returns and immediately after the kernel exits its critical
section. This process involves a static buffer, also known as work queue, used to
store work requests until they are processed. This is all internal to the VxWorks
kernel and users must never make use of the work queue. However, it is possible
for the work queue to overflow in situations where a large number of interrupts
cause work deferral without allowing the kernel to complete execution of the
critical section that was originally interrupted. These situations are uncommon
and are often symptoms of ill-behaved interrupt service routines. A work queue
overflow is also known as a work queue panic in reference to the message the
kernel displays as a result of an overflow:
244
4 Multitasking
4.20 Interrupt Service Routines
workQPanic: Kernel work queue overflow
In order to help reduce the occurrences of work queue overflows, system architects
can use the WIND_JOBS_MAX kernel configuration parameter to increase the size
of the kernel work queue. However in most cases this is simply hiding the root
cause of the overflow.
4.20.5 Special Limitations of ISRs
Many VxWorks facilities are available to ISRs, but there are some important
limitations. These limitations stem from the fact that an ISR does not run in a
regular task context and has no task control block, so all ISRs share a single stack.
For this reason, the basic restriction on ISRs is that they must not invoke routines
that might cause the caller to block. For example, they must not try to take a
semaphore, because if the semaphore is unavailable, the kernel tries to switch the
caller to the pended state. However, ISRs can give semaphores, releasing any tasks
waiting on them.
Because the memory facilities malloc( ) and free( ) take a semaphore, they cannot
be called by ISRs, and neither can routines that make calls to malloc( ) and free( ).
For example, ISRs cannot call any creation or deletion routines.
ISRs also must not perform I/O through VxWorks drivers. Although there are no
inherent restrictions in the I/O system, most device drivers require a task context
because they might block the caller to wait for the device. An important exception
is the VxWorks pipe driver, which is designed to permit writes by ISRs.
VxWorks supplies a logging facility, in which a logging task prints text messages
to the system console. This mechanism was specifically designed for ISR use, and
is the most common way to print messages from ISRs. For more information, see
the VxWorks API reference for logLib.
An ISR also must not call routines that use a floating-point coprocessor. In
VxWorks, the interrupt driver code created by intConnect( ) does not save and
restore floating-point registers; thus, ISRs must not include floating-point
instructions. If an ISR requires floating-point instructions, it must explicitly save
and restore the registers of the floating-point coprocessor using routines in
fppArchLib.
In addition, you should be particularly careful when using C++ for an ISR or when
invoking a C++ method from an ISR written in C or assembly. Some of the issues
involved in using C++ include the following:
245
4
VxWorks
Kernel Programmer's Guide, 6.6
■
The VxWorks intConnect( ) routine require the address of the function to
execute when the interrupt occurs, but the address of a non-static member
function cannot be used, so static member functions must be implement.
■
Objects cannot be instantiated or deleted in ISR code.
■
C++ code used to execute in an ISR should restrict itself to Embedded C++. No
exceptions nor run-time type identification (RTTI) should be used.
All VxWorks utility libraries, such as the linked-list and ring-buffer libraries, can
be used by ISRs. As discussed earlier (4.5 Task Error Status: errno, p.184), the global
variable errno is saved and restored as a part of the interrupt enter and exit code
generated by the intConnect( ) facility. Thus, errno can be referenced and
modified by ISRs as in any other code. Table 4-23 lists routines that can be called
from ISRs.
246
4 Multitasking
4.20 Interrupt Service Routines
Table 4-23
Routines Callable by ISRs
Library
Routine
bLib
All routines
errnoLib
errnoGet( ), errnoSet( )
eventLib
eventSend( )
fppArchLib
fppSave( ), fppRestore( )
intLib
intContext( ), intCount( ), intVecSet( ), intVecGet( )
intArchLib
intLock( ), intUnlock( )
logLib
logMsg( )
lstLib
All routines except lstFree( )
mathALib
All routines, if fppSave( )/fppRestore( ) are used
msgQLib
msgQSend( )
rngLib
All routines except rngCreate( ) and rngDelete( )
pipeDrv
write( )
selectLib
selWakeup( ), selWakeupAll( )
semLib
semGive( ) except mutual-exclusion semaphores, semFlush( )
semPxLib
sem_post( )
sigLib
kill( )
taskLib
taskSuspend( ), taskResume( ), taskPrioritySet( ),
taskPriorityGet( ), taskIdVerify( ), taskIdDefault( ),
taskIsReady( ), taskIsSuspended( ), taskTcb( )
tickLib
tickAnnounce( ), tickSet( ), tickGet( )
tyLib
tyIRd( ), tyITx( )
vxLib
vxTas( ), vxMemProbe( )
wdLib
wdStart( ), wdCancel( )
4
247
VxWorks
Kernel Programmer's Guide, 6.6
4.20.6 Exceptions at Interrupt Level
When a task causes a hardware exception such as an illegal instruction or bus
error, the task is suspended and the rest of the system continues uninterrupted.
However, when an ISR causes such an exception, there is no safe recourse for the
system to handle the exception. The ISR has no context that can be suspended.
Instead, VxWorks stores the description of the exception in a special location in
low memory and executes a system restart.
The VxWorks boot loader tests for the presence of the exception description in low
memory and if it is detected, display it on the system console. The boot loader’s e
command re-displays the exception description; see Wind River Workbench User’s
Guide: Setting up Your Hardware.
One example of such an exception is the following message:
workQPanic: Kernel work queue overflow.
This exception usually occurs when kernel calls are made from interrupt level at a
very high rate. It generally indicates a problem with clearing the interrupt signal
or a similar driver problem. (See 4.20.4 ISRs and the Kernel Work Queue, p.244.)
4.20.7 Reserving High Interrupt Levels
The VxWorks interrupt support described earlier in this section is acceptable for
most applications. However, on occasion, low-level control is required for events
such as critical motion control or system failure response. In such cases it is
desirable to reserve the highest interrupt levels to ensure zero-latency response to
these events. To achieve zero-latency response, VxWorks provides the routine
intLockLevelSet( ), which sets the system-wide interrupt-lockout level to the
specified level. If you do not specify a level, the default is the highest level
supported by the processor architecture. For information about
architecture-specific implementations of intLockLevelSet( ), see the VxWorks
Architecture Supplement.
!
CAUTION: Some hardware prevents masking certain interrupt levels; check the
hardware manufacturer’s documentation.
248
4 Multitasking
4.20 Interrupt Service Routines
4.20.8 Additional Restrictions for ISRs at High Interrupt Levels
ISRs connected to interrupt levels that are not locked out (either an interrupt level
higher than that set by intLockLevelSet( ), or an interrupt level defined in
hardware as non-maskable) have special restrictions:
4
■
The ISR can be connected only with intVecSet( ).
■
The ISR cannot use any VxWorks operating system facilities that depend on
interrupt locks for correct operation. The effective result is that the ISR cannot
safely make any call to any VxWorks function, except reboot.
For more information, see the VxWorks Architecture Supplement for the architecture
in question.
!
WARNING: The use of NMI with any VxWorks functionality, other than reboot, is
not recommended. Routines marked as interrupt safe do not imply they are NMI
safe and, in fact, are usually the very ones that NMI routines must not call (because
they typically use intLock( ) to achieve the interrupt safe condition).
4.20.9 Interrupt-to-Task Communication
While it is important that VxWorks support direct connection of ISRs that run at
interrupt level, interrupt events usually propagate to task-level code. Many
VxWorks facilities are not available to interrupt-level code, including I/O to any
device other than pipes. The following techniques can be used to communicate
from ISRs to task-level code:
■
Shared Memory and Ring Buffers
ISRs can share variables, buffers, and ring buffers with task-level code.
■
Semaphores
ISRs can give semaphores (except for mutual-exclusion semaphores and
VxMP shared semaphores) that tasks can take and wait for.
■
Message Queues
ISRs can send messages to message queues for tasks to receive (except for
shared message queues using VxMP). If the queue is full, the message is
discarded.
249
VxWorks
Kernel Programmer's Guide, 6.6
■
Pipes
ISRs can write messages to pipes that tasks can read. Tasks and ISRs can write
to the same pipes. However, if the pipe is full, the message written is discarded
because the ISR cannot block. ISRs must not invoke any I/O routine on pipes
other than write( ).
■
Signals
ISRs can signal tasks, causing asynchronous scheduling of their signal
handlers.
■
VxWorks Events
ISRs can send VxWorks events to tasks.
250
5
POSIX Facilities
5.1 Introduction 252
5.2 Configuring VxWorks with POSIX Facilities 253
5.3 General POSIX Support 255
5.4 POSIX Header Files 257
5.5 POSIX Namespace 259
5.6 POSIX Clocks and Timers 259
5.7 POSIX Asynchronous I/O 263
5.8 POSIX Advisory File Locking 263
5.9 POSIX Page-Locking Interface 264
5.10 POSIX Threads 264
5.11 POSIX Thread Mutexes and Condition Variables 273
5.12 POSIX and VxWorks Scheduling 277
5.13 POSIX Semaphores 289
5.14 POSIX Message Queues 299
5.15 POSIX Signals 313
5.16 POSIX Memory Management 313
251
VxWorks
Kernel Programmer's Guide, 6.6
5.1 Introduction
VxWorks provides extensive POSIX support in many of its native kernel libraries.
To facilitate application portability, VxWorks provides additional POSIX
interfaces as optional components. In the kernel, VxWorks implements some of the
traditional interfaces described by the POSIX standard IEEE Std 1003.1 (POSIX.1)
as well as many of the real-time interfaces in the POSIX.1 optional functionality.
For detailed information about POSIX standards and facilities, see The Open
Group Web sites at http://www.opengroup.org/ and http://www.unix.org/.
POSIX and Real-Time Systems
While VxWorks provides many POSIX compliant APIs, not all POSIX APIs are
suitable for embedded and real-time systems, or are entirely compatible with the
VxWorks operating system architecture. In a few cases, therefore, Wind River has
imposed minor limitations on POSIX functionality to serve either real-time
systems or VxWorks compatibility. For example:
■
Swapping memory to disk is not appropriate in real-time systems, and
VxWorks provides no facilities for doing so. It does, however, provide POSIX
page-locking routines to facilitate porting code to VxWorks. The routines
otherwise provide no useful function—pages are always locked in VxWorks
systems (for more information see 5.9 POSIX Page-Locking Interface, p.264).
■
VxWorks tasks are scheduled on a system-wide basis; processes themselves
cannot be scheduled. As a consequence, while POSIX access routines allow
two values for contention scope (PTHREAD_SCOPE_SYSTEM and
PTHREAD_SCOPE_PROCESS), only system-wide scope is implemented in
VxWorks for these routines (for more information, see 5.10 POSIX Threads,
p.264 and 5.12 POSIX and VxWorks Scheduling, p.277).
Any such limitations on POSIX functionality are identified in this chapter, or in
other chapters of this guide that provide more detailed information on specific
POSIX APIs.
POSIX and VxWorks Facilities
This chapter describes the POSIX support provided by VxWorks and
VxWorks-specific POSIX extensions. In addition, it compares native VxWorks
facilities with similar POSIX facilities that are also available with VxWorks.
The qualifier VxWorks is used in this chapter to identify native non-POSIX APIs for
purposes of comparison with POSIX APIs. For example, you can find a discussion
of VxWorks semaphores contrasted to POSIX semaphores in 5.13.1 Comparison of
252
5 POSIX Facilities
5.2 Configuring VxWorks with POSIX Facilities
POSIX and VxWorks Semaphores, p.290, although POSIX semaphores are also
implemented in VxWorks.
VxWorks extensions to POSIX are identified as such.
NOTE: This chapter provides information about POSIX facilities available in the
kernel. For information about facilities available for real-time processes (RTPs), see
the corresponding chapter in the VxWorks Application Programmer’s Guide.
5.2 Configuring VxWorks with POSIX Facilities
VxWorks provides extensive POSIX support for the kernel with many of its
libraries (see 5.3 General POSIX Support, p.255), but the default configuration of
VxWorks does not include many other POSIX facilities that may be used. The
optional VxWorks components that provide support for individual POSIX
libraries are described in this section.
General POSIX support for kernel space can be provided with the BUNDLE_POSIX
component bundle. If memory constraints require a finer-grained configuration,
individual components can be used for selected features. See the configuration
instructions for individual POSIX features throughout this chapter for more
information in this regard.
!
CAUTION: The set of components used for POSIX support in kernel space is not the
same as the set of components used for POSIX support in user space. For
information about the components for user space, see Table 5-1, and see the
VxWorks Application Programmer's Guide: POSIX Facilities for the appropriate
component bundle.
5.2.1 VxWorks Components for POSIX Facilities
Table 5-1 provides an overview of the individual VxWorks components that must
be configured in the kernel to provide support for the specified POSIX facilities.
Networking facilities are described in the Wind River Network Stack for VxWorks 6
Programmer’s Guide.
253
5
VxWorks
Kernel Programmer's Guide, 6.6
Table 5-1
VxWorks Components Providing POSIX Facilities
POSIX Facility
Required VxWorks Component
for Kernel
for Processes
Standard C library INCLUDE_ANSI_* components
Dinkum C library (libc)
Asynchronous I/O INCLUDE_POSIX_AIO,
with system driver INCLUDE_POSIX_AIO_SYSDR
V and INCLUDE_PIPES
INCLUDE_POSIX_CLOCKS and
INCLUDE_POSIX_TIMERS
Clocks
INCLUDE_POSIX_CLOCKS
INCLUDE_POSIX_CLOCKS
Directory and file
utilities
INCLUDE_POSIX_DIRLIB
N/A
ftruncate( )
INCLUDE_POSIX_FTRUNC
N/A
Memory locking
INCLUDE_POSIX_MEM
N/A
Memory
management
N/A
INCLUDE_RTP
Memory-mapped
files
N/A
INCLUDE_POSIX_MAPPED_FILES
Shared memory
objects
N/A
INCLUDE_POSIX_MAPPED_FILES and
INCLUDE_POSIX_SHM
Message queues
INCLUDE_POSIX_MQ
INCLUDE_POSIX_MQ
pthreads
INCLUDE_POSIX_THREADS
INCLUDE_POSIX_CLOCKS,
INCLUDE_POSIX_PTHREAD_SCHEDULE,
and INCLUDE_PTHREAD_CPUTIME
Process Scheduling INCLUDE_POSIX_SCHED
API
N/A
Semaphores
INCLUDE_POSIX_SEM
INCLUDE_POSIX_SEM
Signals
INCLUDE_POSIX_SIGNALS
N/A
Timers
INCLUDE_POSIX_TIMERS
INCLUDE_POSIX_TIMERS
Trace
N/A
INCLUDE_POSIX_TRACE
254
5 POSIX Facilities
5.3 General POSIX Support
Table 5-1
VxWorks Components Providing POSIX Facilities (cont’d)
POSIX Facility
POSIX PSE52
support
Required VxWorks Component
for Kernel
for Processes
N/A
BUNDLE_RTP_POSIX_PSE52
5
5.3 General POSIX Support
Many POSIX-compliant libraries are provided for VxWorks. These libraries are
listed in Table 5-2; see the API references for these libraries for detailed
information.
255
VxWorks
Kernel Programmer's Guide, 6.6
Table 5-2
POSIX Libraries
Functionality
Library
Asynchronous I/O
aioPxLib
Buffer manipulation
bLib
Clock facility
clockLib
Directory handling
dirLib
Environment handling
C Library
File duplication
iosLib
File management
fsPxLib and ioLib
I/O functions
ioLib
Options handling
getopt
POSIX message queues
mqPxLib
POSIX semaphores
semPxLib
POSIX timers
timerLib
POSIX threads
pthreadLib
Standard I/O and some ANSI
C Library
Math
C Library
Memory allocation
memLib and memPartLib
Network/Socket APIs
network libraries
String manipulation
C Library
Trace facility
pxTraceLib
The following sections of this chapter describe the optional POSIX API
components that are provided in addition to the native VxWorks APIs.
256
5 POSIX Facilities
5.4 POSIX Header Files
!
CAUTION: Wind River advises that you do not use both POSIX libraries and native
VxWorks libraries that provide similar functionality. Doing so may result in
undesirable interactions between the two, as some POSIX APIs manipulate
resources that are also used by native VxWorks APIs. For example, do not use
tickLib routines to manipulate the system's tick counter if you are also using
clockLib routines, do not use the taskLib API to change the priority of a POSIX
thread instead of the pthread API, and so on.
Checking for POSIX Support at Run-time
A POSIX application can use the following APIs at run-time to determine the
status of POSIX support in the system:
■
The sysconf( ) routine returns the current values of the configurable system
variables, allowing an application to determine whether an optional feature is
supported or not, and the precise value of system's limits.
■
The confstr( ) routine returns a string associated with a system variable. With
this release, the confstr( ) routine returns a string only for the system's default
path.
■
The uname( ) routine lets an application get information about the system on
which it is running. The identification information provided for VxWorks is
the system name (VxWorks), the network name of the system, the system's
release number, the machine name (BSP model), the architecture's endianness,
the kernel version number, the processor name (CPU family), the BSP revision
level, and the system's build date.
5.4 POSIX Header Files
The POSIX 1003.1 standard defines a set of header files as part of the application
development environment. The VxWorks user-side development environment
provides more POSIX header files than the kernel’s, and their content is also more
in agreement with the POSIX standard than the kernel header files.
The POSIX header files available for the kernel development environment are
listed in Table 5-3.
257
5
VxWorks
Kernel Programmer's Guide, 6.6
Table 5-3
POSIX Header Files
Header File
Description
aio.h
asynchronous input and output
assert.h
verify program assertion
ctype.h
character types
dirent.h
format of directory entries
errno.h
system error numbers
fcntl.h
file control options
limits.h
implementation-defined constants
locale.h
category macros
math.h
mathematical declarations
mqueue.h
message queues
pthread.h
pthreads
sched.h
execution scheduling
semaphore.h
semaphores
setjmp.h
stack environment declarations
signal.h
signals
stdio.h
standard buffered input/output
stdlib.h
standard library definitions
string.h
string operations
sys/mman.h
memory management declarations
sys/resource.h
definitions for XSI resource operations
sys/stat.h
data returned by the stat( ) function
sys/types.h
data types
sys/un.h
definitions for UNIX domain sockets
258
5 POSIX Facilities
5.5 POSIX Namespace
Table 5-3
POSIX Header Files (cont’d)
Header File
Description
time.h
time types
trace.h
trace facility
unistd.h
standard symbolic constants and types
utime.h
access and modification times structure
5
5.5 POSIX Namespace
POSIX namespace isolation can be provided for user-mode (RTP) applications. For
information in this regard, see the VxWorks Application Programmer’s Guide: POSIX
Facilities.
5.6 POSIX Clocks and Timers
VxWorks provides POSIX 1003.1b standard clock and timer interfaces.
POSIX Clocks
POSIX defines various software (virtual) clocks, which are identified as the
CLOCK_REALTIME clock, CLOCK_MONOTONIC clock, process CPU-time clocks,
and thread CPU-time clocks. These clocks all use one system hardware timer.
The real-time clock and the monotonic clock are system-wide clocks, and are
therefore supported for both the VxWorks kernel and processes. The process
CPU-time clocks are not supported in VxWorks. The thread CPU-time clocks are
supported for POSIX threads running in processes. A POSIX thread can use the
real-time clock, the monotonic clock, and a thread CPU-time clock for its
application.
259
VxWorks
Kernel Programmer's Guide, 6.6
For information about thread CPU-time clocks, see the VxWorks Application
Programmer’s Guide: POSIX Facilities.
The real-time clock can be reset (but only from the kernel). The monotonic clock
cannot be reset, and provides the time that has elapsed since the system booted.
The real-time clock can be accessed with the POSIX clock and timer routines by
using the clock_id parameter CLOCK_REALTIME. A real-time clock can be reset at
run time with a call to clock_settime( ) from within the kernel (not from a process).
The monotonic clock can be accessed by calling clock_gettime( ) with a clock_id
parameter of CLOCK_MONOTONIC. A monotonic clock keeps track of the time
that has elapsed since system startup; that is, the value returned by
clock_gettime( ) is the amount of time (in seconds and nanoseconds) that has
passed since the system booted. A monotonic clock cannot be reset. Applications
can therefore rely on the fact that any measurement of a time interval that they
might make has not been falsified by a call to clock_settime( ).
Both CLOCK_REALTIME and CLOCK_MONOTONIC are defined in time.h.
See Table 5-4 for a list of the POSIX clock routines. The obsolete VxWorks-specific
POSIX extension clock_setres( ) is provided for backwards-compatibility
purposes. For more information about clock routines, see the API reference for
clockLib.
Table 5-4
POSIX Clock Routines
Routine
clock_getres( )
Description
Get the clock resolution (CLOCK_REALTIME and
CLOCK_MONOTONIC).
clock_setres( )
Set the clock resolution.
Obsolete VxWorks-specific POSIX extension.
clock_gettime( )
Get the current clock time (CLOCK_REALTIME and
CLOCK_MONOTONIC).
clock_settime( )
Set the clock to a specified time for CLOCK_REALTIME
(fails for CLOCK_MONOTONIC; not supported for a
thread CPU-time clock in the kernel).
To include the clockLib library in the system, configure VxWorks with the
INCLUDE_POSIX_CLOCKS component. For thread CPU-time clocks, the
INCLUDE_POSIX_PTHREAD_SCHEDULER and
INCLUDE_POSIX_THREAD_CPUTIME components must be used as well.
260
5 POSIX Facilities
5.6 POSIX Clocks and Timers
POSIX Timers
The POSIX timer facility provides routines for tasks to signal themselves at some
time in the future. Routines are provided to create, set, and delete a timer.
Timers are created based on clocks. In the kernel, the CLOCK_REALTIME and
CLOCK_MONOTONIC clocks are supported for timers. In processes, the
CLOCK_REALTIME clock, CLOCK_MONOTONIC clock, and thread CPU-time
clocks (including CLOCK_THREAD_CPUTIME_ID clock) are supported.
When a timer goes off, the default signal, SIGALRM, is sent to the task. To install a
signal handler that executes when the timer expires, use the sigaction( ) routine
(see 4.18 Signals, p.226).
See Table 5-5 for a list of the POSIX timer routines. The VxWorks timerLib library
includes a set of VxWorks-specific POSIX extensions: timer_open( ),
timer_close( ), timer_cancel( ), timer_connect( ), and timer_unlink( ). These
routines allow for an easier and more powerful use of POSIX timers on VxWorks.
For more information, see the VxWorks API reference for timerLib.
Table 5-5
POSIX Timer Routines
Routine
Description
timer_create( )
Allocate a timer using the specified clock for a timing base
(CLOCK_REALTIME or CLOCK_MONOTONIC).
timer_delete( )
Remove a previously created timer.
timer_open( )
Open a named timer.
VxWorks-specific POSIX extension.
timer_close( )
Close a named timer.
VxWorks-specific POSIX extension.
timer_gettime( )
Get the remaining time before expiration and the reload
value.
timer_getoverrun( ) Return the timer expiration overrun.
timer_settime( )
Set the time until the next expiration and arm timer.
timer_cancel( )
Cancel a timer.
VxWorks-specific POSIX extension.
261
5
VxWorks
Kernel Programmer's Guide, 6.6
Table 5-5
Example 5-1
POSIX Timer Routines (cont’d)
Routine
Description
timer_connect( )
Connect a user routine to the timer signal.
VxWorks-specific POSIX extension.
timer_unlink( )
Unlink a named timer.
VxWorks-specific POSIX extension.
nanosleep( )
Suspend the current pthread (task) until the time interval
elapses.
sleep( )
Delay for a specified amount of time.
alarm( )
Set an alarm clock for delivery of a signal.
POSIX Timers
/* This example creates a new timer and stores it in timerid. */
/* includes */
#include
#include
int createTimer (void)
{
timer_t timerid;
/* create timer */
if (timer_create (CLOCK_REALTIME, NULL, &timerid) == ERROR)
{
printf ("create FAILED\n");
return (ERROR);
}
return (OK);
}
The POSIX nanosleep( ) routine provides specification of sleep or delay time in
units of seconds and nanoseconds, in contrast to the ticks used by the VxWorks
taskDelay( ) function. Nevertheless, the precision of both is the same, and is
determined by the system clock rate; only the units differ.
To include the timerLib library in a system, configure VxWorks with the
INCLUDE_POSIX_TIMERS component.
262
5 POSIX Facilities
5.7 POSIX Asynchronous I/O
5.7 POSIX Asynchronous I/O
POSIX asynchronous I/O (AIO) routines are provided by the aioPxLib library.
The VxWorks AIO implementation meets the specification of the POSIX 1003.1
standard. For more information, see 7.7 Asynchronous Input/Output, p.381.
5
5.8 POSIX Advisory File Locking
POSIX advisory file locking provides byte-range locks on POSIX-conforming files
(for VxWorks, this means files in an HRFS file system). The VxWorks
implementation meets the specification of the POSIX 1003.1 standard.
POSIX advisory file locking is provided through the fcntl( ) file control function.
To include POSIX advisory file locking facilities in VxWorks, configure the system
with the INCLUDE_POSIX_ADVISORY_FILE_LOCKING component.
The VxWorks implementation of advisory file locking involves a behavioral
difference with regard to deadlock detection because VxWorks processes are not
scheduled. Note that this distinction only matters if you have multiple pthreads (or
tasks) within one process (RTP).
According to POSIX, advisory locks are identified by a process ID, and when a
process exits all of its advisory locks are destroyed, which is true for VxWorks. But
because VxWorks processes cannot themselves be scheduled, individual advisory
locks on a given byte range of a file have two owners: the pthread (or task) that
actually holds the lock, and the process that contains the pthread. In addition, the
calculation of whether one lock would deadlock another lock is done on a pthread
basis, rather than a process basis.
This means that deadlocks are detected if the pthread requesting a new lock would
block on any pthread (in any given process) that is currently blocked (whether
directly or indirectly) on any advisory lock that is held by the requesting pthread.
Immediate-blocking detection (F_SETLK requests) always fail immediately if the
requested byte range cannot be locked without waiting for some other lock,
regardless of the identity of the owner of that lock.
263
VxWorks
Kernel Programmer's Guide, 6.6
5.9 POSIX Page-Locking Interface
The real-time extensions of the POSIX 1003.1 standard are used with operating
systems that perform paging and swapping. On such systems, applications that
attempt real-time performance can use the POSIX page-locking facilities to protect
certain blocks of memory from paging and swapping.
VxWorks does not support memory paging and swapping because the serious
delays in execution time that they cause are undesirable in a real-time system.
However, page-locking routines can be included in VxWorks to facilitate porting
POSIX applications to VxWorks.
These routines do not perform any function, as all pages are always kept in
memory.
The POSIX page-locking routines are part of the memory management library,
mmanPxLib, and are listed in Table 5-6.
Table 5-6
POSIX Page-Locking Routines
Routine
Purpose on Systems with Paging or Swapping
mlockall( )
Locks into memory all pages used by a task.
munlockall( )
Unlocks all pages used by a task.
mlock( )
Locks a specified page.
munlock( )
Unlocks a specified page.
To include the mmanPxLib library in the system, configure VxWorks with the
INCLUDE_POSIX_MEM component.
5.10 POSIX Threads
POSIX threads (also known as pthreads) are similar to VxWorks tasks, but with
additional characteristics. In VxWorks pthreads are implemented on top of native
tasks, but maintain pthread IDs that differ from the IDs of the underlying tasks.
The main reasons for including POSIX thread support in VxWorks are the
following:
264
5 POSIX Facilities
5.10 POSIX Threads
■
For porting POSIX applications to VxWorks.
■
To make use of the POSIX thread scheduler in real-time processes (including
concurrent scheduling policies).
For information about POSIX thread scheduler, see 5.12 POSIX and VxWorks
Scheduling, p.277.
5
5.10.1 POSIX Thread Attributes
A major difference between VxWorks tasks and POSIX threads is the way in which
options and settings are specified. For VxWorks tasks these options are set with the
task creation API, usually taskSpawn( ).
POSIX threads, on the other hand, have characteristics that are called attributes.
Each attribute contains a set of values, and a set of access routines to retrieve and set
those values. You specify all pthread attributes before pthread creation in the
attributes object pthread_attr_t. In a few cases, you can dynamically modify the
attribute values of a pthread after its creation.
5.10.2 VxWorks-Specific Pthread Attributes
The VxWorks implementation of POSIX threads provides two additional pthread
attributes (which are POSIX extensions)—pthread name and pthread options—as
well as routines for accessing them.
Pthread Name
While POSIX threads are not named entities, the VxWorks tasks upon which they
are based are named. By default the underlying task elements are named
pthrNumber (for example, pthr3). The number part of the name is incremented
each time a new thread is created (with a roll-over at 2^32 - 1). It is, however,
possible to name these tasks using the thread name attribute.
■
Attribute Name: threadname
■
Possible Values: a null-terminated string of characters
■
Default Value: none (the default naming policy is used)
■
Access Functions (VxWorks-specific POSIX extensions):
pthread_attr_setname( ) and pthread_attr_getname( )
265
VxWorks
Kernel Programmer's Guide, 6.6
Pthread Options
POSIX threads are agnostic with regard to target architecture. Some VxWorks
tasks, on the other hand, may be created with specific options in order to benefit
from certain features of the architecture. For example, for the Altivec-capable
PowerPC architecture, tasks must be created with the VX_ALTIVEC_TASK in order
to make use of the Altivec processor. The pthread options attribute can be used to
set such options for the VxWorks task upon which the POSIX thread is based.
■
Attribute Name: threadoptions
■
Possible Values: the same as the VxWorks task options. See taskLib.h
■
Default Value: none (the default task options are used)
■
Access Functions (VxWorks-specific POSIX extensions):
pthread_attr_setopt( ) and pthread_attr_getopt( )
5.10.3 Specifying Attributes when Creating Pthreads
The following examples create a pthread using the default attributes and use
explicit attributes.
Example 5-2
Creating a pthread Using Explicit Scheduling Attributes
pthread_t tid;
pthread_attr_t attr;
int ret;
pthread_attr_init(&attr);
/* set the inheritsched attribute to explicit */
pthread_attr_setinheritsched(&attr, PTHREAD_EXPLICIT_SCHED);
/* set the schedpolicy attribute to SCHED_FIFO */
pthread_attr_setschedpolicy(&attr, SCHED_FIFO);
/* create the pthread */
ret = pthread_create(&tid, &attr, entryFunction, entryArg);
266
5 POSIX Facilities
5.10 POSIX Threads
Example 5-3
Creating a pthread Using Default Attributes
pthread_t tid;
int ret;
/* create the pthread with NULL attributes to designate default values */
ret = pthread_create(&tid, NULL, entryFunction, entryArg);
Example 5-4
Designating Your Own Stack for a pthread
5
pthread_t threadId;
pthread_attr_t attr;
void * stackaddr = NULL;
int stacksize = 0;
/* initialize the thread's attributes */
pthread_attr_init (&attr);
/*
* Allocate memory for a stack region for the thread. Malloc() is used
* for simplification since a real-life case is likely to use
memPartAlloc()
* on the kernel side, or mmap() on the user side.
*/
stacksize = 2 * 4096 /* let's allocate two pages */ stackaddr = malloc
(stacksize);
if (stackbase == NULL)
{
printf ("FAILED: mystack: malloc failed\n");
return (-1);
}
/* set the stackaddr attribute */
pthread_attr_setstackaddr (&attr, stackaddr);
/* set the stacksize attribute */
pthread_attr_setstacksize (&attr, stacksize);
/* set the schedpolicy attribute to SCHED_FIFO */
pthread_attr_setschedpolicy (&attr, SCHED_FIFO);
/* create the pthread */
ret = pthread_create (&threadId, &attr, mystack_thread, 0);
267
VxWorks
Kernel Programmer's Guide, 6.6
5.10.4 POSIX Thread Creation and Management
VxWorks provides many POSIX thread routines. Table 5-7 lists a few that are
directly relevant to pthread creation or execution. See the VxWorks API reference
for information about the other routines, and more details about all of them.
Table 5-7
POSIX Thread Routines
Routine
Description
pthread_create( )
Create a pthread.
pthread_cancel( )
Cancel the execution of a pthread
pthread_detach( )
Detach a running pthread so that it cannot be
joined by another pthread.
pthread_join( )
Wait for a pthread to terminate.
pthread_getschedparam( )
Dynamically set value of scheduling priority
attribute.
pthread_setschedparam( )
Dynamically set scheduling priority and policy
parameter.
sched_get_priority_max( )
Get the maximum priority that a pthread can get.
sched_get_priority_min( )
Get the minimum priority that a pthread can get.
sched_rr_get_interval( )
Get the time quantum of execution of the
round-robin policy.
sched_yield( )
Relinquishes the CPU.
268
5 POSIX Facilities
5.10 POSIX Threads
5.10.5 POSIX Thread Attribute Access
The POSIX attribute-access routines are described in Table 5-8. The
VxWorks-specific POSIX extension routines are described in section
5.10.2 VxWorks-Specific Pthread Attributes, p.265.
Table 5-8
POSIX Thread Attribute-Access Routines
5
Routine
Description
pthread_attr_getstacksize( )
Get value of the stack size attribute.
pthread_attr_setstacksize( )
Set the stack size attribute.
pthread_attr_getstackaddr( )
Get value of stack address attribute.
pthread_attr_setstackaddr( )
Set value of stack address attribute.
pthread_attr_getdetachstate( )
Get value of detachstate attribute (joinable or
detached).
pthread_attr_setdetachstate( )
Set value of detachstate attribute (joinable or
detached).
pthread_attr_getscope( )
Get contention scope.
Only PTHREAD_SCOPE_SYSTEM is
supported for VxWorks.
pthread_attr_setscope( )
Set contention scope.
Only PTHREAD_SCOPE_SYSTEM is
supported for VxWorks.
pthread_attr_getinheritsched( )
Get value of scheduling-inheritance attribute.
pthread_attr_setinheritsched( )
Set value of scheduling-inheritance attribute.
pthread_attr_getschedpolicy( )
Get value of the scheduling-policy attribute
(which is not used by default).
pthread_attr_setschedpolicy( )
Set scheduling-policy attribute (which is not
used by default).
pthread_attr_getschedparam( )
Get value of scheduling priority attribute.
pthread_attr_setschedparam( )
Set scheduling priority attribute.
269
VxWorks
Kernel Programmer's Guide, 6.6
Table 5-8
POSIX Thread Attribute-Access Routines (cont’d)
Routine
Description
pthread_attr_getopt( )
Get the task options applying to the pthread.
VxWorks-specific POSIX extension.
pthread_attr_setopt( )
Set non-default task options for the pthread.
VxWorks-specific POSIX extension.
pthread_attr_getname( )
Get the name of the pthread.
VxWorks-specific POSIX extension.
pthread_attr_setname( )
Set a non-default name for the pthread.
VxWorks-specific POSIX extension.
5.10.6 POSIX Thread Private Data
POSIX threads can store and access private data; that is, pthread-specific data.
They use a key maintained for each pthread by the pthread library to access that
data. A key corresponds to a location associated with the data. It is created by
calling pthread_key_create( ) and released by calling pthread_key_delete( ). The
location is accessed by calling pthread_getspecific( ) and pthread_setspecific( ).
This location represents a pointer to the data, and not the data itself, so there is no
limitation on the size and content of the data associated with a key.
The pthread library supports a maximum of 256 keys for all the pthreads in the
kernel.
The pthread_key_create( ) routine has an option for a destructor function, which
is called when the creating pthread exits or is cancelled, if the value associated with
the key is non-NULL.
This destructor function frees the storage associated with the data itself, and not
with the key. It is important to set a destructor function for preventing memory
leaks to occur when the pthread that allocated memory for the data is cancelled.
The key itself should be freed as well, by calling pthread_key_delete( ), otherwise
the key cannot be reused by the pthread library.
5.10.7 POSIX Thread Cancellation
POSIX provides a mechanism, called cancellation, to terminate a pthread
gracefully. There are two types of cancellation: deferred and asynchronous.
270
5 POSIX Facilities
5.10 POSIX Threads
Deferred cancellation causes the pthread to explicitly check to see if it was
cancelled. This happens in one of the two following ways:
■
■
The code of the pthread executes calls to pthread_testcancel( ) at regular
intervals.
The pthread calls a function that contains a cancellation point during which the
pthread may be automatically cancelled.
Asynchronous cancellation causes the execution of the pthread to be forcefully
interrupted and a handler to be called, much like a signal.1
Automatic cancellation points are library routines that can block the execution of
the pthread for a lengthy period of time.
NOTE: While the msync( ), fcntl( ), and tcdrain( ) routines are mandated POSIX
1003.1 cancellation points, they are not provided with VxWorks for this release.
The POSIX cancellation points provided in VxWorks libraries are described in
Table 5-9.
1. Asynchronous cancellation is actually implemented with a special signal, SIGCNCL, which
users should be careful not to block or to ignore.
271
5
VxWorks
Kernel Programmer's Guide, 6.6
Table 5-9
Pthread Cancellation Points in VxWorks Libraries
Library
Routines
aioPxLib
aio_suspend( )
ioLib
creat( ), open( ), read( ), write( ), close( ), fsync( ), fdatasync( )
mqPxLib
mq_receive( ), mq_send( )
pthreadLib pthread_cond_timedwait( ), pthread_cond_wait( ), pthread_join( ),
pthread_testcancel( )
semPxLib
sem_wait( )
sigLib
pause( ), sigsuspend( ), sigtimedwait( ), sigwait( ), sigwaitinfo( )
timerLib
sleep( ), nanosleep( )
Routines that can be used with cancellation points of pthreads are listed in
Table 5-10.
Table 5-10
Pthread Cancellation Routines
Routine
Description
pthread_cancel( )
Cancel execution of a pthread.
pthread_testcancel( )
Create a cancellation point in the calling pthread.
pthread_setcancelstate( )
Enables or disables cancellation.
pthread_setcanceltype( )
Selects deferred or asynchronous cancellation.
pthread_cleanup_push( )
Registers a function to be called when the pthread is
cancelled, exits, or calls pthread_cleanup_pop( ) with
a non-null run parameter.
pthread_cleanup_pop( )
Unregisters a function previously registered with
pthread_cleanup_push( ). This function is
immediately executed if the run parameter is non-null.
272
5 POSIX Facilities
5.11 POSIX Thread Mutexes and Condition Variables
5.11 POSIX Thread Mutexes and Condition Variables
Pthread mutexes (mutual exclusion variables) and condition variables provide
compliance with the POSIX 1003.1c standard. Like POSIX threads, mutexes and
condition variables have attributes associated with them. Mutex attributes are held
in a data type called pthread_mutexattr_t, which contains two attributes, protocol
and prioceiling.
The routines used to manage these attributes are described below. For more
information about these and other mutex routines, see the API reference for
pthreadLib.
5.11.1 Thread Mutexes
The routines that can be used to act directly on a mutex object and on the mutex
attribute object are listed in Table 5-11 and Table 5-12 (respectively).
Table 5-11
POSIX Routines Acting on a Mutex Object
Routine
Description
pthread_mutex_destroy( )
Destroy a mutex.
pthread_mutex_init( )
Initialize a mutex.
pthread_mutex_getprioceiling( )
Get the priority ceiling of a mutex.
pthread_mutex_setprioceiling( )
Set the priority ceiling of a mutex.
pthread_mutex_lock( )
Lock a mutex.
pthread_mutex_trylock( )
Check and lock a mutex if available.
pthread_mutex_unlock( )
Unlock a mutex.
273
5
VxWorks
Kernel Programmer's Guide, 6.6
Table 5-12
POSIX Routines Acting on a Mutex Attribute Object
Routine
Description
pthread_mutexattr_init( )
Initialize mutex attributes object.
pthread_mutexattr_destroy( )
Destroy mutex attributes object.
pthread_mutexattr_getprioceiling( )
Get prioceiling attribute of mutex
attributes object.
pthread_mutexattr_setprioceiling( )
Set prioceiling attribute of mutex
attributes object.
pthread_mutexattr_getprotocol( )
Get protocol attribute of mutex
attributes object.
pthread_mutexattr_setprotocol( )
Set protocol attribute of mutex
attributes object.
Protocol Mutex Attribute
The protocol mutex attribute defines how the mutex variable deals with the
priority inversion problem (which is described in the section for VxWorks
mutual-exclusion semaphores; see 4.12.3 Mutual-Exclusion Semaphores, p.205).
■
Attribute Name: protocol
■
Possible Values: PTHREAD_PRIO_NONE, PTHREAD_PRIO_INHERIT and
PTHREAD_PRIO_PROTECT
■
Access Routines: pthread_mutexattr_getprotocol( ) and
pthread_mutexattr_setprotocol( )
The PTHREAD_PRIO_INHERIT option is the default value of the protocol attribute
for pthreads created in the kernel (unlike pthreads created in processes, for which
the default is PTHREAD_PRIO_NONE).
The PTHREAD_PRIO_INHERIT value is used to create a mutex with priority
inheritance—and is equivalent to the association of SEM_Q_PRIORITY and
SEM_INVERSION_SAFE options used with semMCreate( ). A pthread owning a
mutex variable created with the PTHREAD_PRIO_INHERIT value inherits the
priority of any higher-priority pthread waiting for the mutex and executes at this
elevated priority until it releases the mutex, at which points it returns to its original
priority.
274
5 POSIX Facilities
5.11 POSIX Thread Mutexes and Condition Variables
Because it might not be desirable to elevate a lower-priority pthread to a priority
above a certain level, POSIX defines the notion of priority ceiling, described below.
Mutual-exclusion variables created with priority protection use the
PTHREAD_PRIO_PROTECT value.
Priority Ceiling Mutex Attribute
5
The prioceiling attribute is the POSIX priority ceiling for mutex variables created
with the protocol attribute set to PTHREAD_PRIO_PROTECT.
■
Attribute Name: prioceiling
■
Possible Values: any valid (POSIX) priority value (0-255, with zero being the
lowest).
■
Access Routines: pthread_mutexattr_getprioceiling( ) and
pthread_mutexattr_setprioceiling( )
■
Dynamic Access Routines: pthread_mutex_getprioceiling( ) and
pthread_mutex_setprioceiling( )
Note that the POSIX priority numbering scheme is the inverse of the VxWorks
scheme. For more information see 5.12.2 POSIX and VxWorks Priority Numbering,
p.279.
A priority ceiling is defined by the following conditions:
■
Any pthread attempting to acquire a mutex, whose priority is higher than the
ceiling, cannot acquire the mutex.
■
Any pthread whose priority is lower than the ceiling value has its priority
elevated to the ceiling value for the duration that the mutex is held.
■
The pthread’s priority is restored to its previous value when the mutex is
released.
5.11.2 Condition Variables
A pthread condition variable corresponds to an object that permits pthreads to
synchronize on an event or state represented by the value of a variable. This is a
more complicated type of synchronization than the one allowed by mutexes only.
Its main advantage is that is allows for passive waiting (as opposed to active
waiting or polling) on a change in the value of the variable. Condition variables are
used in conjunction with mutexes (one mutex per condition variable). The routines
275
VxWorks
Kernel Programmer's Guide, 6.6
that can be used to act directly on a condition variable and on the condition
variable attribute object are listed in Table 5-11 and Table 5-12 (respectively).
Table 5-13
Table 5-14
POSIX Routines Acting on a Condition Variable Object
Routine
Description
pthread_cond_destroy( )
Destroy condition variables.
pthread_cond_init( )
Initialize condition variables.
pthread_cond_broadcast( )
Broadcast a condition.
pthread_cond_signal( )
Signal a condition.
pthread_cond_wait( )
Wait on a condition.
pthread_cond_timedwait( )
Wait on a condition with timeout.
POSIX Routines Acting on a Condition Variable Attribute Object
Routine
Description
pthread_condattr_destroy( )
Destroy condition variable attributes
object.
pthread_condattr_init( )
Initialize condition variable attributes
object.
276
5 POSIX Facilities
5.12 POSIX and VxWorks Scheduling
5.12 POSIX and VxWorks Scheduling
VxWorks can be configured with either the traditional (native) VxWorks scheduler
or with a POSIX thread scheduler. Neither can be used to schedule processes
(RTPs). The only entities that can be scheduled in VxWorks are tasks and pthreads.
The VxWorks implementation of a POSIX thread scheduler is an enhancement of
the traditional VxWorks scheduler that provides additional scheduling facilities
for pthreads running in processes.
With either scheduler, VxWorks tasks and pthreads share a single priority range
and the same global scheduling scheme. With the POSIX thread scheduler,
however, pthreads running in processes may have individual (concurrent)
scheduling policies. Note that VxWorks must be configured with the POSIX thread
scheduler in order to run pthreads in processes.
NOTE: Wind River recommends that you do not use both POSIX APIs and
VxWorks APIs in the same application. Doing so may make a POSIX application
non-compliant.
Table 5-15 provides an overview of how scheduling works for tasks and pthreads,
for each of the schedulers, in both the kernel and processes (RTPs). The key
differences are the following:
■
The POSIX thread scheduler provides POSIX scheduling support for threads
running in processes.
■
In all other cases, the POSIX thread scheduler schedules pthreads and tasks in
the same (non-POSIX) manner as the traditional VxWorks scheduler. (There is
a minor difference between how it handles tasks and pthreads whose priorities
have been lowered; see Differences in Re-Queuing Pthreads and Tasks With
Lowered Priorities, p.285.)
■
The traditional VxWorks scheduler cannot be used to schedule pthreads in
processes. In fact, pthreads cannot be started in processes unless VxWorks is
configured with the POSIX thread scheduler.
The information provided in Table 5-15 is discussed in detail in subsequent
sections.
277
5
VxWorks
Kernel Programmer's Guide, 6.6
Table 5-15
Task and Pthread Scheduling in the Kernel and in Processes
POSIX Thread Scheduler
Execution
Environment
Kernel
Processes
Tasks
Pthreads
Priority-based
preemptive, or
round-robin
scheduling.
Same as task
scheduling.
Priority-based
preemptive, or
round robin
scheduling.
POSIX FIFO,
round-robin,
sporadic, or other
(system default).
No concurrent
scheduling
policies.
Concurrent
scheduling
policies available.
Traditional VxWorks Scheduler
Tasks
Pthreads
Priority-based
preemptive, or
round-robin
scheduling.
Same as task
scheduling.
Priority-based
preemptive, or
round-robin
scheduling.
N/A.
No concurrent
scheduling
policies.
Pthreads cannot be
run in processes
with traditional
VxWorks
scheduler.a
a. The traditional VxWorks scheduler cannot ensure behavioral compliance with the POSIX 1 standard.
278
5 POSIX Facilities
5.12 POSIX and VxWorks Scheduling
5.12.1 Differences in POSIX and VxWorks Scheduling
In general, the POSIX scheduling model and scheduling in a VxWorks system
differ in the following ways—regardless of whether the system is configured with
the Wind River POSIX thread scheduler or the traditional VxWorks scheduler:
■
POSIX supports a two-level scheduling model that includes the concept of
contention scope, by which the scheduling of pthreads can apply system wide
or on a process basis. In VxWorks, on the other hand, processes (RTPs) cannot
themselves be scheduled, and tasks and pthreads are scheduled on a
system-wide (kernel and processes) basis.
■
POSIX applies scheduling policies on a process-by-process and
thread-by-thread basis. VxWorks applies scheduling policies on a
system-wide basis, for all tasks and pthreads, whether in the kernel or in
processes. This means that all tasks and pthreads use either a preemptive
priority scheme or a round-robin scheme. The only exception to this rule is that
pthreads executing in processes can be subject to concurrent (individual)
scheduling policies, including sporadic scheduling (note that the POSIX
thread scheduler must be used in this case).
■
POSIX supports the concept of scheduling allocation domain; that is, the
association between processes or threads and processors. Since VxWorks does
not support multi-processor hardware, there is only one domain on VxWorks
and all the tasks and pthreads are associated to it.
■
The POSIX priority numbering scheme is the inverse of the VxWorks scheme.
For more information see 5.12.2 POSIX and VxWorks Priority Numbering, p.279.
■
VxWorks does not support the POSIX thread-concurrency feature, as all
threads are scheduled. The POSIX thread-concurrency APIs are provided for
application portability, but they have no effect.
5.12.2 POSIX and VxWorks Priority Numbering
The POSIX priority numbering scheme is the inverse of the VxWorks priority
numbering scheme. In POSIX, the higher the number, the higher the priority. In
VxWorks, the lower the number, the higher the priority, where 0 is the highest
priority.
The priority numbers used with the POSIX scheduling library, schedPxLib, do not,
therefore, match those used and reported by all other components of VxWorks.
You can change the default POSIX numbering scheme by setting the global
variable posixPriorityNumbering to FALSE. If you do so, schedPxLib uses the
279
5
VxWorks
Kernel Programmer's Guide, 6.6
VxWorks numbering scheme (a smaller number means a higher priority) and its
priority numbers match those used by the other components of VxWorks.
In the following sections, discussions of pthreads and tasks at the same priority level
refer to functionally equivalent priority levels, and not to priority numbers.
5.12.3 Default Scheduling Policy
All VxWorks tasks and pthreads are scheduled according to the system-wide
default scheduling policy. The only exception to this rule is for pthreads running
in user mode (in processes). In this case, concurrent scheduling policies that differ
from the system default can be applied to pthreads.
Note that pthreads can be run in processes only if VxWorks is configured with the
POSIX thread scheduler; they cannot be run in processes if VxWorks is configured
with the traditional scheduler.
The system-wide default scheduling policy for VxWorks, regardless of which
scheduler is used, is priority-based preemptive scheduling—which corresponds to
the POSIX SCHED_FIFO scheduling policy.
At run-time the active system-wide default scheduling policy can be changed to
round-robin scheduling with the kernelTimeSlice( ) routine. It can be changed
back by calling kernelTimeSlice( ) with a parameter of zero. VxWorks
round-robin scheduling corresponds to the POSIX SCHED_RR policy.
The kernelTimeSlice( ) routine cannot be called in user mode (that is, from a
process). A call with a non-zero parameter immediately affects all kernel and user
tasks, all kernel pthreads, and all user pthreads using the SCHED_OTHER policy.
Any user pthreads running with the SCHED_RR policy are unaffected by the call;
but those started after it use the newly defined timeslice.
280
5 POSIX Facilities
5.12 POSIX and VxWorks Scheduling
5.12.4 VxWorks Traditional Scheduler
The VxWorks traditional scheduler can be used with both tasks and pthreads in
the kernel. It cannot be used with pthreads in processes. If VxWorks is configured
with the traditional scheduler, a pthread_create( ) call in a process fails and the
errno is set to ENOSYS.
The traditional VxWorks scheduler schedules pthreads as if they were tasks. All
tasks and pthreads executing in a system are therefore subject to the current
default scheduling policy (either the priority-based preemptive policy or the
round-robin scheduling policy; see 5.12.3 Default Scheduling Policy, p.280), and
concurrent policies cannot be applied to individual pthreads. For general
information about the traditional scheduler and how it works with tasks, see
4.3 Task Scheduling, p.166.
The scheduling options provided by the traditional VxWorks scheduler are similar
to the POSIX ones. The following pthreads scheduling policies correspond to the
traditional VxWorks scheduling policies:
■
SCHED_FIFO is similar to VxWorks priority-based preemptive scheduling.
There are differences as to where tasks or pthreads are placed in the ready
queue if their priority is lowered; see Caveats About Scheduling Behavior with the
POSIX Thread Scheduler, p.284.
■
SCHED_RR corresponds to VxWorks round-robin scheduling.
■
SCHED_OTHER corresponds to the current system-wide default scheduling
policy. The SCHED_OTHER policy is the default policy for pthreads in
VxWorks.
There is no VxWorks traditional scheduler policy that corresponds to
SCHED_SPORADIC.
Configuring VxWorks with the Traditional Scheduler
VxWorks is configured with the traditional scheduler by default. This scheduler is
provided by the INCLUDE_VX_TRADITIONAL_SCHEDULER component.
Caveats About Scheduling Behavior with the VxWorks Traditional Scheduler
Concurrent scheduling policies are not supported for pthreads in the kernel, and
care must therefore be taken with pthread scheduling-inheritance and scheduling
policy attributes.
If the scheduling-inheritance attribute is set to PTHREAD_EXPLICIT_SCHED and
the scheduling policy to SCHED_FIFO or SCHED_RR, and this policy does not
281
5
VxWorks
Kernel Programmer's Guide, 6.6
match the current system-wide default scheduling policy, the creation of pthreads
fails.
Wind River therefore recommends that you always use
PTHREAD_INHERIT_SCHED (which is the default) as a scheduling-inheritance
attribute. In this case the current VxWorks scheduling policy applies, and the
parent pthread's priority is used. Or, if the pthread must be started with a different
priority than its parent, the scheduling-inheritance attribute can be set to
PTHREAD_EXPLICIT_SCHED and the scheduling policy attribute set to be
SCHED_OTHER (which corresponds to the current system-wide default
scheduling policy.).
In order to take advantage of the POSIX scheduling model, VxWorks must be
configured with the POSIX thread scheduler, and the pthreads in question must be
run in processes (RTPs). See 5.12.5 POSIX Threads Scheduler, p.282.
5.12.5 POSIX Threads Scheduler
The POSIX thread scheduler can be used to schedule both pthreads and tasks in a
VxWorks system. Note that the purpose of the POSIX thread scheduler is to
provide POSIX scheduling support for pthreads running in processes. There is no
reason to use it in a system that does not require this support (kernel-only systems,
or systems with processes but without pthreads).
The POSIX thread scheduler is required for running pthreads in processes, where it
provides compliance with POSIX 1003.1 for pthread scheduling (including
concurrent scheduling policies). If VxWorks is not configured with the POSIX
thread scheduler, pthreads cannot be created in processes.
NOTE: The POSIX priority numbering scheme is the inverse of the VxWorks
scheme, so references to a given priority level or same level in comparisons of these
schemes refer to functionally equivalent priority levels, and not to priority
numbers. For more information about the numbering schemes see 5.12.2 POSIX
and VxWorks Priority Numbering, p.279.
Scheduling in the Kernel
The POSIX thread scheduler schedules kernel tasks and kernel pthreads in the same
manner as the traditional VxWorks task scheduler. See 4.3 Task Scheduling, p.166
for information about the traditional scheduler and how it works with VxWorks
tasks, and 5.12.4 VxWorks Traditional Scheduler, p.281 for information about how
282
5 POSIX Facilities
5.12 POSIX and VxWorks Scheduling
POSIX scheduling policies correspond to the traditional VxWorks scheduling
policies.
Scheduling in Processes
When VxWorks is configured with the POSIX thread scheduler, tasks executing in
processes are scheduled according to system-wide default scheduling policy. On
the other hand, pthreads executing in processes are scheduled according to POSIX
1003.1. Scheduling policies can be assigned to each pthread and changed
dynamically. The scheduling policies are as follows:
■
SCHED_FIFO is a preemptive priority scheduling policy. For a given priority
level, pthreads scheduled with this policy are handled as peers of the VxWorks
tasks at the same level. There is a slight difference in how pthreads and tasks
are handled if their priorities are lowered (for more information; see Differences
in Re-Queuing Pthreads and Tasks With Lowered Priorities, p.285).
■
SCHED_RR is a per-priority round-robin scheduling policy. For a given
priority level, all pthreads scheduled with this policy are given the same time
of execution (time-slice) before giving up the CPU.
■
SCHED_SPORADIC is a policy used for aperiodic activities, which ensures that
the pthreads associated with the policy are served periodically at a high
priority for a bounded amount of time, and a low background priority at all
other times.
■
SCHED_OTHER corresponds to the scheduling policy currently in use for
VxWorks tasks, which is either preemptive priority or round-robin. Pthreads
scheduled with this policy are submitted to the system's global scheduling
policy, exactly like VxWorks tasks or kernel pthreads.
Note the following with regard to the VxWorks implementation of the
SCHED_SPORADIC policy:
■
The system periodic clock is used for time accounting.
■
Dynamically changing the scheduling policy to SCHED_SPORADIC is not
supported; however, dynamically changing the policy from
SCHED_SPORADIC to another policy is supported.
■
VxWorks does not impose an upper limit on the maximum number of
replenishment events with the SS_REPL_MAX macro. A default of 40 events is
set with the sched_ss_max_repl field of the thread attribute structure, which
can be changed.
283
5
VxWorks
Kernel Programmer's Guide, 6.6
Configuring VxWorks with the POSIX Thread Scheduler
To configure VxWorks with the POSIX thread scheduler, add the
INCLUDE_POSIX_PTHREAD_SCHEDULER component to the kernel.
Note that only the SCHED_FIFO, SCHED_RR, and SCHED_OTHER scheduling
policies are provided with the INCLUDE_POSIX_PTHREAD_SCHEDULER
component. For the SCHED_SPORADIC scheduling policy, the
INCLUDE_PX_SCHED_SPORADIC_POLICY component must be included as well.
The bundle BUNDLE_RTP_POSIX_PSE52 includes the
INCLUDE_PX_SCHED_SPORADIC_POLICY component as well as the
INCLUDE_POSIX_PTHREAD_SCHEDULER component.
The configuration parameter POSIX_PTHREAD_RR_TIMESLICE may be used to
configure the default time slicing interval for pthreads started with the SCHED_RR
policy. To modify the time slice at run time, call kernelTimeSlice( ) with a
different time slice value. The new time slice value only affects pthreads created
after the kernelTimeSlice( ) call.
NOTE: The INCLUDE_POSIX_PTHREAD_SCHEDULER component is a standalone
component. It is not dependent on any other POSIX components nor is it
automatically included with any other components.
The POSIX thread scheduler must be added explicitly with either the
INCLUDE_POSIX_PTHREAD_SCHEDULER component or the
BUNDLE_RTP_POSIX_PSE52 bundle.
The POSIX thread scheduler component is independent because it is intended to
be used only with pthreads in processes; kernel-only systems that use pthreads,
have no need to change from the default VxWorks traditional scheduler.
Caveats About Scheduling Behavior with the POSIX Thread Scheduler
Using the POSIX thread scheduler involves a few complexities that should be
taken into account when designing your system. Care should be taken with regard
to the following:
■
Using both round-robin and priority-based preemptive scheduling policies.
■
Running pthreads with the individual SCHED_OTHER policy.
■
Differences in re-queuing pthreads and tasks with lowered priorities.
■
Backwards compatibility issues for POSIX applications designed for the
VxWorks traditional scheduler.
284
5 POSIX Facilities
5.12 POSIX and VxWorks Scheduling
Using both Round-Robin and Priority-Based Preemptive Policies
Using a combination of round-robin and priority-based preemptive policies for
tasks and pthreads of the same priority level can lead to task or pthread CPU
starvation for the entities running with the round-robin policy.
When VxWorks is running with round-robin scheduling as the system default,
tasks may not run with their expected time slice if there are pthreads running at
the same priority level with the concurrent (individual) SCHED_FIFO policy. This
is because one of the pthreads may monopolize the CPU and starve the tasks. Even
if the usurper pthread is preempted, it stays at the head of its priority lists (as
POSIX mandates), and continues to monopolize the CPU when that priority level
can run again. Pthreads scheduled with the SCHED_RR or SCHED_OTHER policy
are at the same disadvantage as the tasks scheduled with the round-robin policy.
Similarly, when VxWorks is running with preemptive scheduling as the system
default, tasks may starve pthreads with the same priority level if the latter have the
concurrent (individual) SCHED_RR policy.
Running pthreads with the Concurrent SCHED_OTHER Policy
Pthreads created with the concurrent (individual) SCHED_OTHER policy behave
the same as the system-wide default scheduling policy, which means that:
■
If the system default is currently priority-based preemptive scheduling, the
SCHED_OTHER pthreads run with the preemptive policy.
■
If the system default is currently round-robin scheduling, the SCHED_OTHER
pthreads run with the round-robin policy.
While changing the default system policy from priority-based preemptive
scheduling to round-robin scheduling (or the opposite) changes the effective
scheduling policy for pthreads created with SCHED_OTHER, it has no effect on
pthreads created with SCHED_RR or SCHED_FIFO.
Differences in Re-Queuing Pthreads and Tasks With Lowered Priorities
The POSIX thread scheduler re-queues pthreads that have had their priority
lowered differently than it re-queues tasks that have had their priority lowered.
The difference is as follows:
■
When the priority of a pthread is lowered (with the pthread_setschedprio( )
routine), the POSIX thread scheduler places it at the head of the priority list.
■
When the priority of a task is lowered (with the taskPrioritySet( ) routine), the
POSIX thread scheduler places it at the tail of the priority list—which is the
same as what the traditional VxWorks scheduler would do.
285
5
VxWorks
Kernel Programmer's Guide, 6.6
What this means is that lowering the priority of a task and a pthread may have a
different effect on when they will run (if there are other tasks or pthreads in their
priority list). For example, if a task and a pthread each have their priority lowered
to effectively the same level, the pthread will be at the head of the priority list and
the task at the end. The pthread will run before any other pthreads or tasks at this
level, and the task after any other pthreads or tasks.
Backwards Compatibility Issues for Applications
Using the POSIX thread scheduler changes the behavior of POSIX applications that
were written to run with the traditional VxWorks scheduler. For existing POSIX
applications that require backward-compatibility, the scheduling policy can be
changed to SCHED_OTHER for all pthreads. This causes their policy to default to
the active VxWorks task scheduling policy (as was the case before the introduction
of the POSIX thread scheduler).
5.12.6 POSIX Scheduling Routines
The POSIX 1003.1b scheduling routines provided by the schedPxLib library for
VxWorks are described in Table 5-16.
Table 5-16
POSIX Scheduling Routines
Routine
Description
sched_get_priority_max( ) Gets the maximum pthread priority.
sched_get_priority_min( ) Gets the minimum pthread priority.
sched_rr_get_interval( )
If round-robin scheduling is in effect, gets the time
slice length.
sched_yield( )
Relinquishes the CPU.
For more information about these routines, see the schedPxLib API reference.
286
5 POSIX Facilities
5.12 POSIX and VxWorks Scheduling
NOTE: Several scheduling routines that were provided with schedPxLib for
VxWorks 5.x and early versions of VxWorks 6.x are not POSIX compliant, and are
maintained only for backward compatibility in the kernel. The use of these
routines is deprecated: sched_setparam( ), sched_getparam( ),
sched_setscheduler( ), and sched_getscheduler( ).
The native VxWorks routines taskPrioritySet( ) and taskPriorityGet( ) should be
used for task priorities. The POSIX routines pthread_setschedparam( ) and
pthread_getschedparam( ) should be used for pthread priorities.
For information about changing the default system scheduling policy, see
5.12.3 Default Scheduling Policy, p.280. For information about concurrent
scheduling policies, see 5.12.5 POSIX Threads Scheduler, p.282.
Note that the POSIX priority numbering scheme is the inverse of the VxWorks
scheme. For more information see 5.12.2 POSIX and VxWorks Priority Numbering,
p.279.
To include the schedPxLib library in the system, configure VxWorks with the
INCLUDE_POSIX_SCHED component.
5.12.7 Getting Scheduling Parameters: Priority Limits and Time Slice
The routines sched_get_priority_max( ) and sched_get_priority_min( ) return the
maximum and minimum possible POSIX priority, respectively.
If round-robin scheduling is enabled, you can use sched_rr_get_interval( ) to
determine the length of the current time-slice interval. This routine takes as an
argument a pointer to a timespec structure (defined in time.h), and writes the
number of seconds and nanoseconds per time slice to the appropriate elements of
that structure.
287
5
VxWorks
Kernel Programmer's Guide, 6.6
Example 5-5
Getting the POSIX Round-Robin Time Slice
/* The following example checks that round-robin scheduling is enabled,
* gets the length of the time slice, and then displays the time slice.
*/
/* includes */
#include
#include
STATUS rrgetintervalTest (void)
{
struct timespec slice;
/* turn on round robin */
kernelTimeSlice (30);
if (sched_rr_get_interval (0, &slice) == ERROR)
{
printf ("get-interval test failed\n");
return (ERROR);
}
printf ("time slice is %l seconds and %l nanoseconds\n",
slice.tv_sec, slice.tv_nsec);
return (OK);
}
288
5 POSIX Facilities
5.13 POSIX Semaphores
5.13 POSIX Semaphores
POSIX defines both named and unnamed semaphores, which have the same
properties, but which use slightly different interfaces. The POSIX semaphore
library provides routines for creating, opening, and destroying both named and
unnamed semaphores.
When opening a named semaphore, you assign a symbolic name,2 which the other
named-semaphore routines accept as an argument. The POSIX semaphore
routines provided by semPxLib are shown in Table 5-17.
Table 5-17
POSIX Semaphore Routines
Routine
Description
sem_init( )
Initializes an unnamed semaphore.
sem_destroy( )
Destroys an unnamed semaphore.
sem_open( )
Initializes/opens a named semaphore.
sem_close( )
Closes a named semaphore.
sem_unlink( )
Removes a named semaphore.
sem_wait( )
Lock a semaphore.
sem_trywait( )
Lock a semaphore only if it is not already locked.
sem_post( )
Unlock a semaphore.
sem_getvalue( )
Get the value of a semaphore.
sem_timedwait( )
Lock a semaphore with a timeout.
To include the POSIX semPxLib library semaphore routines in the system,
configure VxWorks with the INCLUDE_POSIX_SEM component.
2. Some operating systems, such as UNIX, require symbolic names for objects that are to be
shared among processes. This is because processes do not normally share memory in such
operating systems. In VxWorks, named semaphores can be used to share semaphores
between real-time processes. In the VxWorks kernel there is no need for named semaphores,
because all kernel objects have unique identifiers. However, using named semaphores of
the POSIX variety provides a convenient way of determining the object’s ID.
289
5
VxWorks
Kernel Programmer's Guide, 6.6
VxWorks also provides semPxLibInit( ), a non-POSIX (kernel-only) routine that
initializes the kernel’s POSIX semaphore library. It is called by default at boot time
when POSIX semaphores have been included in the VxWorks configuration.
5.13.1 Comparison of POSIX and VxWorks Semaphores
POSIX semaphores are counting semaphores; that is, they keep track of the number
of times they are given. The VxWorks semaphore mechanism is similar to that
specified by POSIX, except that VxWorks semaphores offer these additional
features:
■
■
■
■
■
■
priority inheritance
task-deletion safety
the ability for a single task to take a semaphore multiple times
ownership of mutual-exclusion semaphores
semaphore timeouts
queuing mechanism options
When these features are important, VxWorks semaphores are preferable to POSIX
semaphores. (For information about these features, see 4. Multitasking.)
The POSIX terms wait (or lock) and post (or unlock) correspond to the VxWorks
terms take and give, respectively. The POSIX routines for locking, unlocking, and
getting the value of semaphores are used for both named and unnamed
semaphores.
The routines sem_init( ) and sem_destroy( ) are used for initializing and
destroying unnamed semaphores only. The sem_destroy( ) call terminates an
unnamed semaphore and deallocates all associated memory.
The routines sem_open( ), sem_unlink( ), and sem_close( ) are for opening and
closing (destroying) named semaphores only. The combination of sem_close( )
and sem_unlink( ) has the same effect for named semaphores as sem_destroy( )
does for unnamed semaphores. That is, it terminates the semaphore and
deallocates the associated memory.
!
WARNING: When deleting semaphores, particularly mutual-exclusion
semaphores, avoid deleting a semaphore still required by another task. Do not
delete a semaphore unless the deleting task first succeeds in locking that
semaphore. Similarly for named semaphores, close semaphores only from the
same task that opens them.
290
5 POSIX Facilities
5.13 POSIX Semaphores
5.13.2 Using Unnamed Semaphores
When using unnamed semaphores, typically one task allocates memory for the
semaphore and initializes it. A semaphore is represented with the data structure
sem_t, defined in semaphore.h. The semaphore initialization routine, sem_init( ),
lets you specify the initial value.
Once the semaphore is initialized, any task can use the semaphore by locking it
with sem_wait( ) (blocking) or sem_trywait( ) (non-blocking), and unlocking it
with sem_post( ).
Semaphores can be used for both synchronization and exclusion. Thus, when a
semaphore is used for synchronization, it is typically initialized to zero (locked).
The task waiting to be synchronized blocks on a sem_wait( ). The task doing the
synchronizing unlocks the semaphore using sem_post( ). If the task that is blocked
on the semaphore is the only one waiting for that semaphore, the task unblocks
and becomes ready to run. If other tasks are blocked on the semaphore, the task
with the highest priority is unblocked.
When a semaphore is used for mutual exclusion, it is typically initialized to a value
greater than zero, meaning that the resource is available. Therefore, the first task
to lock the semaphore does so without blocking, setting the semaphore to 0
(locked). Subsequent tasks will block until the semaphore is released. As with the
previous scenario, when the semaphore is released the task with the highest
priority is unblocked.
291
5
VxWorks
Kernel Programmer's Guide, 6.6
Example 5-6
POSIX Unnamed Semaphores
/*
* This example uses unnamed semaphores to synchronize an action between the
* calling task and a task that it spawns (tSyncTask). To run from the
shell,
* spawn as a task:
*
* -> sp unnameSem
*/
/* includes */
#include
#include
/* forward declarations */
void syncTask (sem_t * pSem);
/************************************************************************
* unnameSem - test case for unamed semaphores
*
* This routine tests unamed semaphores.
*
* RETURNS: N/A
*
* ERRNOS: N/A
*/
void unnameSem (void)
{
sem_t * pSem;
/* reserve memory for semaphore */
pSem = (sem_t *) malloc (sizeof (sem_t));
if (pSem == NULL)
{
printf ("pSem allocation failed\n");
return;
}
/* initialize semaphore to unavailable */
if (sem_init (pSem, 0, 0) == -1)
{
printf ("unnameSem: sem_init failed\n");
free ((char *) pSem);
return;
}
/* create sync task */
printf ("unnameSem: spawning task...\n");
292
5 POSIX Facilities
5.13 POSIX Semaphores
if (taskSpawn ("tSyncTask", 90, 0, 2000, syncTask, pSem) == ERROR)
{
printf ("Failed to spawn tSyncTask\n");
sem_destroy (pSem);
free ((char *) pSem);
return;
}
/* do something useful to synchronize with syncTask */
/* unlock sem */
printf ("unnameSem: posting semaphore - synchronizing action\n");
if (sem_post (pSem) == -1)
{
printf ("unnameSem: posting semaphore failed\n");
sem_destroy (pSem);
free ((char *) pSem);
return;
}
/* all done - destroy semaphore */
if (sem_destroy (pSem) == -1)
{
printf ("unnameSem: sem_destroy failed\n");
return;
}
free ((char *) pSem);
}
void syncTask
(
sem_t * pSem
)
{
/* wait for synchronization from unnameSem */
if (sem_wait (pSem) == -1)
{
printf ("syncTask: sem_wait failed \n");
return;
}
else
printf ("syncTask: sem locked; doing sync’ed action...\n");
/* do something useful here */
}
293
5
VxWorks
Kernel Programmer's Guide, 6.6
5.13.3 Using Named Semaphores
The sem_open( ) routine either opens a named semaphore that already exists or,
as an option, creates a new semaphore. You can specify which of these possibilities
you want by combining the following flag values:
O_CREAT
Create the semaphore if it does not already exist. If it exists, either fail or open
the semaphore, depending on whether O_EXCL is specified.
O_EXCL
Open the semaphore only if newly created; fail if the semaphore exists.
The results, based on the flags and whether the semaphore accessed already exists,
are shown in Table 5-18.
Table 5-18
Possible Outcomes of Calling sem_open( )
Flag Settings
If Semaphore Exists
If Semaphore Does Not Exist
None
Semaphore is opened.
Routine fails.
O_CREAT
Semaphore is opened.
Semaphore is created.
O_CREAT and O_EXCL
Routine fails.
Semaphore is created.
O_EXCL
Routine fails.
Routine fails.
Once initialized, a semaphore remains usable until explicitly destroyed. Tasks can
explicitly mark a semaphore for destruction at any time, but the system only
destroys the semaphore when no task has the semaphore open.
If VxWorks is configured with INCLUDE_POSIX_SEM_SHOW, you can use show( )
from the shell (with the C interpreter) to display information about a POSIX
semaphore. 3
3. The show( ) routine is not a POSIX routine, nor is it meant to be used programmatically. It
is designed for interactive use with the shell (with the shell’s C interpreter).
294
5 POSIX Facilities
5.13 POSIX Semaphores
This example shows information about the POSIX semaphore mySem with two
tasks blocked and waiting for it:
-> show semId
value = 0 = 0x0
Semaphore name
sem_open() count
Semaphore value
No. of blocked tasks
:mySem
:3
:0
:2
5
Note that show( ) takes the semaphore ID as the argument.
For a group of collaborating tasks to use a named semaphore, one of the tasks first
creates and initializes the semaphore, by calling sem_open( ) with the O_CREAT
flag. Any task that must use the semaphore thereafter, opens it by calling
sem_open( ) with the same name, but without setting O_CREAT. Any task that has
opened the semaphore can use it by locking it with sem_wait( ) (blocking) or
sem_trywait( ) (non-blocking), and then unlocking it with sem_post( ) when the
task is finished with the semaphore.
To remove a semaphore, all tasks using it must first close it with sem_close( ), and
one of the tasks must also unlink it. Unlinking a semaphore with sem_unlink( )
removes the semaphore name from the name table. After the name is removed
from the name table, tasks that currently have the semaphore open can still use it,
but no new tasks can open this semaphore. If a task tries to open the semaphore
without the O_CREAT flag, the operation fails. An unlinked semaphore is deleted
by the system when the last task closes it.
NOTE: POSIX named semaphores may be shared between processes only if their
names start with a / (forward slash) character. They are otherwise private to the
process in which they were created, and cannot be accessed from another process.
See 4.9 Public and Private Objects, p.194.
295
VxWorks
Kernel Programmer's Guide, 6.6
Example 5-7
POSIX Named Semaphores
/*
* In this example, nameSem() creates a task for synchronization. The
* new task, tSyncSemTask, blocks on the semaphore created in nameSem().
* Once the synchronization takes place, both tasks close the semaphore,
* and nameSem() unlinks it. To run this task from the shell, spawn
* nameSem as a task:
*
-> sp nameSem, "myTest"
*/
/* includes */
#include
#include
#include
#include
#include
/* forward declaration */
void syncSemTask (char * name);
/****************************************************************************
*
* nameSem - test program for POSIX semaphores
*
* This routine opens a named semaphore and spawns a task, tSyncSemTask, which
* waits on the named semaphore.
*
* RETURNS: N/A
*
* ERRNO: N/A
*/
void nameSem
(
char * name
)
{
sem_t * semId;
/* create a named semaphore, initialize to 0*/
printf ("nameSem: creating semaphore\n");
if ((semId = sem_open (name, O_CREAT, 0, 0)) == (sem_t *) -1)
{
printf ("nameSem: sem_open failed\n");
return;
}
printf ("nameSem: spawning sync task\n");
if (taskSpawn ("tSyncSemTask", 90, 0, 4000, (FUNCPTR) syncSemTask,
(int) name, 0, 0, 0, 0, 0, 0, 0, 0, 0) == ERROR)
{
printf ("nameSem: unable to spawn tSyncSemTask\n");
sem_close(semId);
296
5 POSIX Facilities
5.13 POSIX Semaphores
return;
}
/* do something useful to synchronize with syncSemTask */
/* give semaphore */
printf ("nameSem: posting semaphore - synchronizing action\n");
if (sem_post (semId) == -1)
{
printf ("nameSem: sem_post failed\n");
sem_close(semId);
return;
}
/* all done */
if (sem_close (semId) == -1)
{
printf ("nameSem: sem_close failed\n");
return;
}
if (sem_unlink (name) == -1)
{
printf ("nameSem: sem_unlink failed\n");
return;
}
printf ("nameSem: closed and unlinked semaphore\n");
}
/****************************************************************************
*
* syncSemTask - waits on a named POSIX semaphore
*
* This routine waits on the named semaphore created by nameSem().
*
* RETURNS: N/A
*
* ERRNO: N/A
*/
void syncSemTask
(
char * name
)
{
sem_t * semId;
/* open semaphore */
printf ("syncSemTask: opening semaphore\n");
if ((semId = sem_open (name, 0)) == (sem_t *) -1)
{
printf ("syncSemTask: sem_open failed\n");
return;
}
297
5
VxWorks
Kernel Programmer's Guide, 6.6
/* block waiting for synchronization from nameSem */
printf ("syncSemTask: attempting to take semaphore...\n");
if (sem_wait (semId) == -1)
{
printf ("syncSemTask: taking sem failed\n");
return;
}
printf ("syncSemTask: has semaphore, doing sync'ed action ...\n");
/* do something useful here */
if (sem_close (semId) == -1)
{
printf ("syncSemTask: sem_close failed\n");
return;
}
}
298
5 POSIX Facilities
5.14 POSIX Message Queues
5.14 POSIX Message Queues
The POSIX message queue routines, provided by mqPxLib, are shown in
Table 5-19.
Table 5-19
POSIX Message Queue Routines
5
Routine
Description
mq_open( )
Opens a message queue.
mq_close( )
Closes a message queue.
mq_unlink( )
Removes a message queue.
mq_send( )
Sends a message to a queue.
mq_receive( )
Gets a message from a queue.
mq_notify( )
Signals a task that a message is waiting on a queue.
mq_setattr( )
Sets a queue attribute.
mq_getattr( )
Gets a queue attribute.
mq_timedsend( )
Sends a message to a queue, with a timeout.
mq_timedreceive( )
Gets a message from a queue, with a timeout.
Note that there are behavioral differences between the kernel and user space
versions of mq_open( ). The kernel version allows for creation of a message queue
for any permission specified by the oflags parameter. The user-space version
complies with the POSIX PSE52 profile, so that after the first call, any subsequent
calls in the same process are only allowed if an equivalent or lower permission is
specified.
For information about the use of permissions with the user-space version of
mq_open( ), see the VxWorks Application Programmer’s Guide: POSIX Facilities.
The VxWorks initialization routine mqPxLibInit( ) initializes the kernel’s POSIX
message queue library (this is a kernel-only routine). It is called automatically at
boot time when the INCLUDE_POSIX_MQ component is part of the system.
For information about the VxWorks message queue library, see the msgQLib API
reference.
299
VxWorks
Kernel Programmer's Guide, 6.6
5.14.1 Comparison of POSIX and VxWorks Message Queues
POSIX message queues are similar to VxWorks message queues, except that POSIX
message queues provide messages with a range of priorities. The differences are
summarized in Table 5-20.
Table 5-20
Message Queue Feature Comparison
Feature
VxWorks Message Queues
POSIX Message Queues
Maximum Message Queue
Levels
1
(specified by
32
(specified by
MSG_PRI_NORMAL | M
SG_PRI_URGENT)
MAX_PRIO_MAX)
Blocked Message Queues
FIFO or priority-based
Priority-based
Received with Timeout
msgQReceive( ) option
mq_timedreceive( )
(user-space only)
Task Notification
With VxWorks message
queue events
mq_notify( )
Close/Unlink Semantics
With msgQOpen library Yes
Send with Timeout
msgQsend( ) option
mq_timesend( )
(user-space only)
5.14.2 POSIX Message Queue Attributes
A POSIX message queue has the following attributes:
■
an optional O_NONBLOCK flag, which prevents a mq_receive( ) call from
being a blocking call if the message queue is empty
■
the maximum number of messages in the message queue
■
the maximum message size
■
the number of messages currently on the queue
Tasks can set or clear the O_NONBLOCK flag using mq_setattr( ), and get the
values of all the attributes using mq_getattr( ). (As allowed by POSIX, this
implementation of message queues makes use of a number of internal flags that
are not public.)
300
5 POSIX Facilities
5.14 POSIX Message Queues
Example 5-8
Setting and Getting Message Queue Attributes
/*
* This example sets the O_NONBLOCK flag and examines message queue
* attributes.
*/
/* includes */
#include
#include
#include
#include
/* defines */
#define MSG_SIZE
16
int attrEx
(
char * name
)
{
mqd_t
struct mq_attr
struct mq_attr
char
int
mqPXId;
attr;
oldAttr;
buffer[MSG_SIZE];
prio;
5
/* mq descriptor */
/* queue attribute structure */
/* old queue attributes */
/* create read write queue that is blocking */
attr.mq_flags = 0;
attr.mq_maxmsg = 1;
attr.mq_msgsize = 16;
if ((mqPXId = mq_open (name, O_CREAT | O_RDWR , 0, &attr))
== (mqd_t) -1)
return (ERROR);
else
printf ("mq_open with non-block succeeded\n");
/* change attributes on queue - turn on non-blocking */
attr.mq_flags = O_NONBLOCK;
if (mq_setattr (mqPXId, &attr, &oldAttr) == -1)
return (ERROR);
else
{
/* paranoia check - oldAttr should not include non-blocking. */
if (oldAttr.mq_flags & O_NONBLOCK)
return (ERROR);
else
printf ("mq_setattr turning on non-blocking succeeded\n");
}
301
VxWorks
Kernel Programmer's Guide, 6.6
/* try receiving - there are no messages but this shouldn't block */
if (mq_receive (mqPXId, buffer, MSG_SIZE, &prio) == -1)
{
if (errno != EAGAIN)
return (ERROR);
else
printf ("mq_receive with non-blocking didn’t block on empty queue\n");
}
else
return (ERROR);
/* use mq_getattr to verify success */
if (mq_getattr (mqPXId, &oldAttr) == -1)
return (ERROR);
else
{
/* test that we got the values we think we should */
if (!(oldAttr.mq_flags & O_NONBLOCK) || (oldAttr.mq_curmsgs != 0))
return (ERROR);
else
printf ("queue attributes are:\n\tblocking is %s\n\t
message size is: %d\n\t
max messages in queue: %d\n\t
no. of current msgs in queue: %d\n",
oldAttr.mq_flags & O_NONBLOCK ? "on" : "off",
oldAttr.mq_msgsize, oldAttr.mq_maxmsg,
oldAttr.mq_curmsgs);
}
/* clean up - close and unlink mq */
if (mq_unlink (name) == -1)
return (ERROR);
if (mq_close (mqPXId) == -1)
return (ERROR);
return (OK);
}
302
5 POSIX Facilities
5.14 POSIX Message Queues
5.14.3 Displaying Message Queue Attributes
The mqPxShow( ) routine can be used to display information about POSIX
message queues, as illustrated below.
-> mq_open ("mymq4", 0x4201, 0)
value = 8380448 = 0x7fe020
-> mqPxShow 0x7fe020
Message queue name
: mymq4
No. of messages in queue
: 0
Maximum no. of messages
: 16
Maximum message size
: 16
Flags
: O_WRONLY
5
O_NONBLOCK
(0x4001
)
5.14.4 Communicating Through a Message Queue
Before a set of tasks can communicate through a POSIX message queue, one of the
tasks must create the message queue by calling mq_open( ) with the O_CREAT flag
set. Once a message queue is created, other tasks can open that queue by name to
send and receive messages on it. Only the first task opens the queue with the
O_CREAT flag; subsequent tasks can open the queue for receiving only
(O_RDONLY), sending only (O_WRONLY), or both sending and receiving
(O_RDWR).
To put messages on a queue, use mq_send( ). If a task attempts to put a message
on the queue when the queue is full, the task blocks until some other task reads a
message from the queue, making space available. To avoid blocking on
mq_send( ), set O_NONBLOCK when you open the message queue. In that case,
when the queue is full, mq_send( ) returns -1 and sets errno to EAGAIN instead of
pending, allowing you to try again or take other action as appropriate.
One of the arguments to mq_send( ) specifies a message priority. Priorities range
from 0 (lowest priority) to 31 (highest priority).
When a task receives a message using mq_receive( ), the task receives the
highest-priority message currently on the queue. Among multiple messages with
the same priority, the first message placed on the queue is the first received (FIFO
order). If the queue is empty, the task blocks until a message is placed on the
queue.
To avoid pending (blocking) on mq_receive( ), open the message queue with
O_NONBLOCK; in that case, when a task attempts to read from an empty queue,
mq_receive( ) returns -1 and sets errno to EAGAIN.
To close a message queue, call mq_close( ). Closing the queue does not destroy it,
but only asserts that your task is no longer using the queue. To request that the
303
VxWorks
Kernel Programmer's Guide, 6.6
queue be destroyed, call mq_unlink( ). Unlinking a message queue does not
destroy the queue immediately, but it does prevent any further tasks from opening
that queue, by removing the queue name from the name table. Tasks that currently
have the queue open can continue to use it. When the last task closes an unlinked
queue, the queue is destroyed.
NOTE: In VxWorks, a POSIX message queue whose name does not start with a
forward-slash (/) character is considered private to the process that has opened it
and can not be accessed from another process. A message queue whose name
starts with a forward-slash (/) character is a public object, and other processes can
access it (as according to the POSIX standard). See 4.9 Public and Private Objects,
p.194.
304
5 POSIX Facilities
5.14 POSIX Message Queues
Example 5-9
POSIX Message Queues
/*
* In this example, the mqExInit() routine spawns two tasks that
* communicate using the message queue.
* To run this test case on the target shell:
*
* -> sp mqExInit
*/
/* mqEx.h - message example header */
/* defines */
#define MQ_NAME "exampleMessageQueue"
/* forward declarations */
void receiveTask (void);
void sendTask (void);
/* testMQ.c - example using POSIX message queues */
/* includes */
#include
#include
#include
#include
#include
#include
#include
/* defines */
#define HI_PRIO 31
#define MSG_SIZE 16
#define MSG "greetings"
/****************************************************************************
*
* mqExInit - main for message queue send and receive test case
*
* This routine spawns to tasks to perform the message queue send and receive
* test case.
*
* RETURNS: OK, or ERROR
*
* ERRNOS: N/A
*/
305
5
VxWorks
Kernel Programmer's Guide, 6.6
int mqExInit (void)
{
/* create two tasks */
if (taskSpawn ("tRcvTask", 151, 0, 4000, (FUNCPTR) receiveTask,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0) == ERROR)
{
printf ("taskSpawn of tRcvTask failed\n");
return (ERROR);
}
if (taskSpawn ("tSndTask", 152, 0, 4000, (FUNCPTR) sendTask,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0) == ERROR)
{
printf ("taskSpawn of tSendTask failed\n");
return (ERROR);
}
return (OK);
}
/****************************************************************************
*
* receiveTask - receive messages from the message queue
*
* This routine creates a message queue and calls mq_receive() to wait for
* a message arriving in the message queue.
*
* RETURNS: OK, or ERROR
*
* ERRNOS: N/A
*/
void receiveTask (void)
{
mqd_t mqPXId; /* msg queue descriptor */
char msg[MSG_SIZE]; /* msg buffer */
int prio; /* priority of message */
/* open message queue using default attributes */
if ((mqPXId = mq_open (MQ_NAME, O_RDWR |
O_CREAT, 0, NULL)) == (mqd_t) -1)
{
printf ("receiveTask: mq_open failed\n");
return;
}
/* try reading from queue */
if (mq_receive (mqPXId, msg, MSG_SIZE, &prio) == -1)
{
printf ("receiveTask: mq_receive failed\n");
return;
}
else
306
5 POSIX Facilities
5.14 POSIX Message Queues
{
printf ("receiveTask: Msg of priority %d received:\n\t\t%s\n",
prio, msg);
}
}
/****************************************************************************
*
* sendTask - send a message to a message queue
*
* This routine opens an already created message queue and
* calls mq_send() to send a message to the opened message queue.
*
* RETURNS: OK, or ERROR
*
* ERRNOS: N/A
*/
void sendTask (void)
{
mqd_t mqPXId; /* msg queue descriptor */
/* open msg queue; should already exist with default attributes */
if ((mqPXId = mq_open (MQ_NAME, O_RDWR, 0, NULL)) == (mqd_t) -1)
{
printf ("sendTask: mq_open failed\n");
return;
}
/* try writing to queue */
if (mq_send (mqPXId, MSG, sizeof (MSG), HI_PRIO) == -1)
{
printf ("sendTask: mq_send failed\n");
return;
}
else
printf ("sendTask: mq_send succeeded\n");
}
5.14.5 Notification of Message Arrival
A pthread (or task) can use the mq_notify( ) routine to request notification of the
arrival of a message at an empty queue. The pthread can thereby avoid blocking
or polling to wait for a message.
Each queue can register only one pthread for notification at a time. Once a queue
has a pthread to notify, no further attempts to register with mq_notify( ) can
succeed until the notification request is satisfied or cancelled.
307
5
VxWorks
Kernel Programmer's Guide, 6.6
Once a queue sends notification to a pthread, the notification request is satisfied,
and the queue has no further special relationship with that particular pthread; that
is, the queue sends a notification signal only once for each mq_notify( ) request. To
arrange for one specific pthread to continue receiving notification signals, the best
approach is to call mq_notify( ) from the same signal handler that receives the
notification signals.
To cancel a notification request, specify NULL instead of a notification signal. Only
the currently registered pthread can cancel its notification request.
The mq_notify( ) mechanism does not send notification:
Example 5-10
■
When additional messages arrive at a message queue that is not empty. That
is, notification is only sent when a message arrives at an empty message queue.
■
If another pthread was blocked on the queue with mq_receive( ).
■
After a response has been made to the call to mq_notify( ). That is, only one
notification is sent per mq_notify( ) call.
Message Queue Notification
/*
* In this example, a task uses mq_notify() to discover when a message
* has arrived on a previously empty queue. To run this from the shell:
*
* -> ld < mq_notify_test.o
* -> sp exMqNotify, "greetings"
* -> mq_send
*
*/
/* includes */
#include
#include
#include
#include
#include
#include
#include
/* defines */
#define QNAM "PxQ1"
#define MSG_SIZE 64 /* limit on message sizes */
/* forward declarations */
static void exNotificationHandle (int, siginfo_t *, void *);
static void exMqRead (mqd_t);
308
5 POSIX Facilities
5.14 POSIX Message Queues
/****************************************************************************
* exMqNotify - example of how to use mq_notify()
*
* This routine illustrates the use of mq_notify() to request notification
* via signal of new messages in a queue. To simplify the example, a
* single task both sends and receives a message.
*
* RETURNS: 0 on success, or -1
*
* ERRNOS: N/A
*/
int exMqNotify
(
char * pMessage,
/* text for message to self */
int loopCnt
/* number of times to send a msg */
)
{
struct mq_attr attr;
/* queue attribute structure */
struct sigevent sigNotify;
/* to attach notification */
struct sigaction mySigAction; /* to attach signal handler */
mqd_t exMqId;
/* id of message queue */
int cnt = 0;
/* Minor sanity check; avoid exceeding msg buffer */
if (MSG_SIZE <= strlen (pMessage))
{
printf ("exMqNotify: message too long\n");
return (-1);
}
/*
* Install signal handler for the notify signal and fill in
* a sigaction structure and pass it to sigaction(). Because the handler
* needs the siginfo structure as an argument, the SA_SIGINFO flag is
* set in sa_flags.
*/
mySigAction.sa_sigaction = exNotificationHandle;
mySigAction.sa_flags = SA_SIGINFO;
sigemptyset (&mySigAction.sa_mask);
if (sigaction (SIGUSR1, &mySigAction, NULL) == -1)
{
printf ("sigaction failed\n");
return (-1);
}
/*
* Create a message queue - fill in a mq_attr structure with the
* size and no. of messages required, and pass it to mq_open().
*/
attr.mq_flags = 0;
attr.mq_maxmsg = 2;
attr.mq_msgsize = MSG_SIZE;
309
5
VxWorks
Kernel Programmer's Guide, 6.6
if ((exMqId = mq_open (QNAM, O_CREAT | O_RDWR | O_NONBLOCK, 0, &attr))
== (mqd_t) - 1 )
{
printf ("mq_open failed\n");
return (-1);
}
/*
* Set up notification: fill in a sigevent structure and pass it
* to mq_notify(). The queue ID is passed as an argument to the
* signal handler.
*/
sigNotify.sigev_signo = SIGUSR1;
sigNotify.sigev_notify = SIGEV_SIGNAL;
sigNotify.sigev_value.sival_int = (int) exMqId;
if (mq_notify (exMqId, &sigNotify) == -1)
{
printf ("mq_notify failed\n");
return (-1);
}
/*
* We just created the message queue, but it may not be empty;
* a higher-priority task may have placed a message there while
* we were requesting notification. mq_notify() does nothing if
* messages are already in the queue; therefore we try to
* retrieve any messages already in the queue.
*/
exMqRead (exMqId);
/*
* Now we know the queue is empty, so we will receive a signal
* the next time a message arrives.
*
* We send a message, which causes the notify handler to be invoked.
* It is a little silly to have the task that gets the notification
* be the one that puts the messages on the queue, but we do it here
* to simplify the example. A real application would do other work
* instead at this point.
*/
if (mq_send (exMqId, pMessage, 1 + strlen (pMessage), 0) == -1)
{
printf ("mq_send failed\n");
}
/* Cleanup */
if (mq_close (exMqId) == -1)
{
printf ("mq_close failed\n");
return (-1);
310
5 POSIX Facilities
5.14 POSIX Message Queues
}
/* More cleanup */
if (mq_unlink (QNAM) == -1)
{
printf ("mq_unlink failed\n");
return (-1);
}
5
return (0);
}
/****************************************************************************
* exNotificationHandle - handler to read in messages
*
* This routine is a signal handler; it reads in messages from a
* message queue.
*
* RETURNS: N/A
*
* ERRNOS: N/A
*/
static void exNotificationHandle
(
int sig,
/* signal number */
siginfo_t * pInfo,
/* signal information */
void * pSigContext
/* unused (required by posix) */
)
{
struct sigevent sigNotify;
mqd_t exMqId;
/* Get the ID of the message queue out of the siginfo structure. */
exMqId = (mqd_t) pInfo->si_value.sival_int;
/*
* Request notification again; it resets each time
* a notification signal goes out.
*/
sigNotify.sigev_signo = pInfo->si_signo;
sigNotify.sigev_value = pInfo->si_value;
sigNotify.sigev_notify = SIGEV_SIGNAL;
if (mq_notify (exMqId, &sigNotify) == -1)
{
printf ("mq_notify failed\n");
return;
}
311
VxWorks
Kernel Programmer's Guide, 6.6
/* Read in the messages */
exMqRead (exMqId);
}
/****************************************************************************
* exMqRead - read in messages
*
* This small utility routine receives and displays all messages
* currently in a POSIX message queue; assumes queue has O_NONBLOCK.
*
* RETURNS: N/A
*
* ERRNOS: N/A
*/
static void exMqRead
(
mqd_t exMqId
)
{
char msg[MSG_SIZE];
int prio;
/*
* Read in the messages - uses a loop to read in the messages
* because a notification is sent ONLY when a message is sent on
* an EMPTY message queue. There could be multiple msgs if, for
* example, a higher-priority task was sending them. Because the
* message queue was opened with the O_NONBLOCK flag, eventually
* this loop exits with errno set to EAGAIN (meaning we did an
* mq_receive() on an empty message queue).
*/
while (mq_receive (exMqId, msg, MSG_SIZE, &prio) != -1)
{
printf ("exMqRead: mqId (0x%x) received message: %s\n", exMqId, msg);
}
if (errno != EAGAIN)
{
printf ("mq_receive: errno = %d\n", errno);
}
}
312
5 POSIX Facilities
5.15 POSIX Signals
5.15 POSIX Signals
VxWorks provides POSIX signal routines, as well as BSD-compatible routines and
native VxWorks routines in the kernel. For information about these facilities, see
4.18 Signals, p.226.
5
5.16 POSIX Memory Management
The VxWorks kernel provides POSIX memory management support for dynamic
memory allocation with calloc( ), malloc( ), realloc( ), and free( ). For more
information in this regard, see 6.7 Kernel Heap and Memory Partition Management,
p.329.
The kernel also provides the following POSIX memory locking routines: mlock( ),
munlock( ), mlockall( ), and munlockall( ). However, memory mappings in
VxWorks are always memory-resident. This ensures deterministic memory access
for mapped files, but it also means that physical memory is continuously
associated with mappings, until it is unmapped. Therefore, these POSIX memory
locking routines do not do anything, and are provided simply for application
portability.
For information about the POSIX memory management facilities that are available
only in processes, see the VxWorks Application Programmer’s Guide: POSIX Facilities.
313
VxWorks
Kernel Programmer's Guide, 6.6
314
6
Memory Management
Kernel Facilities
6.1 Introduction 316
6.2 Configuring VxWorks With Memory Management Facilities 317
6.3 System Memory Maps 317
6.4 Shell Commands 327
6.5 System RAM Autosizing 327
6.6 Reserved Memory 328
6.7 Kernel Heap and Memory Partition Management 329
6.8 Memory Error Detection 331
6.9 Virtual Memory Management 343
6.10 Additional Memory Protection Features 353
6.11 Processes Without MMU Support 355
315
VxWorks
Kernel Programmer's Guide, 6.6
6.1 Introduction
VxWorks provides memory management facilities for all code that executes in the
kernel, as well as memory management facilities for applications that execute as
real-time processes. This chapter deals primarily with kernel-space memory
management, although it also provides information about what memory maps
look like for systems that include support for processes (and related facilities).
This chapter discusses the following topics:
■
The VxWorks components required for different types of memory
management support.
■
The layout of memory for different configurations of VxWorks.
■
Excluding memory from VxWorks use.
■
Using run-time memory autosizing.
■
The kernel heap and memory partition management facilities that are
available in the kernel.
■
Memory error detection facilities, including instrumentation provided by
VxWorks components and the Wind River compiler.
■
Virtual memory management, both automated and programmatic.
■
Using the real-time process environment without an MMU.
For information about the memory management facilities available to
process-based applications, see VxWorks Application Programmer’s Guide: Memory
Management.
For information about additional error detection facilities useful for debugging
software faults, see 11. Error Detection and Reporting.
NOTE: This chapter provides information about facilities available in the VxWorks
kernel. For information about facilities available to real-time processes, see the
corresponding chapter in the VxWorks Application Programmer’s Guide.
316
6 Memory Management
6.2 Configuring VxWorks With Memory Management Facilities
6.2 Configuring VxWorks With Memory Management Facilities
Information about configuring VxWorks with various memory management
facilities is provided in the context of the discussions of those facilities. See:
■
■
■
■
■
■
■
6.4 Shell Commands, p.327
6.5 System RAM Autosizing, p.327
6.6 Reserved Memory, p.328
6.7 Kernel Heap and Memory Partition Management, p.329
6.8 Memory Error Detection, p.331
6.9 Virtual Memory Management, p.343
6.11 Processes Without MMU Support, p.355
6
6.3 System Memory Maps
This section describes the VxWorks memory map as it appears with different
configurations and run-time activity:
■
A system without process support.
■
A system with process support, but without processes running.
■
A system with process support and two processes running, as well as a shared
library and a shared data region.
In addition, it describes various memory views within a single system.
317
VxWorks
Kernel Programmer's Guide, 6.6
6.3.1 System Memory Map Without Process Support
In a VxWorks system RAM is delimited by:
■
The LOCAL_MEM_LOCAL_ADRS BSP configuration parameter, which defines
the start of the system RAM.
■
The address returned by the routine sysPhysMemTop( ), which is at the top of
system RAM. This address is either determined at run-time if RAM autosizing
is enabled (see 6.5 System RAM Autosizing, p.327). If autosizing is disabled,
then sysPhysMemTop( ) is calculated using the BSP configuration parameter
LOCAL_MEM_SIZE; that is sysPhysMemTop( ) returns
LOCAL_MEM_LOCAL_ADRS + LOCAL_MEM_SIZE.
(LOCAL_MEM_LOCAL_ADRS and LOCAL_MEM_SIZE are configuration
parameters of the INCLUDE_MEMORY_CONFIG component.)
System RAM must be contiguous. For systems without an MMU or with the MMU
disabled, this means that the system RAM must be in contiguous physical
memory. For systems with and MMU enabled, the system RAM must be mapped
contiguously in virtual memory. In the latter case, the physical space may be
non-contiguous for some architectures that do not require an identity mapped
kernel. For the architecture specific requirements, see the VxWorks Architecture
Supplement.
!
CAUTION: The SMP configuration of VxWorks does not support MMU-less
configurations. For information about VxWorks SMP, see 15. VxWorks SMP.
Within system RAM, the elements of a VxWorks system are arranged as follows:
■
Below RAM_LOW_ADRS there is an architecture specific layout of memory
blocks used for saving boot parameters, the system exception message area,
the exception or interrupt vector table, and so on. For specific details, see the
VxWorks Architecture Supplement.
■
The kernel code (text, data, and bss) starting at address RAM_LOW_ADRS.
ROM-resident images are an exception, for which the text segment is located
outside of the system RAM (see 2.4.1 VxWorks Image Types, p.15).
■
The WDB target agent memory pool is located immediately above the kernel
code, if WDB is configured into the system (see 12.6 WDB Target Agent, p.626).
■
The kernel heap follows the WDB memory pool.
■
An optional area of persistent memory.
318
6 Memory Management
6.3 System Memory Maps
■
An optional area of user-reserved memory may be located above the kernel
heap.
Figure 6-1 illustrates a typical memory map for a system without process support.
For a comparable illustration of a system with process support—which means it
has unmapped memory available for processes that have not yet been created—see
Figure 6-2.
Note that the memory pool for the WDB target agent is present only if WDB is
configured into the kernel. Without WDB, the kernel heap starts right above the
end of the kernel BSS ELF segment.
The routine sysMemTop( ) returns the end of the kernel heap area. If both the
user-reserved memory size (USER_RESERVED_MEM) and the persistent memory
size (PM_RESERVED_MEM) are zero, then sysMemTop( ) returns the same value
than sysPhysMemTop( ), and the kernel heap extends to the end of the system
RAM area. For more information about configuring user-reserved memory and
persistent memory. See 6.6 Reserved Memory, p.328 for more information. Also see
6.3.2 System Memory Map with Process Support, p.321.
319
6
VxWorks
Kernel Programmer's Guide, 6.6
Figure 6-1
Fig 5-1: Memory Map of a System Without Process Support
LOCAL_MEM_LOCAL_ADRS
(start of system RAM)
RAM_LOW_ADRS
(start of kernel text)
Kernel Image
(text + rodata + data + bss)
WDB Memory Pool
End of Kernel BSS
(optional)
Start of Kernel Heap
Kernel Heap
(System Memory Partition)
User-Reserved Memory Area
(optional)
ED&R Persistent Memory Area
sysMemTop( ) = sysPhysMemTop( ) USR_RESERVED_MEM PM_RESERVED_MEM
sysPhysMemTop( ) - PM_RESERVED_MEM
(optional)
sysPhysMemTop( )
(end of system RAM)
320
6 Memory Management
6.3 System Memory Maps
6.3.2 System Memory Map with Process Support
Kernel applications have access to the same memory management facilities as
described in 6.3 System Memory Maps, p.317, whether process support is included
or not.
The only difference between the two configurations relates to the size of the kernel
heap. Without process support, the kernel heap extends up to sysMemTop( ).
With process support the kernel heap does not extend up to sysMemTop( ), but
instead uses the KERNEL_HEAP_SIZE parameter (set in the INCLUDE_RTP
component) as its size. This parameter is disregarded if process support is not
included in VxWorks.
By default, KERNEL_HEAP_SIZE is set to two-thirds of the RAM located between
sysMemTop( ) and the end of the kernel code, or the end of the WDB memory pool
when the WDB component is included into the system configuration.
Figure 6-2 illustrates this configuration. The RAM located between sysMemTop( )
and the end of the kernel heap is left unmapped. RAM pages are allocated from
that unmapped RAM area when process, shared library, or shared data region
space must be mapped. For a comparable image of a system without process
support, see Figure 6-1.
321
6
VxWorks
Kernel Programmer's Guide, 6.6
Figure 6-2
Fig 5-2: Memory Map of a System With Process Support
LOCAL_MEM_LOCAL_ADRS
RAM_LOW_ADRS
Kernel Image
(text + rodata + data + bss)
WDB Memory Pool
End of Kernel BSS
(optional)
Kernel Heap
Start of Kernel Heap
KERNEL_HEAP_SIZE
End of Kernel Heap
RAM initially unmapped
(dynamically mapped when
creating RTPs, SLs, or SDs)
User-Reserved Memory Area
(optional)
ED&R Persistent Memory Area
sysMemTop( ) = sysPhysMemTop( ) USR_RESERVED_MEM PM_RESERVED_MEM
sysPhysMemTop( ) - PM_RESERVED_MEM
(optional)
sysPhysMemTop( )
322
6 Memory Management
6.3 System Memory Maps
The default setting of KERNEL_HEAP_SIZE should be adjusted to meet the
requirements of the system.
6.3.3 System Memory Map with Processes Running
A VxWorks system configured for real-time processes may have one or more
applications executing as processes at run-time. It may also have shared libraries
and shared data regions instantiated. The kernel, each of the processes, shared
libraries, and shared data regions occupy a discrete space in virtual memory.
Each VxWorks process has its own region of virtual memory; processes do not
overlap in virtual memory. This flat virtual-memory map provides advantages in
speed, in a programming model that accommodates systems with and without an
MMU, and in debugging applications (see 6.11 Processes Without MMU Support,
p.355).
The virtual space assigned to a process is not necessarily composed of one large
contiguous block of virtual memory. In some cases it will be composed of several
smaller blocks of virtual space which are discontinuous from each other.
Figure 6-3 illustrates the memory map of a system with the kernel areas (RAM and
I/O), two different processes (RTP A and RTP B), as well as one shared library, and
one shared data region.
323
6
VxWorks
Kernel Programmer's Guide, 6.6
Figure 6-3
Fig 5-3: Memory Map of a System With Two Processes
Kernel Code
+
Kernel Heap
+
Kernel Task Stacks
+
Kernel Object Modules
RTP A Code
RTP B Heap
Kernel Memory Space
RTP B Task Stacks
RTP A Memory Space
RTP A Heap
RTP B Memory Space
Shared Library
Shared Library and
Shared Data Memory Space
RTP B Code
RTP A Tasks Stacks
Shared Data Region
I/O Region 1
I/O Region 2
324
6 Memory Management
6.3 System Memory Maps
Each process has its own virtual memory context, defined by its MMU translation
table used to map virtual and physical memory, and other information about each
page of memory. This memory context describes the virtual space that all of the
tasks in a the process can access. In other words, it defines the memory view of a
process.
The kernel space is mapped with supervisor access privilege in the memory
context of each process (but not with user mode privilege). Therefore tasks
executing in a process can access kernel memory space only in system calls, during
which the execution is switched to supervisor mode. (For information about
system calls, see VxWorks Application Programmer’s Guide: Applications and
Processes.)
A shared library or shared data region is mapped into the virtual context of a
process only when the process’ application code opens or creates it, and it
effectively disappears from the process’ memory view when the application closes
or deletes the shared library or shared data region.
Figure 6-4 illustrates the different memory views of a system with two processes
(RTP A and RTP B), a shared library that both RTP A and RTP B opened, as well
as a shared data region that both a kernel application and RTP B opened.
The first memory view corresponds to the memory space accessible by kernel
tasks. The second and third memory views correspond to the memory space
accessible by tasks executing in process A, respectively process B. Note that the
grayed areas are only accessible during system calls.
325
6
VxWorks
Kernel Programmer's Guide, 6.6
Figure 6-4
Fig 5-4: System Memory Views
Kernel
Memory View
Kernel Code
+
Kernel Heap
+
Kernel Task Stacks
+
Kernel Object Modules
RTP A
Memory View
Kernel Code
+
Kernel Heap
+
Kernel Task Stacks
+
Kernel Object Modules
RTP B
Memory View
Kernel Code
+
Kernel Heap
+
Kernel Task Stacks
+
Kernel Object Modules
RTP A Code
RTP B Heap
RTP B Task Stacks
RTP A Heap
Shared Library
Shared Library
RTP B Code
RTP A Tasks Stacks
Shared Data Region
326
Shared Data Region
I/O Region 1
I/O Region 1
I/O Region 1
I/O Region 2
I/O Region 2
I/O Region 2
6 Memory Management
6.4 Shell Commands
Note that on system without an MMU, or with the MMU disabled, there is only
one memory view shared by the kernel and all process tasks. This memory view
corresponds to Figure 6-3. Any task in the system, whether it is a kernel or a task
executing in a process, has access to all the memory: kernel space, I/O regions, any
processes memory, shared libraries, and shared data regions. In other words, such
configurations do not provide any memory protection. For more information, see
6.11 Processes Without MMU Support, p.355.
6
6.4 Shell Commands
The shell’s adrSpaceShow( ) show routine (for the C interpreter) or the adrsp info
command (for the command interpreter) can be used to display an overview of the
address space usage at time of the call. These are included in the kernel with the
INCLUDE_ADR_SPACE_SHOW and INCLUDE_ADR_SPACE_SHELL_CMD
components, respectively.
The rtpMemShow( ) show routine or the rtp meminfo command can be used to
display the private mappings of a process. These are included with the
INCLUDE_RTP_SHOW and INCLUDE_RTP_SHOW_SHELL_CMD components,
respectively.
The kernel mappings can be displayed with the vmContextShow( ) show routine
or the vm context command. These are included with the INCLUDE_VM_SHOW
and INCLUDE_VM_SHOW_SHELL_CMD components, respectively.
6.5 System RAM Autosizing
When RAM autosizing is supported by the BSP, defining the configuration
parameter LOCAL_MEM_AUTOSIZE will enable run time memory sizing. The
default definition state for this parameter and the implementation itself is
BSP-dependent. Check the BSP reference to see if this feature is supported or not.
When autosizing is supported by the BSP and LOCAL_MEM_AUTOSIZE is defined,
the top of system RAM as reported by sysPhysMemTop( ) is the value determined
at run-time.
327
VxWorks
Kernel Programmer's Guide, 6.6
If the LOCAL_MEM_AUTOSIZE is not defined, the top of the system RAM as
reported by sysPhysMemTop( ) is the address calculated as:
(LOCAL_MEM_LOCAL_ADRS + LOCAL_MEM_SIZE)
If the BSP is unable to perform run time memory sizing then a compile time error
should be generated, informing the user of the limitation.
LOCAL_MEM_AUTOSIZE, LOCAL_MEM_LOCAL_ADRS and LOCAL_MEM_SIZE
are parameters of the INCLUDE_MEM_CONFIG component.
6.6 Reserved Memory
Two types of reserved memory can be configured in VxWorks system RAM:
user-reserved memory and persistent memory. Reserved memory is not cleared by
VxWorks at startup or during system operation. Boot loaders may or may not clear
the area; see Boot Loaders and Reserved Memory, p.328.
User-reserved memory, configured with the BSP parameter
USER_RESERVED_MEM, is part of the system RAM that can managed by kernel
applications independently of the kernel heap.
Persistent memory, configured with the parameter PM_RESERVED_MEM, is the
part of system RAM that is used by the error detection and reporting facilities (see
11. Error Detection and Reporting).
For the layout of the user-reserved memory and the persistent memory, see figures
Figure 6-1 and Figure 6-2.
Boot Loaders and Reserved Memory
Boot loaders may or may not clear reserved memory, depending on the
configuration that was used to create them. If the boot loader is built with both
USER_RESERVED_MEM and PM_RESERVED_MEM set to zero, the system RAM is
cleared through the address calculated as:
(LOCAL_MEM_LOCAL_ADRS + LOCAL_MEM_SIZE)
To ensure that reserved memory is not cleared, the boot loader should be created
with the USER_RESERVED_MEM and the PM_RESERVED_MEM parameter set to
the desired sizes; that is, the same values that are used to build the downloaded
VxWorks image.
328
6 Memory Management
6.7 Kernel Heap and Memory Partition Management
For information about VxWorks boot loaders, see 3. Boot Loader.
NOTE: If autosizing of system RAM is enabled, the top of the system RAM
detected at run-time may be different from the address calculated as
LOCAL_MEM_LOCAL_ADRS + LOCAL_MEM_SIZE, resulting in non-identical
location of the memory range not being cleared by the boot loader. For more
information about autosizing, see 6.5 System RAM Autosizing, p.327.
6
6.7 Kernel Heap and Memory Partition Management
VxWorks provides facilities for heap access and memory partition management.
The memLib and memPartLib libraries provide routines to access the kernel heap,
including standard ANSI-compatible routines as well as routines to manipulate
kernel memory partitions. The kernel heap is used by all code running in the
kernel, including kernel libraries and components, kernel applications, and by
processes when executing system calls.
Memory partitions consist of areas of memory that are used for dynamic memory
allocations by applications and kernel components. Memory partitions may be
used to reserve portions of memory for specific applications, or to isolate dynamic
memory usage on an application basis.
The kernel heap is a specific memory partition, which is also referred to as the
system memory partition.
6.7.1 Configuring the Kernel Heap and the Memory Partition Manager
There are two kernel components for configuring the kernel heap and the memory
partition manager. The core functionality for both the kernel heap and memory
partition is provided by the INCLUDE_MEM_MGR_BASIC component (see the
VxWorks API reference for memPartLib). The INCLUDE_MEM_MGR_FULL
component extends the functionality required for a full-featured heap and
memory partition manager (see the VxWorks API reference for memLib).
The kernel heap is automatically created by the system when either one of these
components are included in the VxWorks configuration. The size of the kernel
heap is set as described in 6.3 System Memory Maps, p.317; see Figure 6-1 and
Figure 6-2.
329
VxWorks
Kernel Programmer's Guide, 6.6
Information about allocation statistics in the kernel heap and in kernel memory
partitions can be obtained with the show routines provided with the
INCLUDE_MEM_SHOW component. For more information, see the VxWorks API
reference for memShow.
6.7.2 Basic Heap and Memory Partition Manager
The memPartLib library (INCLUDE_MEM_MGR_BASIC) provides the core
facilities for memory partition support, including some of the standard
ANSI-compatible routines such as malloc( ), and free( ). The core functionality of
memPartLib provides for the following API:
■
Creation and deletion of memory partitions with memPartCreate( ) and
memPartDelete( ).
■
Addition of memory to a specific memory partition with
memPartAddToPool( ), or to the heap with memAddToPool( ).
■
Allocation and freeing of memory blocks from a specific memory partitions
with memPartAlloc( ), memPartAlignedAlloc( ), and memPartFree( ); and
from the heap with malloc( ) and free( ).
6.7.3 Full Heap and Memory Partition Manager
The memLib library (provided by the INCLUDE_MEM_MGR_FULL component)
adds a few more routines to provide a full-featured memory partition and heap
allocator. The features provided in this library are:
■
Allocation of memory aligned to a specific boundary with memalign( ), and
alignment to a page with valloc( ).
■
Reallocation of blocks of memory in a specific partition with
memPartRealloc( ), or in the heap with realloc( ).
■
The ANSI-compatible routines calloc( ), and cfree( ).
■
Obtaining memory partition statistics with routines memPartInfoGet( ) and
memPartFindMax( ), or in the heap with memFindMax( ) and
memInfoGet( ).
■
Built-in error checking. This feature is controlled with the heap and partition
options. Two types of errors can be enabled. The first type, block error, is
detected during block validation in free( ), realloc( ), memPartFree( ) and
memPartRealloc( ). The second type, allocation error, is detected by any of the
330
6 Memory Management
6.8 Memory Error Detection
allocation and re-allocation routines. There are options to enable logging an
error message and/or to suspend the task hitting the error. Setting and getting
error handling options of a specific memory partition can be done with
memPartOptionsSet( ) and memPartOptionsGet( ). The debugging options
for the heap are controlled with memOptionsSet( ) and memOptionGet( ).
Additional heap and memory partition error detection is provided with heap
and partition memory instrumentation (see 6.8.1 Heap and Partition Memory
Instrumentation, p.331).
For more information, see the VxWorks API references for memPartLib and
memLib.
6.8 Memory Error Detection
Support for memory error detection is provided by two optional instrumentation
libraries. The memEdrLib library performs error checks of operations in the kernel
heap and memory partitions in the kernel. The Run-Time Error Checking (RTEC)
feature of the Wind River Compiler can be used to check for additional errors, such
as buffer overruns and underruns, static and automatic variable reference checks.
Errors detected by these facilities are reported by the error detection and reporting
facility, which must, therefore be included in the VxWorks kernel configuration.
See 11. Error Detection and Reporting.
6.8.1 Heap and Partition Memory Instrumentation
To supplement the error detection features built into memLib and memPartLib
(such as valid block checking), components can be added to VxWorks to perform
automatic, programmatic, and interactive error checks on memLib and
memPartLib operations.
The components help detect common programming errors such as double-freeing
an allocated block, freeing or reallocating an invalid pointer, memory leaks. In
addition, with compiler-assisted code instrumentation, they help detect
bounds-check violations, buffer over-runs and under-runs, pointer references to
free memory blocks, pointer references to automatic variables outside the scope of
the variable, and so on. Note that compiler-assisted instrumentation must be used
331
6
VxWorks
Kernel Programmer's Guide, 6.6
in order to track buffer underruns and overruns. For information about compiler
instrumentation, see 6.8.2 Compiler Instrumentation, p.338.
Errors detected by the automatic checks are logged by the error detection and
reporting facility.
Configuring VxWorks with Memory Partition and Heap Instrumentation
To enable the basic level of memory partition and heap instrumentation, the
following components must be included into the kernel configuration:
■
INCLUDE_MEM_EDR, includes the basic memory partition debugging
functionality and instrumentation code.
■
INCLUDE_EDR_ERRLOG, INCLUDE_EDR_POLICIES and
INCLUDE_EDR_SHOW for error detection, reporting, and persistent memory.
For more information see 11. Error Detection and Reporting.
The following component may also be included:
■
INCLUDE_MEM_EDR_SHOW, for enabling the show routines.
In addition, the following parameters of the INCLUDE_MEM_EDR component can
be modified:
MEDR_EXTENDED_ENABLE
Set to TRUE to enable logging trace information for each allocated block, but
at the cost of increased memory used to store entries in the allocation database.
The default setting is FALSE.
MEDR_FILL_FREE_ENABLE
Set to TRUE to enable pattern-filling queued free blocks. This aids detecting
writes into freed buffers. The default setting is FALSE.
MEDR_FREE_QUEUE_LEN
Maximum length of the free queue. When a memory block is freed, instead of
immediately returning it to the partition's memory pool, it is kept in a queue.
This is useful for detecting references to a memory block after it has been freed.
When the queue reaches the maximum length allowed, the blocks are returned
to the respective memory pool in a FIFO order. Queuing is disabled when this
parameter is 0. Default setting for this parameter is 64.
MEDR_BLOCK_GUARD_ENABLE
Enable guard signatures in the front and the end of each allocated block.
Enabling this feature aids in detecting buffer overruns, underruns, and some
heap memory corruption. The default setting is FALSE.
332
6 Memory Management
6.8 Memory Error Detection
MEDR_POOL_SIZE
Set the size of the memory pool used to maintain the memory block database.
Default setting in the kernel is 1MB. The database uses 32 bytes per memory
block without extended information enabled, and 64 bytes per block with
extended information enabled (call stack trace). This pool is allocated from the
kernel heap.
Error Types
During execution, errors are automatically logged when the allocation, free, and
re-allocation functions are called. The following error types are automatically
identified and logged:
■
Allocation returns block address within an already allocated block from the
same partition. This would indicate corruption in the partition data structures.
■
Allocation returns block address that is in the task's stack space. This would
indicate corruption in the partition data structures.
■
Allocation returns block address that is in the kernel's static data section. This
would indicate corruption in the partition data structures.
■
Freeing a pointer that is in the task’s stack space.
■
Freeing memory that was already freed and is still in the free queue.
■
Freeing memory that is in the kernel’s static data section.
■
Freeing memory in a different partition than the one in which it was allocated.
■
Freeing a partial memory block.
■
Freeing a memory block with the guard zone corrupted, when the
MEDR_BLOCK_GUARD_ENABLE environment variable is TRUE.
■
Pattern in a memory block which is in the free queue has been corrupted, when
the MEDR_FILL_FREE_ENABLE environment variable is TRUE.
Shell Commands
The show routines and commands described in Table 6-1 are available for use with
the shell’s C and command interpreters to display information.
333
6
VxWorks
Kernel Programmer's Guide, 6.6
Table 6-1
Shell Commands
C Interpreter
Command Interpreter
Description
edrShow( )
edr show
Displays error records.
memEdrPartShow( )
mem part list
Displays a summary of the
instrumentation
information for memory
partitions in the kernel.
memEdrBlockShow( )
mem block list
Displays information about
allocated blocks. Blocks can
be selected using a
combination of various
querying criteria: partition
ID, block address, allocating
task ID, block type.
memEdrFreeQueueFlush( ) mem queue flush
Flushes the free queue.
When this routine is called,
freeing of all blocks in the
free queue is finalized so
that all corresponding
memory blocks are returned
the free pool of the
respective partition.
memEdrBlockMark( )
Marks or unmarks selected
blocks allocated at the time
of the call. The selection
criteria may include
partition ID and/or
allocating task ID. This
routine can be used to
monitor memory leaks by
displaying information of
unmarked blocks with
memBlockShow( ) or
mem block list.
334
mem block mark
and
mem block unmark
6 Memory Management
6.8 Memory Error Detection
Code Example
The following kernel application code is used to demonstrate various errors
detected with the heap and partition memory instrumentation (line numbers are
included for reference purposes). Its use is illustrated in Shell Session Example,
p.335.
#include
#include
6
void heapErrors (void)
{
char * pChar;
pChar = malloc (24);
free (pChar + 2);
free (pChar);
free (pChar);
pChar = malloc (32);
}
/* free partial block */
/* double-free block */
/* leaked memory */
Shell Session Example
The following shell session is executed with the C interpreter. The sample code
listed above is compiled and linked in the VxWorks kernel (see 2.6.8 Linking Kernel
Application Object Modules with VxWorks, p.64). The kernel must include the
INCLUDE_MEM_EDR and INCLUDE_MEM_EDR_SHOW components. In order to
enable saving call stack information, the parameter MEDR_EXTENDED_ENABLE is
set TRUE. Also, the kernel should be configured with the error detection and
reporting facility, including the show routines, as described in 11.2 Configuring
Error Detection and Reporting Facilities, p.564.
First mark all allocated blocks:
-> memEdrBlockMark
value = 6390 = 0x18f6
Next, clear the error log. This step is optional, and is done only to start with a clean
log:
-> edrClear
value = 0 = 0x0
The kernel application may be started in a new task spawned with the sp( ) utility,
as follows:
-> taskId = sp (heapErrors)
New symbol "taskId" added to kernel symbol table.
Task spawned: id = 0x246d010, name = t1
taskId = 0x2469ed0: value = 38195216 = 0x246d010
335
VxWorks
Kernel Programmer's Guide, 6.6
At this point the application finished execution. The following command lists the
memory blocks allocated, but not freed by the application task. Note that the
listing shows the call stack at the time of the allocation:
-> memEdrBlockShow 0, 0, taskId, 5, 1
Addr
Type
Size
Part ID Task ID Task Name
Trace
-------- ------ -------- -------- -------- ------------ -----------246d7a0 alloc
32
269888 246d010
-t1 heapErrors()
memPartAlloc()
0x001bdc88()
Errors detected while executing the application are logged in persistent memory
region.
336
6 Memory Management
6.8 Memory Error Detection
Display the log using edrShow( ). The first error corresponds to line 9 in the test
code; the second error corresponds to line 12.
-> edrShow
ERROR LOG
=========
Log Size:
Record Size:
Max Records:
CPU Type:
Errors Missed:
Error count:
Boot count:
Generation count:
524288 bytes (128 pages)
4096 bytes
123
0x5a
0 (old) + 0 (recent)
2
20
94
6
==[1/2]==============================================================
Severity/Facility:
NON-FATAL/KERNEL
Boot Cycle:
20
OS Version:
6.0.0
Time:
THU JAN 01 00:00:31 1970 (ticks = 1880)
Task:
"t1" (0x0246d010)
freeing part of allocated memory block
PARTITION: 0x269888
PTR=0x246bea2
BLOCK: allocated at 0x0246bea0, 24 bytes
<<<<>>>>
0x0011d240
0x00111364
0x001c26f8
0x001bdc6c
vxTaskEntry +0x54 :
heapErrors
+0x24 :
memPartFree +0xa4 :
memEdrItemGet+0x588:
heapErrors ()
free ()
0x001bdbb4 ()
0x001bd71c ()
==[2/2]==============================================================
Severity/Facility:
NON-FATAL/KERNEL
Boot Cycle:
20
OS Version:
6.0.0
Time:
THU JAN 01 00:00:31 1970 (ticks = 1880)
Task:
"t1" (0x0246d010)
freeing memory in free list
PARTITION: 0x269888
PTR=0x246bea0
BLOCK: free block at 0x0246bea0, 24 bytes
<<<<>>>>
0x0011d240 vxTaskEntry +0x54 :
0x00111374 heapErrors
+0x34 :
0x001c26f8 memPartFree +0xa4 :
0x001bdc6c memEdrItemGet+0x588:
value = 0 = 0x0
heapErrors ()
free ()
0x001bdbb4 ()
0x001bd71c ()
337
VxWorks
Kernel Programmer's Guide, 6.6
6.8.2 Compiler Instrumentation
Additional errors are detected if the application is compiled using the Run-Time
Error Checking (RTEC) feature of the Wind River Compiler (Diab). The following
flag should be used:
-Xrtc=option
NOTE: This feature is not available with the GNU compiler.
Code compiled with the -Xrtc flag is instrumented for run-time checks such as
pointer reference check and pointer arithmetic validation, standard library
parameter validation, and so on. These instrumentations are supported through
the memory partition run-time error detection library. Table 6-2 lists the -Xrtc
options that are supported.
Note that using the -Xrtc flag without specifying any options is the same as using
them all.
Table 6-2
-Xrtc Options
Option
Description
0x01
register and check static (global) variables
0x02
register and check automatic variables
0x08
pointer reference checks
0x10
pointer arithmetic checks
0x20
pointer increment/decrement checks
0x40
standard function checks; for example memset( ) and bcopy( )
0x80
report source code filename and line number in error logs
338
6 Memory Management
6.8 Memory Error Detection
The errors and warnings detected by the RTEC compile-in instrumentation are
logged by the error detection and reporting facility (see 11. Error Detection and
Reporting). The following error types are identified:
■
■
■
■
■
■
Bounds-check violation for allocated memory blocks.
Bounds-check violation of static (global) variables.
Bounds-check violation of automatic variables.
Reference to a block in the free queue.
Reference to the free part of the task’s stack.
De-referencing a NULL pointer.
6
Configuring VxWorks for RTEC Support
Support for this feature in the kernel is enabled by adding the
INCLUDE_MEM_EDR_RTC component, as well as the components described in
section Configuring VxWorks with Memory Partition and Heap Instrumentation, p.332.
Shell Commands
The compiler provided instrumentation automatically logs errors detected in
applications using the error detection and reporting facility. For a list of the shell
commands available for error logs see 11.4 Displaying and Clearing Error Records,
p.568.
Code Example
Application code built with the RTEC instrumentation has compiler-generated
constructors. To ensure that the constructors are called when a module is
dynamically downloaded, the module must be processed similarly to a C++
application. For example, the following make rule can be used:
TGT_DIR=$(WIND_BASE)/target
%.out : %.c
@ $(RM) $@
$(CC) $(CFLAGS) -Xrtc=0xfb $(OPTION_OBJECT_ONLY) $<
@ $(RM) ctdt_$(BUILD_EXT).c
$(NM) $(basename $@).o | $(MUNCH) > ctdt_$(BUILD_EXT).c
$(MAKE) CC_COMPILER=$(OPTION_DOLLAR_SYMBOLS) ctdt_$(BUILD_EXT).o
$(LD_PARTIAL) $(LD_PARTIAL_LAST_FLAGS) $(OPTION_OBJECT_NAME)$@ $(basename
$@).o ctdt_$(BUILD_EXT).o
include $(TGT_DIR)/h/make/rules.library
339
VxWorks
Kernel Programmer's Guide, 6.6
The the following application code generates various errors that can be recorded
and displayed (line numbers are included for reference purposes). Its use is
illustrated in Shell Session Example, p.340.
#include
#include
#include
void refErrors ()
{
char name[] = "very_long_name";
char * pChar;
int state[] = { 0, 1, 2, 3 };
int ix = 0;
pChar = malloc (13);
memcpy (pChar, name, strlen (name)); /* bounds check violation */
/* of allocated block */
for (ix = 0; ix < 4; ix++)
state[ix] = state [ix + 1];
/* bounds check violation */
free (pChar);
memcpy (pChar, "another_name", 12);
}
/* reference a free block */
Shell Session Example
The following shell session log is executed with the C interpreter. The sample code
listed above is compiled and linked in the VxWorks kernel (see 2.6.8 Linking Kernel
Application Object Modules with VxWorks, p.64). The kernel must include the
INCLUDE_MEM_EDR and INCLUDE_MEM_EDR_RTC components. Also, the
kernel should be configured with the error detection and reporting facility,
including the show routines, as described in 11.2 Configuring Error Detection and
Reporting Facilities, p.564.
First, clear the error log to start with a clean log:
-> edrClear
value = 0 = 0x0
Start the kernel application in a new task spawned with the sp( ) utility:
-> sp refErrors
Task spawned: id = 0x246d7d0, name = t2
value = 38197200 = 0x246d7d0
At this point the application finished execution. Errors detected while executing
the application are logged in the persistent memory region. Display the log using
340
6 Memory Management
6.8 Memory Error Detection
edrShow( ). In the example below, the log display is interspersed with description
of the errors.
-> edrShow
ERROR LOG
=========
Log Size:
Record Size:
Max Records:
CPU Type:
Errors Missed:
Error count:
Boot count:
Generation count:
524288 bytes (128 pages)
4096 bytes
123
0x5a
0 (old) + 0 (recent)
3
21
97
6
The first error corresponds to line 13 in the test code. A string of length 14 is copied
into a allocated buffer of size 13:
==[1/3]==============================================================
Severity/Facility:
NON-FATAL/KERNEL
Boot Cycle:
21
OS Version:
6.0.0
Time:
THU JAN 01 00:14:22 1970 (ticks = 51738)
Task:
"t2" (0x0246d7d0)
Injection Point:
refErr.c:13
memory block bounds-check violation
PTR=0x246be60 OFFSET=0 SIZE=14
BLOCK: allocated at 0x0246be60, 13 bytes
<<<<>>>>
0x0011d240 vxTaskEntry +0x54 : 0x00111390 ()
0x00111470 refErrors
+0xe4 : __rtc_chk_at ()
0x001bd02c memEdrErrorLog+0x13c: _sigCtxSave ()
341
VxWorks
Kernel Programmer's Guide, 6.6
The second error refers to line 18: the local state array is referenced with index 4.
Since the array has only four elements, the range of valid indexes is 0 to 3:
==[2/3]==============================================================
Severity/Facility:
NON-FATAL/KERNEL
Boot Cycle:
21
OS Version:
6.0.0
Time:
THU JAN 01 00:14:22 1970 (ticks = 51738)
Task:
"t2" (0x0246d7d0)
Injection Point:
refErr.c:18
memory block bounds-check violation
PTR=0x278ba94 OFFSET=16 SIZE=4
BLOCK: automatic at 0x0278ba94, 16 bytes
<<<<>>>>
0x0011d240 vxTaskEntry +0x54 : 0x00111390 ()
0x001114a0 refErrors
+0x114: __rtc_chk_at ()
0x001bd02c memEdrErrorLog+0x13c: _sigCtxSave ()
The last error is caused by the code on line 22. A memory block that has been freed
is being modified:
==[3/3]==============================================================
Severity/Facility:
NON-FATAL/KERNEL
Boot Cycle:
21
OS Version:
6.0.0
Time:
THU JAN 01 00:14:22 1970 (ticks = 51739)
Task:
"t2" (0x0246d7d0)
Injection Point:
refErr.c:22
pointer to free memory block
PTR=0x246be60 OFFSET=0 SIZE=12
BLOCK: free block at 0x0246be60, 13 bytes
<<<<>>>>
0x0011d240 vxTaskEntry +0x54 : 0x00111390 ()
0x00111518 refErrors
+0x18c: __rtc_chk_at ()
0x001bd02c memEdrErrorLog+0x13c: _sigCtxSave ()
342
6 Memory Management
6.9 Virtual Memory Management
6.9 Virtual Memory Management
VxWorks can be configured with an architecture-independent interface to the
CPU’s memory management unit (MMU) to provide virtual memory support.
This support includes the following features:
■
Setting up the kernel memory context at boot time.
■
Mapping pages in virtual space to physical memory.
■
Setting caching attributes on a per-page basis.
■
Setting protection attributes on a per-page basis.
■
Setting a page mapping as valid or invalid.
■
Locking and unlocking TLB entries for pages of memory.
■
Enabling page optimization.
6
The programmable elements of virtual memory (VM) support are provided by the
vmBaseLib library.
NOTE: There are differences in the vmBaseLib library provided for the symmetric
multiprocessor (SMP) and uniprocessor (UP) configurations of VxWorks, and
special guidelines for its use in optimizing SMP applications. For more
information about vmBaseLib and SMP, see vmBaseLib Restrictions, p.709 and
Using vmBaseLib, p.701. For general information about VxWorks SMP and about
migration, see 15. VxWorks SMP and 15.15 Migrating Code to VxWorks SMP, p.702.
When process (RTP) support is included in VxWorks with the INCLUDE_RTP
component, the virtual memory facilities also provide system support for
managing multiple virtual memory contexts, such as creation and deletion of
process memory context.
For information about additional MMU-based memory protection features
beyond basic virtual memory support, see 6.10 Additional Memory Protection
Features, p.353.
Also note that errors (exceptions) generated with the use of virtual memory
features can be detected and managed with additional VxWorks facilities. See
11. Error Detection and Reporting for more information.
343
VxWorks
Kernel Programmer's Guide, 6.6
6.9.1 Configuring Virtual Memory Management
The components listed in Table 6-3 provide basic virtual memory management, as
well as show routines for use from the shell.
Table 6-3
MMU Components
Constant
Description
INCLUDE_MMU_GLOBAL_MAP
Initialize the kernel's global MMU
mappings according to the BSP's
sysPhysMemDesc[ ] table. See
Configuring the Kernel Virtual Memory
Context, p.344.
INCLUDE_MMU_BASIC
Include the vmBaseLib API, which is
used for programmatic management of
virtual memory (see 6.9.2 Managing
Virtual Memory Programmatically, p.346).
INCLUDE_LOCK_TEXT_SECTION
Kernel text TLB locking optimization.
INCLUDE_PAGE_SIZE_OPTIMIZATION Page size optimization for the kernel.
INCLUDE_VM_SHOW
Virtual memory show routines for the
shell C interpreter.
INCLUDE_VM_SHOW_SHELL_CMD
Virtual memory show commands for
the shell command interpreter.
For information about related components see 6.10 Additional Memory Protection
Features, p.353.
Configuring the Kernel Virtual Memory Context
The kernel virtual memory context is created automatically at boot time based on
configuration data provided by the BSP. The primary data is in the
sysPhysMemDesc[ ] table, which is usually defined in the BSP’s sysLib.c file. The
table defines the initial kernel mappings and initial attributes. The entries in this
table are of PHYS_MEM_DESC structure type, which is defined in vmLib.h.
There is usually no need to change the default sysPhysMemDesc[ ] configuration.
However, modification may be required or advisable, for example, when:
344
6 Memory Management
6.9 Virtual Memory Management
■
New driver support or new devices (for example, flash memory) are added to
the system.
■
The protection or cache attributes of certain entries must be changed. For
example, entries for flash memory can be read-only if the content of the flash
is never written from VxWorks. However, if a flash driver such as TrueFFS is
used, the protection attribute has to be set to writable.
■
There are unused entries in the table. In general, it is best to keep only those
entries that actually describe the system, as each entry may require additional
system RAM for page tables (depending on size of the entry, its location
relative to other entries, and architecture-specific MMU parameters). The
larger the memory blocks mapped, the more memory is used for page tables.
The sysPhysMemDesc[ ] table can be modified at run-time. This is useful, for
example, with PCI drivers that can be auto-configured, which means that memory
requirements are detected at run-time. In this case the size and address fields can
be updated programmatically for the corresponding sysPhysMemDesc[ ] entries.
It is important to make such updates before the VM subsystem is initialized by
usrMmuInit( ), for example during execution of sysHwInit( ).
!
CAUTION: The regions of memory defined in sysPhysMemDesc[ ] must be
page-aligned, and must span complete pages. In other words, the first three fields
(virtual address, physical address, and length) of a PHYS_MEM_DESC structure
must all be even multiples of the MMU page size. Specifying elements of
sysPhysMemDesc[ ] that are not page-aligned causes the target to reboot during
initialization. See the VxWorks Architecture Supplement to determine what page size
is supported for the architecture in question.
Configuration Example
This example is based on multiple CPUs using the shared-memory network. A
separate memory board is used for the shared-memory pool. Because this memory
is not mapped by default, it must be added to sysPhysMemDesc[ ] for all the
boards on the network. The memory starts at 0x4000000 and must be made
non-cacheable, as shown in the following code fragment:
345
6
VxWorks
Kernel Programmer's Guide, 6.6
/* shared memory */
{
(VIRT_ADDR) 0x4000000,
/* virtual address */
(PHYS_ADDR) 0x4000000,
/* physical address */
0x20000,
/* length */
/* initial state mask */
MMU_ATTR_VALID_MSK | MMU_ATTR_PROT_MSK | MMU_ATTR_CACHE_MSK,
/* initial state */
MMU_ATTR_VALID | MMU_ATTR_PROT_SUP_READ | MMU_ATTR_PROT_SUP_WRITE |
MMU_ATTR_CACHE_OFF
}
For some architectures, the system RAM (the memory used for the VxWorks
kernel image, kernel heap, and so on) must be identity mapped. This means that
for the corresponding entry in the sysPhysMemDesc[ ] table, the virtual address
must be the same as the physical address. For more information see 6.3 System
Memory Maps, p.317 and the VxWorks Architecture Supplement.
6.9.2 Managing Virtual Memory Programmatically
This section describes the facilities provided for manipulating the MMU
programmatically using low-level routines in vmBaseLib. You can make portions
of memory non-cacheable, write-protect portions of memory, invalidate pages,
lock TLB entries, or optimize the size of memory pages.
For more information about the virtual memory routines, see the VxWorks API
reference for vmBaseLib.
NOTE: There are differences in the vmBaseLib library provided for the symmetric
multiprocessor (SMP) and uniprocessor (UP) configurations of VxWorks, and
special guidelines for its use in optimizing SMP applications. For more
information about vmBaseLib and SMP, see vmBaseLib Restrictions, p.709 and
Using vmBaseLib, p.701. For general information about VxWorks SMP and about
migration, see 15. VxWorks SMP and 15.15 Migrating Code to VxWorks SMP, p.702.
Modifying Page States
Each virtual memory page (typically 4 KB) has a state associated with it. A page
can be valid/invalid, readable, writable, executable, or cacheable/non-cacheable.
The state of a page can be changed with the vmStateSet( ) routine. See Table 6-4
and Table 6-5 for lists of the page state constants and page state masks that can be
used with vmStateSet( ). A page state mask must be used to describe which flags
346
6 Memory Management
6.9 Virtual Memory Management
are being changed. A logical OR operator can be used with states and masks to
define both mapping protection and cache attributes.
Table 6-4
Page State Constants
Constant
Description
MMU_ATTR_VALID
Valid translation
MMU_ATTR_VALID_NOT
Invalid translation
MMU_ATTR_PRO_SUP_READ
Readable memory in kernel mode
MMU_ATTR_PRO_SUP_WRITE
Writable memory in kernel mode
MMU_ATTR_PRO_SUP_EXE
Executable memory in kernel mode
MMU_ATTR_PRO_USR_READ
Readable memory in user mode
MMU_ATTR_PRO_USR_WRITE
Writable memory in user mode
MMU_ATTR_PRO_USR_EXE
Executable memory in user mode
MMU_ATTR_CACHE_OFF
Non-cacheable memory
MMU_ATTR_CACHE_COPYBACK
Cacheable memory, copyback mode
MMU_ATTR_CACHE_WRITETHRU
Cacheable memory, writethrough mode
MMU_ATTR_CACHE_DEFAULT
Default cache mode (equal to either
COPYBACK, WRITETHRU, or
CACHE_OFF, depending on the setting of
USER_D_CACHE_MODE)
MMU_ATTR_CACHE_COHERENCY
Memory coherency is enforced (not
supported on all architectures; for more
information, see the VxWorks Architecture
Supplement)
MMU_ATTR_CACHE_GUARDED
Prevent out-of-order load operations, and
pre-fetches (not supported on all
architectures; for more information, see
the VxWorks Architecture Supplement)
MMU_ATTR_NO_BLOCK
Page attributes can be changed from ISR.
6
347
VxWorks
Kernel Programmer's Guide, 6.6
Table 6-4
Page State Constants (cont’d)
Constant
Description
MMU_ATTR_SPL_0
Optional Architecture Specific States (only
used by some architectures; for more
information, see the VxWorks Architecture
Supplement)
...
MMU_ATTR_SPL_7
Table 6-5
Page State Masks
Constant
Description
MMU_ATTR_VALID_MSK
Modify valid flag
MMU_ATTR_PROT_MSK
Modify protection flags
MMU_ATTR_CACHE_MSK
Modify cache flags
MMU_ATTR_SPL_MSK
Modify architecture specific flags
Not all combinations of protection settings are supported by all CPUs. For
example, many processor types do not provide setting for execute or non-execute
settings. On such processors, readable also means executable.
For information about architecture-specific page states and their combination, see
the VxWorks Architecture Supplement.
Making Memory Non-Writable
Sections of memory can be write-protected using vmStateSet( ) to prevent
inadvertent access. This can be used, for example, to restrict modification of a data
object to a particular routine. If a data object is global but read-only, tasks can read
the object but not modify it. Any task that must modify this object must call the
associated routine. Inside the routine, the data is made writable for the duration of
the routine, and on exit, the memory is set to MMU_ATTR_PROT_SUP_READ.
Nonwritable Memory Example
In this code example, a task calls dataModify( ) to modify the data structure
pointed to by pData. This routine makes the memory writable, modifies the data,
and sets the memory back to nonwritable. If a task subsequently tries to modify the
data without using dataModify( ), a data access exception occurs.
348
6 Memory Management
6.9 Virtual Memory Management
/* privateCode.h - header file to make data writable from routine only */
#define MAX 1024
typedef struct myData
{
char stuff[MAX];
int moreStuff;
} MY_DATA;
/* privateCode.c - uses VM contexts to make data private to a code segment */
#include
#include
#include
#include
#include "privateCode.h"
MY_DATA * pData;
SEM_ID dataSemId;
int pageSize;
/***********************************************************************
*
* initData - allocate memory and make it nonwritable
*
* This routine initializes data and should be called only once.
*
*/
STATUS initData (void)
{
pageSize = vmPageSizeGet();
/* create semaphore to protect data */
dataSemId = semBCreate (SEM_Q_PRIORITY, SEM_EMPTY);
/* allocate memory = to a page */
pData = (MY_DATA *) valloc (pageSize);
/* initialize data and make it read-only */
bzero ((char *) pData, pageSize);
if (vmStateSet (NULL, (VIRT_ADDR) pData, pageSize, MMU_ATTR_PROT_MSK,
MMU_ATTR_PROT_SUP_READ) == ERROR)
{
semGive (dataSemId);
return (ERROR);
}
/* release semaphore */
semGive (dataSemId);
return (OK);
}
/********************************************************************
*
* dataModify - modify data
*
* To modify data, tasks must call this routine, passing a pointer to
* the new data.
* To test from the shell use:
*
-> initData
*
-> sp dataModify
*
-> d pData
*
-> bfill (pdata, 1024, ‘X’)
*/
349
6
VxWorks
Kernel Programmer's Guide, 6.6
STATUS dataModify
(
MY_DATA * pNewData
)
{
/* take semaphore for exclusive access to data */
semTake (dataSemId, WAIT_FOREVER);
/* make memory writable */
if (vmStateSet (NULL, (VIRT_ADDR) pData, pageSize, MMU_ATTR_PROT_MSK,
MMU_ATTR_PROT_SUP_READ | MMU_ATTR_PROT_SUP_WRITE) == ERROR)
{
semGive (dataSemId);
return (ERROR);
}
/* update data*/
bcopy ((char *) pNewData, (char *) pData, sizeof(MY_DATA));
/* make memory not writable */
if (vmStateSet (NULL, (VIRT_ADDR) pData, pageSize, MMU_ATTR_PROT_MSK,
MMU_ATTR_PROT_SUP_READ) == ERROR)
{
semGive (dataSemId);
return (ERROR);
}
semGive (dataSemId);
return (OK);
}
Invalidating Memory Pages
To invalidate memory on a page basis, use vmStateSet( ) as follows:
vmStateSet (NULL, address, len, MMU_ATTR_VALID_MSK, MMU_ATTR_VALID_NOT);
Any access to a mapping made invalid generates an exception whether it is a read
or a write access.
To re-validate the page, use vmStateSet( ) as follows:
vmStateSet (NULL, address, len, MMU_ATTR_VALID_MSK, MMU_ATTR_VALID);
Locking TLB Entries
For some processors it is possible to force individual entries in the Translation
Look-aside Buffer (TLB) to remain permanently in the TLB. When the
architecture-specific MMU library supports this feature, the vmPageLock( )
routine can be used to lock page entries, and vmPageUnlock( ) to unlock page
entries.
350
6 Memory Management
6.9 Virtual Memory Management
The INCLUDE_LOCK_TEXT_SECTION component provides facilities for TLB
locking. When this component is included in VxWorks, the kernel image text
section is automatically locked at system startup.
This feature can be used for performance optimizations in a manner similar to
cache locking. When often-used page entries are locked in the TLB, the number of
TLB misses can be reduced. Note that the number of TLB entries are generally
limited on all processors types, so locking too many entries can result in contention
for the remaining entries that are used dynamically.
For more information, see the VxWorks Architecture Supplement.
Page Size Optimization
For some processors it is possible to enable larger page sizes than the default
(defined by VM_PAGE_SIZE) for large, contiguous memory blocks that have
homogeneous memory attributes. There are several advantages to using such
optimization, including:
■
Reducing the number of page table entries (PTE) needed to map memory,
resulting in less memory used.
■
More efficient TLB entry usage, resulting in fewer TLB misses, therefore
potentially better performance.
Optimization of the entire kernel memory space (including I/O blocks) at startup
can be accomplished by configuring VxWorks with the
INCLUDE_PAGE_SIZE_OPTIMIZATION component.
Page size optimization for specific blocks of memory can be accomplished at
run-time with the vmPageOptimize( ) routine.
De-optimization is performed automatically when necessary. For example, if part
of a memory block that has been optimized is set with different attributes, the large
page is automatically broken up into multiple smaller pages and the new attribute
is set to the requested pages only.
Setting Page States in ISRs
For many types of processors, vmStateSet( ) is a non-blocking routine, and can
therefore be called safely from ISRs. However, it may block in some cases, such as
on processors that support page size optimization (see Page Size Optimization,
p.351).
351
6
VxWorks
Kernel Programmer's Guide, 6.6
To make sure that vmStateSet( ) can be called safely from an ISR for specific pages,
the page must first have the MMU_ATTR_NO_BLOCK attribute set. The following
code example shows how this can be done:
#include
#include
#define DATA_SIZE
0x10000
char * pData;
void someInitFunction ()
{
/* allocate buffer */
pData = (char *) valloc (DATA_SIZE);
/* set no-block attribute for the buffer */
vmStateSet (NULL, (VIRT_ADDR) pData, DATA_SIZE,
MMU_ATTR_SPL_MSK, MMU_ATTR_NO_BLOCK);
}
void someISR ()
{
...
/* now it's safe to set any attribute for the buffer in an ISR */
vmStateSet (NULL, (VIRT_ADDR) pData, DATA_SIZE,
MMU_ATTR_PROT_MSK, MMU_ATTR_SUP_RWX);
...
}
6.9.3 Troubleshooting
The show routines and commands described in Table 6-6 are available to assist
with trouble-shooting virtual memory problems.
352
6 Memory Management
6.10 Additional Memory Protection Features
Table 6-6
Virtual Memory Shell Commands
C Interpreter
Command Interpreter
Description
vmContextShow( )
vm context
Lists information about the entire
process context, including private
mappings and kernel mappings
(for supervisor access), as well as
any shared data contexts attached
to the process.
rtpMemShow( )
rtp meminfo
Lists only the process’ private
mappings.
These routines and commands are provided by the INCLUDE_VM_SHOW,
INCLUDE_VM_SHOW_SHELL_CMD, INCLUDE_RTP_SHOW, and
INCLUDE_RTP_SHOW_SHELL_CMD components.
For more details and usage example of the show routines see the VxWorks shell
references.
6.10 Additional Memory Protection Features
VxWorks provides MMU-based features that supplement basic virtual memory
support to provide a more reliable run-time environment. These additional
memory-protection features are:
■
■
■
■
■
task stack overrun and underrun detection
interrupt stack overrun and underrun detection
non-executable task stacks
text segment write-protection
exception vector table write-protection
For information about basic virtual memory support, see 6.9 Virtual Memory
Management, p.343.
Errors generated with the use of these features can be detected and managed with
additional VxWorks facilities. See 11. Error Detection and Reporting for more
information.
353
6
VxWorks
Kernel Programmer's Guide, 6.6
6.10.1 Configuring VxWorks for Additional Memory Protection
The components listed in Table 6-7 provide additional memory-protection
features. They can be added to VxWorks as a unit with the
INCLUDE_KERNEL_HARDENING component. The individual and composite
components all include the basic virtual memory component
INCLUDE_MMU_BASIC by default.
Table 6-7
Additional Memory Protection Components
Component
Description
INCLUDE_PROTECT_TASK_STACK
Task stack overrun and underrun
protection.
INCLUDE_TASK_STACK_NO_EXEC
Non-executable task stacks.
INCLUDE_PROTECT_TEXT
Text segment write-protection.
INCLUDE_PROTECT_VEC_TABLE
Exception vector table
write-protection and NULL pointer
reference detection.
INCLUDE_PROTECT_INTERRUPT_STACK Interrupt stack overrun and underrun
protection.
Note that protection of the kernel text segment—and the text segments of kernel
modules dynamically loaded into the kernel space—is not provided by default. On
the other hand, the text segment of processes and shared libraries is always
write-protected, whether or not VxWorks is configured with the
INCLUDE_PROTECT_TEXT component. Similarly, the execution stack of a process
task is not affected by the INCLUDE_PROTECT_TASK_STACK or
INCLUDE_TASK_STACK_NO_EXEC components—it is always protected unless
the task is spawned with the taskSpawn( ) option VX_NO_STACK_PROTECT.
6.10.2 Stack Overrun and Underrun Detection
VxWorks can be configured so that guard zones are inserted at the beginning and
end of task execution stacks. For more information, see Task Stack Guard Zones,
p.176.
The operating system can also be configured to insert guard zones at both ends of
the interrupt stack. For more information, see Interrupt Stack Protection, p.243.
354
6 Memory Management
6.11 Processes Without MMU Support
6.10.3 Non-Executable Task Stack
VxWorks can be configured so that task stacks are non-executable. For more
information, see Non-Executable Task Stacks, p.177.
6.10.4 Text Segment Write Protection
All text segments are write-protected when VxWorks is configured with the
INCLUDE_PROTECT_TEXT component. When VxWorks is loaded, all text
segments are write-protected The text segments of any additional object modules
loaded in the kernel space using ld( ) are automatically marked as read-only.
When object modules are loaded, memory that is to be write-protected is allocated
in page-size increments. No additional steps are required to write-protect kernel
application code.
6.10.5 Exception Vector Table Write Protection
When VxWorks is configured with the INCLUDE_PROTECT_VEC_TABLE
component, the exception vector table is write-protected during system
initialization.
The architecture-specific API provided to modify the vector table automatically
write-enables the exception vector table for the duration of the call. For more
information about these APIs, see the VxWorks Architecture Supplement for the
architecture in question.
6.11 Processes Without MMU Support
VxWorks can be configured to provide support for real-time processes on a system
based on a processor without an MMU, or based on a processor with MMU but
with the MMU disabled.
With this configuration, a software simulation-based memory page management
library keeps track of identity mappings only. This means that there is no address
translation, and memory page attributes (protection attributes and cache
attributes) are not supported.
355
6
VxWorks
Kernel Programmer's Guide, 6.6
!
CAUTION: VxWorks SMP does not support MMU-less configurations. For
information about VxWorks SMP, see 15. VxWorks SMP.
The advantages of a configuration without MMU support are that it:
■
Enables the process environment on systems without an MMU. It provides
private namespace for applications, for building applications independently
from the kernel, and for simple migration from systems without an MMU to
those with one.
■
Allows application code be run in non-privileged (user) mode.
■
Under certain conditions, it may provide increased performance by
eliminating overhead of the TLB miss and reload. This assumes, however, that
there is no negative impact due to the changed cache conditions.
The limitations of this configuration are:
■
Depending on the processor type, BSP configuration, drivers and OS facilities
used, disabling the MMU may require disabling the data cache as well.
Disabling the data cache results in significant performance penalty that is
much greater than the benefit derived from avoiding TLB misses.
■
There is no memory protection. That is, memory cannot be write-protected,
and neither the kernel or any process are protected from other processes.
■
The address space is limited to the available system RAM, which is typically
smaller than it would be available on systems with MMU-based address
translation enabled. Because of the smaller address space, a system is more
likely to run out of large contiguous blocks of memory due to fragmentation.
■
Not all processors and target boards can be used with the MMU disabled. For
the requirements of your system see the hardware manual of the board and
processor used.
For information about architecture and processor-specific limitations, see the
VxWorks Architecture Supplement.
Configuring VxWorks With Process Support for Systems Without an MMU
There are no special components needed for the process environment with
software-simulated paging. As with any configurations that provide process
support, the INCLUDE_RTP component must be added to the kernel.
The steps required to enable software-simulated paging are:
356
6 Memory Management
6.11 Processes Without MMU Support
1.
Add the INCLUDE_RTP component to include process support. This
automatically includes all dependent subsystems, among them
INCLUDE_MMU_BASIC.
2.
Change the SW_MMU_ENABLE parameter of the INCLUDE_MMU_BASIC
component to TRUE (the default value is FALSE).
In addition, the following optional configuration steps can reduce the footprint of
the system:
3.
4.
Change the VM_PAGE_SIZE parameter of the INCLUDE_MMU_BASIC
component. The default is architecture-dependent; usually 4K or 8K. Allowed
values are 1K, 2K, 4K, 8K, 16K, 32K, 64K. Typically, a smaller page size results
in finer granularity and therefore more efficient use of the memory space.
However, smaller page size requires more memory needed for keeping track
the mapping information.
Disable stack guard page protection by changing the
TASK_STACK_OVERFLOW_SIZE and TASK_STACK_UNDERFLOW_SIZE
configuration parameters to zero. Without protection provided by an MMU,
stack overflow and underflow cannot be detected, so the guard pages serve no
purpose.
5.
Remove the following components from the VxWorks configuration:
INCLUDE_KERNEL_HARDENING, INCLUDE_PROTECT_TEXT,
INCLUDE_PROTECT_VEC_TABLE, INCLUDE_PROTECT_TASK_STACK,
INCLUDE_TASK_STACK_NO_EXEC, and
INCLUDE_PROTECT_INTERRUPT_STACK. Without an MMU, these features
do not work. Including them only results in unnecessary consumption of
resources.
357
6
VxWorks
Kernel Programmer's Guide, 6.6
358
7
I/O System
7.1 Introduction 360
7.2 Configuring VxWorks With I/O Facilities 362
7.3 Files, Devices, and Drivers 363
7.4 Basic I/O 365
7.5 Buffered I/O: stdio 378
7.6 Other Formatted I/O 380
7.7 Asynchronous Input/Output 381
7.8 Devices in VxWorks 391
7.9 Differences Between VxWorks and Host System I/O 422
7.10 Internal I/O System Structure 423
7.11 PCMCIA 450
7.12 Peripheral Component Interconnect: PCI 450
359
VxWorks
Kernel Programmer's Guide, 6.6
7.1 Introduction
The VxWorks I/O system is designed to present a simple, uniform,
device-independent interface to any kind of device, including:
■
character-oriented devices such as terminals or communications lines
■
random-access block devices such as disks
■
virtual devices such as intertask pipes and sockets
■
monitor and control devices such as digital and analog I/O devices
■
network devices that give access to remote devices
The VxWorks I/O system provides standard C libraries for both basic and
buffered I/O. The basic I/O libraries are UNIX-compatible; the buffered I/O
libraries are ANSI C-compatible.
Internally, the VxWorks I/O system has a unique design that makes it faster and
more flexible than most other I/O systems. These are important attributes in a
real-time system.
The diagram in Figure 7-1 illustrates the relationships between the different
elements of the VxWorks I/O system. All of these elements are discussed in this
chapter, except for file system routines (which are dealt with in 8. Local File
Systems), and the network elements (which are covered in the Wind River Network
Stack for VxWorks 6 Programmer’s Guide).
360
7 I/O System
7.1 Introduction
Figure 7-1
Overview of the VxWorks I/O System
Application
Buffered I/O: stdio
fioLib
fread( )
fwrite( )
fioRead( )
printf( )
sprintf( )
Basic I/O Routines
(device independent)
Network Sockets
Interface
read( )
write( )
send( )
recv( )
Driver Routines
File System Routines
NetWork Stack
Routines
Library Routines
xxRead( )
xxWrite( )
bsdSend( )
bsdReceive( )
tyLib
xxRead( )
xxWrite( )
XBD Routines
Hardware Devices
xyzStrategy( )
Network
Disk Drive
Serial Device
NOTE: The dotted lines in Figure 7-1 indicate that the XBD facility is required for
some file systems, but not others. For example, HRFS and dosFs require XBD,
while ROMFS has its own interface to drivers. See 7.8.8 Extended Block Device
Facility: XBD, p.402.
NOTE: This chapter provides information about facilities available in the VxWorks
kernel. For information about facilities available to real-time processes, see the
corresponding chapter in the VxWorks Application Programmer’s Guide.
361
7
VxWorks
Kernel Programmer's Guide, 6.6
7.2 Configuring VxWorks With I/O Facilities
The components providing the primary VxWorks I/O facilities are as follows:
■
INCLUDE_IO_BASIC—provides basic I/O functionality.
■
INCLUDE_IO_FILE_SYSTEM—provides file system support.
■
INCLUDE_POSIX_DIRLIB—provides POSIX directory utilities.
■
INCLUDE_POSIX_FS—provides POSIX file system APIs.
■
INCLUDE_IO_REMOVABLE—provides support for removable file systems.
■
INCLUDE_IO_POSIX—Provides POSIX I/O support.
■
INCLUDE_IO_RTP—provides I/O support for RTPs.
■
INCLUDE_IO_MISC—miscellaneous IO functions that are no longer
referenced but are provided for backwards compatibility.
The component INCLUDE_IO_SYSTEM is provided for backward compatibility. It
includes all the components listed above.
Components that provide support for additional features are described
throughout this chapter.
362
7 I/O System
7.3 Files, Devices, and Drivers
7.3 Files, Devices, and Drivers
In VxWorks, applications access I/O devices by opening named files. A file can
refer to one of two things:
■
An unstructured raw device such as a serial communications channel or an
intertask pipe.
■
A logical file on a structured, random-access device containing a file system.
Consider the following named files:
7
/usr/myfile
/pipe/mypipe
/tyCo/0
The first refers to a file called myfile, on a disk device called /usr. The second is a
named pipe (by convention, pipe names begin with /pipe). The third refers to a
physical serial channel. However, I/O can be done to or from any of these in the
same way. Within VxWorks, they are all called files, even though they refer to very
different physical objects.
Devices are handled by device drivers. In general, using the I/O system does not
require any further understanding of the implementation of devices and drivers.
Note, however, that the VxWorks I/O system gives drivers considerable flexibility
in the way they handle each specific device. Drivers conform to the conventional
user view presented here, but can differ in the specifics. See 7.8 Devices in VxWorks,
p.391.
Although all I/O is directed at named files, it can be done at two different levels:
basic and buffered. The two differ in the way data is buffered and in the types of calls
that can be made. These two levels are discussed in later sections.
7.3.1 Filenames and the Default Device
A filename is specified as a character string. An unstructured device is specified
with the device name. In the case of file system devices, the device name is
followed by a filename. Thus, the name /tyCo/0 might name a particular serial I/O
channel, and the name DEV1:/file1 indicates the file file1 on the DEV1: device.
When a filename is specified in an I/O call, the I/O system searches for a device
with a name that matches at least an initial substring of the filename. The I/O
function is then directed at this device.
363
VxWorks
Kernel Programmer's Guide, 6.6
If a matching device name cannot be found, then the I/O function is directed at a
default device. You can set this default device to be any device in the system,
including no device at all, in which case failure to match a device name returns an
error. You can obtain the current default path by using ioDefPathGet( ). You can
set the default path by using ioDefPathSet( ).
Non-block devices are named when they are added to the I/O system, usually at
system initialization time. Block devices are named when they are initialized for
use with a specific file system. The VxWorks I/O system imposes no restrictions
on the names given to devices. The I/O system does not interpret device or
filenames in any way, other than during the search for matching device and
filenames.
It is useful to adopt some naming conventions for device and file names: most
device names begin with a forward-slash (/), except non-NFS network devices,
and VxWorks HRFS and dosFs file system devices.
NOTE: To be recognized by the virtual root file system, device names must begin
with a single leading forward-slash, and must not contain any other slash
characters. For more information, see 8.3 Virtual Root File System: VRFS, p.457.
By convention, NFS-based network devices are mounted with names that begin
with a slash. For example:
/usr
Non-NFS network devices are named with the remote machine name followed by
a colon. For example:
host:
The remainder of the name is the filename in the remote directory on the remote
system.
File system devices using dosFs are often named with uppercase letters and digits
followed by a colon. For example:
DEV1:
NOTE: Filenames and directory names on dosFs devices are often separated by
backslashes (\). These can be used interchangeably with forward slashes (/).
!
CAUTION: Because device names are recognized by the I/O system using simple
substring matching, a slash (/ or \) should not be used alone as a device name, nor
should a slash be used as any part of a device name itself.
364
7 I/O System
7.4 Basic I/O
7.4 Basic I/O
Basic I/O is the lowest level of I/O in VxWorks. The basic I/O interface is
source-compatible with the I/O primitives in the standard C library. There are
seven basic I/O calls, shown in Table 7-1.
Table 7-1
Basic I/O Routines
Routine
Description
creat( )
Creates a file.
remove( )
Deletes a file.
open( )
Opens a file (optionally, creates a file if it does not already exist.)
close( )
Closes a file.
read( )
Reads a previously created or opened file.
write( )
Writes to a previously created or opened file.
ioctl( )
Performs special control functions on files.
7
7.4.1 File Descriptors
At the basic I/O level, files are referred to by a file descriptor. A file descriptor is a
small integer returned by a call to open( ) or creat( ). The other basic I/O calls take
a file descriptor as a parameter to specify a file.
File descriptors are not global. The kernel has its own set of file descriptors, and
each process (RTP) has its own set. Tasks within the kernel, or within a specific
process share file descriptors. The only instance in which file descriptors may be
shared across these boundaries, is when one process is a child of another process
or of the kernel and it does not explicitly close a file using the descriptors it inherits
from its parent. (Processes created by kernel tasks share only the spawning kernel
task's standard I/O file descriptors 0, 1 and 2.) For example:
■
If task A and task B are running in process foo, and they each perform a
write( ) on file descriptor 7, they will write to the same file (and device).
■
If process bar is started independently of process foo (it is not foo’s child) and
its tasks X and Y each perform a write( ) on file descriptor 7, they will be
writing to a different file than tasks A and B in process foo.
365
VxWorks
Kernel Programmer's Guide, 6.6
■
If process foobar is started by process foo (it is foo’s child) and its tasks M and
N each perform a write( ) on file descriptor 7, they will be writing to the same
file as tasks A and B in process foo. However, this is only true as long as the
tasks do not close the file. If they close it, and subsequently open file descriptor
7 they will operate on a different file.
When a file is opened, a file descriptor is allocated and returned. When the file is
closed, the file descriptor is deallocated.
File Descriptor Table
The number of file descriptors available in the kernel is defined with the
NUM_FILES configuration macro. This specifies the size of the file descriptor table,
which controls how many file descriptors can be simultaneously in use. The
default number is 50, but it can be changed to suit the needs of the system.
To avoid running out of file descriptors, and encountering errors on file creation,
applications should close any descriptors that are no longer in use.
The size of the file descriptor table for the kernel can also be changed at
programmatically. The rtpIoTableSizeGet( ) routine reads the size of the file
descriptor table and the rtpIoTableSizeSet( ) routine changes it. Note that these
routines can be used with both the kernel and processes (the I/O system treats the
kernel as a special kind of process).
The calling entity—whether kernel or process—can be specified with an
rtpIoTableSizeSet( ) call by setting the first parameter to zero. The new size of the
file descriptor table is set with the second parameter. Note that you can only
increase the size.
7.4.2 Standard Input, Standard Output, and Standard Error
Three file descriptors have special meanings:
■
■
■
0 is used for standard input (stdin).
1 is used for standard output (stdout).
2 is used for standard error output (stderr).
All tasks read their standard input—like getchar( )—from file descriptor 0.
Similarly file descriptor 1 is used for standard output—like printf( ). And file
descriptor 2 is used for outputting error messages. Using these descriptors, you
can manipulate the input and output for many tasks at once by redirecting the files
associated with the descriptors.
366
7 I/O System
7.4 Basic I/O
These standard file descriptors are used to make tasks and modules independent
of their actual I/O assignments. If a module sends its output to standard output
(file descriptor 1), its output can then be redirected to any file or device, without
altering the module.
VxWorks allows two levels of redirection. First, there is a global assignment of the
three standard file descriptors. Second, individual tasks can override the global
assignment of these file descriptors with assignments that apply only to that task.
7.4.3 Standard I/O Redirection
7
When VxWorks is initialized, the global standard I/O file descriptors, stdin (0),
stdout (1) and stderr (2), are set to the system console device file descriptor by
default, which is usually the serial tty device.
Each kernel task uses these global standard I/O file descriptors by default. Thus,
any standard I/O operations like calls to printf( ) and getchar( ) use the global
standard I/O file descriptors.
Standard I/O can be redirected, however, either at the individual task level, or
globally for the kernel.
The standard I/O of a specific task can be changed with the ioTaskStdSet( )
routine. The parameters of this routine are the task ID of the task for which the
change is to be made (0 is used for the calling task itself), the standard file
descriptor to be redirected, and the file descriptor to direct it to. For example, a task
can make the following call to write standard output to the fileFd file descriptor:
ioTaskStdSet (0, 1, fileFd);
The third argument (fileFd in this case) can be any valid open file descriptor. If it
is a file system file, all the task's subsequent standard output, such as that from
printf( ), is written to it.
To reset the task standard I/O back to global standard I/O, the third argument can
be 0, 1, or 2.
The global standard I/O file descriptors can also be changed from the default
setting, which affects all kernel tasks except that have had their task-specific
standard I/O file descriptors changed from the global ones.
Global standard I/O file descriptors are changed by calling ioGlobalStdSet( ). The
parameters to this routine are the standard I/O file descriptor to be redirected, and
the file descriptor to direct it to. For example:
ioGlobalStdSet (1, newFd);
367
VxWorks
Kernel Programmer's Guide, 6.6
This call sets the global standard output to newFd, which can be any valid open
file descriptor. All tasks that do not have their individual task standard output
redirected are affected by this redirection, and all their subsequent standard I/O
output goes to newFd.
The current settings of the global and any task's task standard I/O can be
determined by calling ioGlobalStdGet( ) and ioTaskStdGet( ). For more
information, see the VxWorks API references for these routines.
Issues with Standard I/O Redirection
Be careful with file descriptors used for task standard I/O redirection to ensure
that data corruption does not occur. Before any task’s standard I/O file descriptors
are closed, they should be replaced with new file descriptors with a call to
ioTaskStdSet( ).
If a task’s standard I/O is set with ioTaskStdSet( ), the file descriptor number is
stored in that task’s memory. In some cases, this file descriptor may be closed,
released by some other task or the one that opened it. Once it is released, it may be
reused and opened to track a different file. Should the task holding it as a task
standard I/O descriptor continue to use it for I/O, data corruption is unavoidable.
As an example, consider a task spawned from a telnet or rlogin session. The task
inherits the network session task's standard I/O file descriptors. If the session
exits, the standard I/O file descriptors of the network session task are closed.
However, the spawned task still holds those file descriptors as its task standard
I/O continued with input and output to them. If the closed file descriptors are
recycled and re-used by other open( ) call, however, data corruption results,
perhaps with serious consequences for the system. To prevent this from
happening, all spawned tasks must have their standard I/O file descriptors
redirected before the network session is terminated.
The following example illustrates this scenario, with redirection of a spawned
task’s standard I/O to the global standard I/O from the shell before logout. The
taskspawn( ) call is abbreviated to simplify presentation.
-> taskSpawn "someTask",......
Task spawned: id = 0x52a010, name = t4
value = 5414928 = 0x52a010
-> ioTaskStdSet 0x52a010,0,0
value = 0 = 0x0
-> ioTaskStdSet 0x52a010,1,1
value = 0 = 0x0
-> ioTaskStdSet 0x52a010,2,2
value = 0 = 0x0
-> logout
368
7 I/O System
7.4 Basic I/O
The next example illustrates task standard I/O redirection to other file descriptors.
-> taskSpawn "someTask",......
Task spawned: id = 0x52a010, name = t4
value = 5414928 = 0x52a010
-> ioTaskStdSet 0x52a010,0,someOtherFdx
value = 0 = 0x0
-> ioTaskStdSet 0x52a010,1,someOtherFdy
value = 0 = 0x0
-> ioTaskStdSet 0x52a010,2,someOtherFdz
value = 0 = 0x0
-> logout
7
7.4.4 Open and Close
Before I/O can be performed on a device, a file descriptor must be opened to the
device by invoking the open( ) routine—or creat( ), as discussed in the next section.
The arguments to open( ) are the filename, the type of access, and the mode (file
permissions):
fd = open ("name", flags, mode);
For open( ) calls made in the kernel, the mode parameter can be set to zero if file
permissions do not need to be specified.
The file-access options that can be used with the flags parameter to open( ) are
listed in Table 7-2.
Table 7-2
File Access Options
Flag
Description
O_RDONLY
Open for reading only.
O_WRONLY
Open for writing only.
O_RDWR
Open for reading and writing.
O_CREAT
Create a file if it does not already exist.
O_EXCL
Error on open if the file exists and O_CREAT is also set.
O_SYNC
Write on the file descriptor complete as defined by
synchronized I/O file integrity completion.
369
VxWorks
Kernel Programmer's Guide, 6.6
Table 7-2
!
File Access Options (cont’d)
Flag
Description
O_DSYNC
Write on the file descriptor complete as defined by
synchronized I/O data integrity completion.
O_RSYNC
Read on the file descriptor complete at the same sync level as
O_DSYNC and O_SYNC flags.
O_APPEND
Set the file offset to the end of the file prior to each write, which
guarantees that writes are made at the end of the file. It has no
effect on devices other than the regular file system.
O_NONBLOCK
Non-blocking I/O.
O_NOCTTY
If the named file is a terminal device, don't make it the
controlling terminal for the process.
O_TRUNC
Open with truncation. If the file exists and is a regular file, and
the file is successfully opened, its length is truncated to 0. It has
no effect on devices other than the regular file system.
WARNING: While the third parameter to open( )—mode, for file permissions—is
usually optional for other operating systems, it is required for the VxWorks
implementation of open( ) in the kernel (but not in processes). When the mode
parameter is not appropriate for a given call, it should be set to zero. Note that this
can be an issue when porting software from UNIX to VxWorks.
Note the following special cases with regard to use of the file access and mode (file
permissions) parameters to open( ):
■
In general, you can open only preexisting devices and files with open( ).
However, with NFS network, dosFs, and HRFS devices, you can also create
files with open( ) by OR’ing O_CREAT with one of the other access flags.
■
HRFS directories can be opened with the open( ) routine, but only using the
O_RDONLY flag.
■
With both dosFs and NFS devices, you can use the O_CREAT flag to create a
subdirectory by setting mode to FSTAT_DIR. Other uses of the mode parameter
with dosFs devices are ignored.
370
7 I/O System
7.4 Basic I/O
■
With an HRFS device you cannot use the O_CREAT flag and the FSTAT_DIR
mode option to create a subdirectory. HRFS ignores the mode option and
simply creates a regular file.
■
The netDrv default file system does not support the F_STAT_DIR mode option
or the O_CREAT flag.
■
For NFS devices, the third parameter to open( ) is normally used to specify the
mode of the file. For example:
myFd = open ("fooFile", O_CREAT | O_RDWR, 0644);
■
While HRFS supports setting the permission mode for a file, it is not used by
the VxWorks operating system.
■
Files can be opened with the O_SYNC flag, indicating that each write should be
immediately written to the backing media. This flag is currently supported by
the dosFs file system, and includes synchronizing the FAT and the directory
entries.
■
The O_SYNC flag has no effect with HRFS because file system is always
synchronous. HRFS updates files as though the O_SYNC flag were set.
NOTE: Drivers or file systems may or may not honor the flag values or the mode
values. A file opened with O_RDONLY mode may in fact be writable if the driver
allows it. Consult the driver or file system information for specifics.
See the VxWorks file system API references for more information about the
features that each file system supports.
The open( ) routine, if successful, returns a file descriptor. This file descriptor is
then used in subsequent I/O calls to specify that file. The file descriptor is an
identifier that is not task specific; that is, it is shared by all tasks within the memory
space. Within a given process or the kernel, therefore, one task can open a file and
any other task can then use the file descriptor. The file descriptor remains valid
until close( ) is invoked with that file descriptor, as follows:
close (fd);
At that point, I/O to the file is flushed (completely written out) and the file
descriptor can no longer be used by any task within the process (or kernel).
However, the same file descriptor number can again be assigned by the I/O
system in any subsequent open( ).
Since the kernel only terminates when the system shuts down, there is no situation
analogous to file descriptors being closed automatically when a process
terminates. File descriptors in the kernel can only be closed by direct command.
371
7
VxWorks
Kernel Programmer's Guide, 6.6
7.4.5 Create and Remove
File-oriented devices must be able to create and remove files as well as open
existing files.
The creat( ) routine directs a file-oriented device to make a new file on the device
and return a file descriptor for it. The arguments to creat( ) are similar to those of
open( ) except that the filename specifies the name of the new file rather than an
existing one; the creat( ) routine returns a file descriptor identifying the new file.
fd = creat ("name", flag);
Note that with the HRFS file system the creat( ) routine is POSIX-compliant, and
the second parameter is used to specify file permissions; the file is opened in
O_RDWR mode.
With dosFs, however, the creat( ) routine is not POSIX-compliant and the second
parameter is used for open mode flags.
The remove( ) routine deletes a named file on a file-system device:
remove ("name");
Files should be closed before they are removed.
With non-file-system devices, the creat( ) routine performs the same function as
open( ). The remove( ) routine, however has no effect.
7.4.6 Read and Write
After a file descriptor is obtained by invoking open( ) or creat( ), tasks can read
bytes from a file with read( ) and write bytes to a file with write( ). The arguments
to read( ) are the file descriptor, the address of the buffer to receive input, and the
maximum number of bytes to read:
nBytes = read (fd, &buffer, maxBytes);
The read( ) routine waits for input to be available from the specified file, and
returns the number of bytes actually read. For file-system devices, if the number of
bytes read is less than the number requested, a subsequent read( ) returns 0 (zero),
indicating end-of-file. For non-file-system devices, the number of bytes read can be
less than the number requested even if more bytes are available; a subsequent
read( ) may or may not return 0. In the case of serial devices and TCP sockets,
repeated calls to read( ) are sometimes necessary to read a specific number of
bytes. (See the reference entry for fioRead( ) in fioLib). A return value of ERROR
(-1) indicates an unsuccessful read.
372
7 I/O System
7.4 Basic I/O
The arguments to write( ) are the file descriptor, the address of the buffer that
contains the data to be output, and the number of bytes to be written:
actualBytes = write (fd, &buffer, nBytes);
The write( ) routine ensures that all specified data is at least queued for output
before returning to the caller, though the data may not yet have been written to the
device (this is driver dependent). The write( ) routine returns the number of bytes
written; if the number returned is not equal to the number requested, an error has
occurred.
7
7.4.7 File Truncation
It is sometimes convenient to discard part of the data in a file. After a file is open
for writing, you can use the ftruncate( ) routine to truncate a file to a specified size.
Its arguments are a file descriptor and the desired length of the file in bytes:
status = ftruncate (fd, length);
If it succeeds in truncating the file, ftruncate( ) returns OK.
If the file descriptor refers to a device that cannot be truncated, ftruncate( ) returns
ERROR, and sets errno to EINVAL.
If the size specified is larger than the actual size of the file, the result depends on
the file system. For both dosFs and HRFS, the size of the file is extended to the
specified size; however, for other file systems, ftruncate( ) returns ERROR, and sets
errno to EINVAL (just as if the file descriptor referred to a device that cannot be
truncated).
The ftruncate( ) routine is part of the POSIX 1003.1b standard. It is fully supported
as such by the HRFS. The dosFs implementation is, however, only partially
compliant: creation and modification times are not changed.
Also note that with HRFS the seek position is not modified by truncation, but with
dosFs the seek position is set to the end of the file.
7.4.8 I/O Control
The ioctl( ) routine provides a flexible mechanism for performing I/O functions
that are not performed by the other basic I/O calls. Examples include determining
how many bytes are currently available for input, setting device-specific options,
obtaining information about a file system, and positioning random-access files to
specific byte positions.
373
VxWorks
Kernel Programmer's Guide, 6.6
The arguments to the ioctl( ) routine are the file descriptor, a code that identifies
the control function requested, and an optional function-dependent argument:
result = ioctl (fd, function, arg);
For example, the following call uses the FIOBAUDRATE function to set the baud
rate of a tty device to 9600:
status = ioctl (fd, FIOBAUDRATE, 9600);
The discussion of specific devices in 7.8 Devices in VxWorks, p.391 summarizes the
ioctl( ) functions available for each device. The ioctl( ) control codes are defined in
ioLib.h. For more information, see the reference entries for specific device drivers
or file systems.
7.4.9 Pending on Multiple File Descriptors with select( )
The VxWorks select facility provides a UNIX- and Windows-compatible method
for pending on multiple file descriptors. The library selectLib provides both
task-level support, allowing tasks to wait for multiple devices to become active,
and device driver support, giving drivers the ability to detect tasks that are pended
while waiting for I/O on the device. To use this facility, the header file selectLib.h
must be included in your application code.
Task-level support not only gives tasks the ability to simultaneously wait for I/O
on multiple devices, but it also allows tasks to specify the maximum time to wait
for I/O to become available. An example of using the select facility to pend on
multiple file descriptors is a client-server model, in which the server is servicing
both local and remote clients. The server task uses a pipe to communicate with
local clients and a socket to communicate with remote clients. The server task must
respond to clients as quickly as possible. If the server blocks waiting for a request
on only one of the communication streams, it cannot service requests that come in
on the other stream until it gets a request on the first stream. For example, if the
server blocks waiting for a request to arrive in the socket, it cannot service requests
that arrive in the pipe until a request arrives in the socket to unblock it. This can
delay local tasks waiting to get their requests serviced. The select facility solves this
problem by giving the server task the ability to monitor both the socket and the
pipe and service requests as they come in, regardless of the communication stream
used.
Tasks can block until data becomes available or the device is ready for writing. The
select( ) routine returns when one or more file descriptors are ready or a timeout
has occurred. Using the select( ) routine, a task specifies the file descriptors on
which to wait for activity. Bit fields are used in the select( ) call to specify the read
374
7 I/O System
7.4 Basic I/O
and write file descriptors of interest. When select( ) returns, the bit fields are
modified to reflect the file descriptors that have become available. The macros for
building and manipulating these bit fields are listed in Table 7-3.
Table 7-3
Select Macros
Macro
Description
FD_ZERO
Zeroes all bits.
FD_SET
Sets the bit corresponding to a specified file descriptor.
FD_CLR
Clears a specified bit.
FD_ISSET
Returns non-zero if the specified bit is set; otherwise returns 0.
Applications can use select( ) with any character I/O devices that provide support
for this facility (for example, pipes, serial devices, and sockets).
For information on writing a device driver that supports select( ), see Implementing
select( ), p.441.
375
7
VxWorks
Kernel Programmer's Guide, 6.6
Example 7-1
Using select( )
/* selServer.c - select example
* In this example, a server task uses two pipes: one for normal-priority
* requests, the other for high-priority requests. The server opens both
* pipes and blocks while waiting for data to be available in at least one
* of the pipes.
*/
#include
#include
#include
#define
#define
#define
#define
MAX_FDS 2
MAX_DATA 1024
PIPEHI
"/pipe/highPriority"
PIPENORM "/pipe/normalPriority"
/************************************************************************
* selServer - reads data as it becomes available from two different pipes
*
* Opens two pipe fds, reading from whichever becomes available. The
* server code assumes the pipes have been created from either another
* task or the shell. To test this code from the shell do the following:
* -> ld < selServer.o
* -> pipeDevCreate ("/pipe/highPriority", 5, 1024)
* -> pipeDevCreate ("/pipe/normalPriority", 5, 1024)
* -> fdHi = open
("/pipe/highPriority", 1, 0)
* -> fdNorm = open ("/pipe/normalPriority", 1, 0)
* -> iosFdShow
* -> sp selServer
* -> i
* At this point you should see selServer’s state as pended. You can now
* write to either pipe to make the selServer display your message.
* -> write fdNorm, "Howdy", 6
* -> write fdHi, "Urgent", 7
*/
STATUS selServer (void)
{
struct fd_set readFds;
int
fds[MAX_FDS];
int
width;
int
i;
char
buffer[MAX_DATA];
/*
/*
/*
/*
/*
bit mask of fds to read from */
array of fds on which to pend */
number of fds on which to pend */
index for fd array */
buffer for data that is read */
/* open file descriptors */
if ((fds[0] = open (PIPEHI, O_RDONLY, 0)) == ERROR)
{
close (fds[0]);
return (ERROR);
}
if ((fds[1] = open (PIPENORM, O_RDONLY, 0)) == ERROR)
376
7 I/O System
7.4 Basic I/O
{
close (fds[0]);
close (fds[1]);
return (ERROR);
}
/* loop forever reading data and servicing clients */
FOREVER
{
/* clear bits in read bit mask */
FD_ZERO (&readFds);
7
/* initialize bit mask */
FD_SET (fds[0], &readFds);
FD_SET (fds[1], &readFds);
width = (fds[0] > fds[1]) ? fds[0] : fds[1];
width++;
/* pend, waiting for one or more fds to become ready */
if (select (width, &readFds, NULL, NULL, NULL) == ERROR)
{
close (fds[0]);
close (fds[1]);
return (ERROR);
}
/* step through array and read from fds that are ready */
for (i=0; i< MAX_FDS; i++)
{
/* check if this fd has data to read */
if (FD_ISSET (fds[i], &readFds))
{
/* typically read from fd now that it is ready */
read (fds[i], buffer, MAX_DATA);
/* normally service request, for this example print it */
printf ("SELSERVER Reading from %s: %s\n",
(i == 0) ? PIPEHI : PIPENORM, buffer);
}
}
}
}
377
VxWorks
Kernel Programmer's Guide, 6.6
7.4.10 POSIX File System Routines
The POSIX fsPxLib library provides I/O and file system routines for various file
manipulations. These routines are described in Table 7-4.
Table 7-4
File System Routines
Routine
Description
unlink( )
Unlink a file.
link( )
Link a file.
fsync( )
Synchronize a file.
fdatasync( )
Synchronize the data of a file.
rename( )
Change the name of a file.
fpathconf( )
Determine the current value of a configurable limit.
pathconf( )
Determine the current value of a configurable limit.
access( )
Determine accessibility of a file.
chmod( )
Change the permission mode of a file.
fcntl( )
Perform control functions over open files.
For more information, see the API references for fsPxLib and ioLib.
7.5 Buffered I/O: stdio
The VxWorks I/O library provides a buffered I/O package that is compatible with
the UNIX and Windows stdio package, and provides full ANSI C support.
Configure VxWorks with the ANSI Standard component bundle to provide
buffered I/O support.
378
7 I/O System
7.5 Buffered I/O: stdio
NOTE: The implementation of printf( ), sprintf( ), and sscanf( ), traditionally
considered part of the stdio package, is part of a different package in VxWorks.
These routines are discussed in 7.6 Other Formatted I/O, p.380.
7.5.1 Using stdio
Although the VxWorks I/O system is efficient, some overhead is associated with
each low-level call. First, the I/O system must dispatch from the
device-independent user call (read( ), write( ), and so on) to the driver-specific
routine for that function. Second, most drivers invoke a mutual exclusion or
queuing mechanism to prevent simultaneous requests by multiple users from
interfering with each other.
This overhead is quite small because the VxWorks primitives are fast. However, an
application processing a single character at a time from a file incurs that overhead
for each character if it reads each character with a separate read( ) call:
n = read (fd, &char, 1);
To make this type of I/O more efficient and flexible, the stdio package implements
a buffering scheme in which data is read and written in large chunks and buffered
privately. This buffering is transparent to the application; it is handled
automatically by the stdio routines and macros. To access a file with stdio, a file is
opened with fopen( ) instead of open( ) (many stdio calls begin with the letter f):
fp = fopen ("/usr/foo", "r");
The returned value, a file pointer is a handle for the opened file and its associated
buffers and pointers. A file pointer is actually a pointer to the associated data
structure of type FILE (that is, it is declared as FILE *). By contrast, the low-level I/O
routines identify a file with a file descriptor, which is a small integer. In fact, the
FILE structure pointed to by the file pointer contains the underlying file descriptor
of the open file.
A file descriptor that is already open can be associated subsequently with a FILE
buffer by calling fdopen( ):
fp = fdopen (fd, "r");
After a file is opened with fopen( ), data can be read with fread( ), or a character at
a time with getc( ), and data can be written with fwrite( ), or a character at a time
with putc( ).
The routines and macros to get data into or out of a file are extremely efficient.
They access the buffer with direct pointers that are incremented as data is read or
379
7
VxWorks
Kernel Programmer's Guide, 6.6
written by the user. They pause to call the low-level read or write routines only
when a read buffer is empty or a write buffer is full.
!
WARNING: The stdio buffers and pointers are private to a particular task. They are
not interlocked with semaphores or any other mutual exclusion mechanism,
because this defeats the point of an efficient private buffering scheme. Therefore,
multiple tasks must not perform I/O to the same stdio FILE pointer at the same
time.
The FILE buffer is deallocated when fclose( ) is called.
7.5.2 Standard Input, Standard Output, and Standard Error
As discussed in 7.4 Basic I/O, p.365, there are three special file descriptors (0, 1, and
2) reserved for standard input, standard output, and standard error. Three
corresponding stdio FILE buffers are automatically created when a task uses the
standard file descriptors, stdin, stdout, and stderr, to do buffered I/O to the
standard file descriptors. Each task using the standard I/O file descriptors has its
own stdio FILE buffers. The FILE buffers are deallocated when the task exits.
Additional routines in fioLib provide formatted but unbuffered output. The
routine printErr( ) is analogous to printf( ) but outputs formatted strings to the
standard error file descriptor (2). The routine fdprintf( ) outputs formatted strings
to a specified file descriptor.
7.6 Other Formatted I/O
This section describes additional formatting routines and facilities.
7.6.1 Special Cases: printf( ), sprintf( ), and sscanf( )
The routines printf( ), sprintf( ), and sscanf( ) are generally considered to be part
of the standard stdio package. However, the VxWorks implementation of these
routines, while functionally the same, does not use the stdio package. Instead, it
uses a self-contained, formatted, non-buffered interface to the I/O system in the
library fioLib.
380
7 I/O System
7.7 Asynchronous Input/Output
Note that these routines provide the functionality specified by ANSI; however,
printf( ) is not buffered.
Because these routines are implemented in this way, the full stdio package, which
is optional, can be omitted from a VxWorks configuration without sacrificing their
availability. Applications requiring printf-style output that is buffered can still
accomplish this by calling fprintf( ) explicitly to stdout.
While sscanf( ) is implemented in fioLib and can be used even if stdio is omitted,
the same is not true of scanf( ), which is implemented in the usual way in stdio.
7
7.6.2 Additional Routines: printErr( ) and fdprintf( )
Additional routines in fioLib provide formatted but unbuffered output. The
routine printErr( ) is analogous to printf( ) but outputs formatted strings to the
standard error file descriptor (2). The routine fdprintf( ) outputs formatted strings
to a specified file descriptor.
7.6.3 Message Logging
Another higher-level I/O facility is provided by the library logLib, which allows
formatted messages to be logged without having to do I/O in the current task’s
context, or when there is no task context. The message format and parameters are
sent on a message queue to a logging task, which then formats and outputs the
message. This is useful when messages must be logged from interrupt level, or
when it is desirable not to delay the current task for I/O or use the current task’s
stack for message formatting (which can take up significant stack space). The
message is displayed on the console unless otherwise redirected at system startup
using logInit( ) or dynamically using logFdSet( ).
7.7 Asynchronous Input/Output
Asynchronous Input/Output (AIO) is the ability to perform input and output
operations concurrently with ordinary internal processing. AIO enables you to
de-couple I/O operations from the activities of a particular task when these are
logically independent.
381
VxWorks
Kernel Programmer's Guide, 6.6
The VxWorks AIO implementation meets the specification in the POSIX 1003.1b
standard.
The benefit of AIO is greater processing efficiency: it permits I/O operations to
take place whenever resources are available, rather than making them await
arbitrary events such as the completion of independent operations. AIO eliminates
some of the unnecessary blocking of tasks that is caused by ordinary synchronous
I/O; this decreases contention for resources between input/output and internal
processing, and expedites throughput.
Include AIO in your VxWorks configuration with the INCLUDE_POSIX_AIO and
INCLUDE_POSIX_AIO_SYSDRV components. The second configuration constant
enables the auxiliary AIO system driver, required for asynchronous I/O on all
current VxWorks devices.
7.7.1 The POSIX AIO Routines
The VxWorks library aioPxLib provides POSIX AIO routines. To access a file
asynchronously, open it with the open( ) routine, like any other file. Thereafter, use
the file descriptor returned by open( ) in calls to the AIO routines. The POSIX AIO
routines (and two associated non-POSIX routines) are listed in Table 7-5.
The default VxWorks initialization code calls aioPxLibInit( ) automatically when
the POSIX AIO component is included in VxWorks with INCLUDE_POSIX_AIO.
The aioPxLibInit( ) routine takes one parameter, the maximum number of
lio_listio( ) calls that can be outstanding at one time. By default this parameter is
MAX_LIO_CALLS. When the parameter is 0 (the default), the value is taken from
AIO_CLUST_MAX (defined in
installDir/vxworks-6.x/target/h/private/aioPxLibP.h).
The AIO system driver, aioSysDrv, is initialized by default with the routine
aioSysInit( ) when both INCLUDE_POSIX_AIO and
INCLUDE_POSIX_AIO_SYSDRV are included in VxWorks. The purpose of
aioSysDrv is to provide request queues independent of any particular device
driver, so that you can use any VxWorks device driver with AIO.
382
7 I/O System
7.7 Asynchronous Input/Output
Table 7-5
Asynchronous Input/Output Routines
Function
Description
aioPxLibInit( )
Initializes the AIO library (non-POSIX).
aioShow( )
Displays the outstanding AIO requests (non-POSIX).a
aio_read( )
Initiates an asynchronous read operation.
aio_write( )
Initiates an asynchronous write operation.
lio_listio( )
Initiates a list of up to LIO_MAX asynchronous I/O requests.
aio_error( )
Retrieves the error status of an AIO operation.
aio_return( )
Retrieves the return status of a completed AIO operation.
aio_cancel( )
Cancels a previously submitted AIO operation.
aio_suspend( )
Waits until an AIO operation is done, interrupted, or timed out.
aio_fsync( )
Asynchronously forces file synchronization.
7
a. This function is not built into the host shell. To use it from the host shell, VxWorks must
be configured with the INCLUDE_POSIX_AIO_SHOW component. When you
invoke the function, its output is sent to the standard output device.
The routine aioSysInit( ) takes three parameters: the number of AIO system tasks
to spawn, and the priority and stack size for these system tasks. The number of
AIO system tasks spawned equals the number of AIO requests that can be handled
in parallel. The default initialization call uses three constants:
MAX_AIO_SYS_TASKS, AIO_TASK_PRIORITY, and AIO_TASK_STACK_SIZE.
When any of the parameters passed to aioSysInit( ) is 0, the corresponding value
is taken from AIO_IO_TASKS_DFLT, AIO_IO_PRIO_DFLT, and
AIO_IO_STACK_DFLT (all defined in
installDir/vxworks-6.x/target/h/aioSysDrv.h).
Table 7-6 lists the names of the constants, and shows the constants used within
initialization routines when the parameters are left at their default values of 0, and
where these constants are defined.
383
VxWorks
Kernel Programmer's Guide, 6.6
Table 7-6
Init Routine
AIO Initialization Functions and Related Constants
Configuration Parameter
Def. Header File Constant
Value used when arg = 0
Def.
Value Header File
aioPxLibInit( ) MAX_LIO_CALLS
0
AIO_CLUST_MAX
100
aioSysInit( )
MAX_AIO_SYS_TASKS
0
AIO_IO_TASKS_DFLT
2
aioSysDrv.h
AIO_TASK_PRIORITY
0
AIO_IO_PRIO_DFLT
50
aioSysDrv.h
AIO_TASK_STACK_SIZE
0
AIO_IO_STACK_DFLT
private/aioPxLibP.h
0x7000 aioSysDrv.h
7.7.2 AIO Control Block
Each of the AIO calls takes an AIO control block (aiocb) as an argument. The
calling routine must allocate space for the aiocb, and this space must remain
available for the duration of the AIO operation. (Thus the aiocb must not be
created on the task's stack unless the calling routine will not return until after the
AIO operation is complete and aio_return( ) has been called.) Each aiocb describes
a single AIO operation. Therefore, simultaneous asynchronous I/O operations
using the same aiocb are not valid and produce undefined results.
The aiocb structure is defined in aio.h. It contains the following fields:
aio_fildes
The file descriptor for I/O.
aio_offset
The offset from the beginning of the file.
aio_buf
The address of the buffer from/to which AIO is requested.
aio_nbytes
The number of bytes to read or write.
aio_reqprio
The priority reduction for this AIO request.
aio_sigevent
The signal to return on completion of an operation (optional).
aio_lio_opcode
An operation to be performed by a lio_listio( ) call.
384
7 I/O System
7.7 Asynchronous Input/Output
aio_sys
The address of VxWorks-specific data (non-POSIX).
For full definitions and important additional information, see the reference entry
for aioPxLib.
!
CAUTION: The aiocb structure and the data buffers referenced by it are used by the
system to perform the AIO request. Therefore, once the aiocb has been submitted
to the system, the application must not modify the aiocb structure until after a
subsequent call to aio_return( ). The aio_return( ) call retrieves the previously
submitted AIO data structures from the system. After the aio_return( ) call, the
calling application can modify the aiocb, free the memory it occupies, or reuse it
for another AIO call. If space for the aiocb is allocated from the stack, the task
should not be deleted (or complete running) until the aiocb has been retrieved
from the system with an aio_return( ) call.
7.7.3 Using AIO
The routines aio_read( ), aio_write( ), or lio_listio( ) initiate AIO operations. The
last of these, lio_listio( ), allows you to submit a number of asynchronous requests
(read and/or write) at one time. In general, the actual I/O (reads and writes)
initiated by these routines does not happen immediately after the AIO request. For
this reason, their return values do not reflect the outcome of the actual I/O
operation, but only whether a request is successful—that is, whether the AIO
routine is able to put the operation on a queue for eventual execution.
After the I/O operations themselves execute, they also generate return values that
reflect the success or failure of the I/O. There are two routines that you can use to
get information about the success or failure of the I/O operation: aio_error( ) and
aio_return( ). You can use aio_error( ) to get the status of an AIO operation
(success, failure, or in progress), and aio_return( ) to obtain the return values from
the individual I/O operations. Until an AIO operation completes, its error status is
EINPROGRESS. To cancel an AIO operation, call aio_cancel( ). To force all I/O
operations to the synchronized I/O completion state, use aio_fsync( ).
AIO with Periodic Checks for Completion
The following code uses a pipe for the asynchronous I/O operations. The example
creates the pipe, submits an AIO read request, verifies that the read request is still
in progress, and submits an AIO write request. Under normal circumstances, a
385
7
VxWorks
Kernel Programmer's Guide, 6.6
synchronous read to an empty pipe blocks and the task does not execute the write,
but in the case of AIO, we initiate the read request and continue. After the write
request is submitted, the example task loops, checking the status of the AIO
requests periodically until both the read and write complete. Because the AIO
control blocks are on the stack, we must call aio_return( ) before returning from
aioExample( ).
Example 7-2
Asynchronous I/O
/* aioEx.c - example code for using asynchronous I/O */
/* includes */
#include
#include
#include
#include
#include
#include
/* defines */
#define BUFFER_SIZE 200
struct aiocb
struct aiocb
aiocb_read; /* read aiocb */
aiocb_write; /* write aiocb */
/************************************************************************
* aioExample - use AIO library * This example shows the basic functions of
the AIO library.
* RETURNS: OK if successful, otherwise ERROR.
*/
STATUS aioExample (const char *exFile)
{
int
fd;
char
buffer [BUFFER_SIZE]; /* buffer for read aiocb */
static char * test_string = "testing 1 2 3";
int
error;
if ((fd = open (exFile, O_CREAT | O_TRUNC | O_RDWR, 0666)) ==
ERROR)
{
printf ("aioExample: cannot open %s for writing. errno 0x%x\n", exFile,
errno);
return (ERROR);
}
printf ("aioExample: Example file = %s\tFile descriptor = %d\n",
exFile, fd);
/* initialize read and write aiocbs */
386
7 I/O System
7.7 Asynchronous Input/Output
memset ( &aiocb_read, 0, sizeof (struct aiocb));
memset ( buffer, 0, sizeof (buffer));
aiocb_read.aio_fildes = fd;
aiocb_read.aio_buf = buffer;
aiocb_read.aio_nbytes = BUFFER_SIZE;
aiocb_read.aio_reqprio = 0;
memset ( &aiocb_write, 0, sizeof (struct aiocb));
aiocb_write.aio_fildes = fd;
aiocb_write.aio_buf = test_string;
aiocb_write.aio_nbytes = strlen (test_string);
aiocb_write.aio_reqprio = 0;
7
/* initiate the read */
if (aio_read (&aiocb_read) == -1)
printf ("aioExample: aio_read failed\n");
/* verify that it is in progress */
if (aio_error (&aiocb_read) == EINPROGRESS)
printf ("aioExample: read is still in progress\n");
/* write to pipe - the read should be able to complete */
printf ("aioExample: getting ready to initiate the write\n");
if (aio_write (&aiocb_write) == -1)
printf ("aioExample: aio_write failed\n");
/* wait til both read and write are complete */
while ((error = aio_error (&aiocb_read) == EINPROGRESS) ||
(aio_error (&aiocb_write) == EINPROGRESS))
sleep (1);
printf ("aioExample: error = %d\n", error);
/* print out what was read */
printf ("aioExample: message = %s\n", buffer);
/* clean up */
if (aio_return (&aiocb_read) == -1)
printf ("aioExample: aio_return for aiocb_read failed\n");
if (aio_return (&aiocb_write) == -1)
printf ("aioExample: aio_return for aiocb_write failed\n");
close (fd);
return (OK);
}
387
VxWorks
Kernel Programmer's Guide, 6.6
Alternatives for Testing AIO Completion
A task can determine whether an AIO request is complete in any of the following
ways:
■
Check the result of aio_error( ) periodically, as in the previous example, until
the status of an AIO request is no longer EINPROGRESS.
■
Use aio_suspend( ) to suspend the task until the AIO request is complete.
■
Use signals to be informed when the AIO request is complete.
The following example is similar to the preceding aioExample( ), except that it
uses signals for notification that the write operation has finished. If you test this
from the shell, spawn the routine to run at a lower priority than the AIO system
tasks to assure that the test routine does not block completion of the AIO request.
Example 7-3
Asynchronous I/O with Signals
#include
#include
#include
#include
#include
#include
/* defines */
#define
#define
#define
#define
BUFFER_SIZE
200
LIST_SIZE
1
WRITE_EXAMPLE_SIG_NO 25 /* signal number */
READ_EXAMPLE_SIG_NO 26 /* signal number */
/* forward declarations */
void writeSigHandler (int sig, struct siginfo * info, void * pContext);
void readSigHandler (int sig, struct siginfo * info, void * pContext);
static
static
static
static
static
struct
struct
struct
struct
char
aiocb
aiocb
sigaction
sigaction
aiocb_read;
/* read aiocb */
aiocb_write; /* write aiocb */
write_action; /* signal info */
read_action; /* signal info */
buffer [BUFFER_SIZE]; /* aiocb read buffer */
/************************************************************************
* aioExampleSig - use AIO library.
*
* This example shows the basic functions of the AIO library.
* Note if this is run from the shell it must be spawned. Use:
* -> sp aioExampleSig
*
* RETURNS: OK if successful, otherwise ERROR.
*/
388
7 I/O System
7.7 Asynchronous Input/Output
STATUS aioExampleSig (const char *exFile)
{
int
fd;
static char * test_string = "testing 1 2 3";
if ((fd = open (exFile, O_CREAT | O_TRUNC| O_RDWR, 0666)) == ERROR)
{
printf ("aioExample: cannot open %s errno 0x%x\n", exFile, errno);
return (ERROR);
}
printf ("aioExampleSig: Example file = %s\tFile descriptor = %d\n",
exFile, fd);
/* set up signal handler for WRITE_EXAMPLE_SIG_NO */
write_action.sa_sigaction = writeSigHandler;
write_action.sa_flags = SA_SIGINFO;
sigemptyset (&write_action.sa_mask);
sigaction (WRITE_EXAMPLE_SIG_NO, &write_action, NULL);
/* set up signal handler for READ_EXAMPLE_SIG_NO */
read_action.sa_sigaction = readSigHandler;
read_action.sa_flags = SA_SIGINFO;
sigemptyset (&read_action.sa_mask);
sigaction (READ_EXAMPLE_SIG_NO, &read_action, NULL);
/* initialize read and write aiocbs */
memset ( &aiocb_read, 0, sizeof (struct aiocb));
memset ( buffer, 0, sizeof (buffer));
aiocb_read.aio_fildes = fd;
aiocb_read.aio_buf = buffer;
aiocb_read.aio_nbytes = BUFFER_SIZE;
aiocb_read.aio_reqprio = 0;
/* set up signal info */
aiocb_read.aio_sigevent.sigev_signo = READ_EXAMPLE_SIG_NO;
aiocb_read.aio_sigevent.sigev_notify = SIGEV_SIGNAL;
aiocb_read.aio_sigevent.sigev_value.sival_ptr =
(void *) &aiocb_read;
memset ( &aiocb_write, 0, sizeof (struct aiocb));
aiocb_write.aio_fildes = fd;
aiocb_write.aio_buf = test_string;
aiocb_write.aio_nbytes = strlen (test_string);
aiocb_write.aio_reqprio = 0;
389
7
VxWorks
Kernel Programmer's Guide, 6.6
/* set up signal info */
aiocb_write.aio_sigevent.sigev_signo = WRITE_EXAMPLE_SIG_NO;
aiocb_write.aio_sigevent.sigev_notify = SIGEV_SIGNAL;
aiocb_write.aio_sigevent.sigev_value.sival_ptr =
(void *) &aiocb_write;
/* initiate the read */
if (aio_read (&aiocb_read) == -1)
printf ("aioExampleSig: aio_read failed\n");
/* verify that it is in progress */
if (aio_error (&aiocb_read) == EINPROGRESS)
printf ("aioExampleSig: read is still in progress\n");
/* write to pipe - the read should be able to complete */
printf ("aioExampleSig: getting ready to initiate the write\n");
if (aio_write (&aiocb_write) == -1)
printf ("aioExampleSig: aio_write failed\n");
close (fd);
return (OK);
}
void writeSigHandler
(
int
sig,
struct siginfo *
info,
void *
pContext
)
{
/* print out what was written */
printf ("writeSigHandler: Got signal for aio write\n");
/* write is complete so let’s do cleanup for it here */
if (aio_return (info->si_value.sival_ptr) == -1)
{
printf ("writeSigHandler: aio_return for aiocb_write failed\n");
}
}
void readSigHandler
(
int
sig,
struct siginfo *
info,
void *
pContext
)
{
/* print out what was read */
printf ("readSigHandler: Got signal for aio read\n");
390
7 I/O System
7.8 Devices in VxWorks
/* write is complete so let’s do cleanup for it here */
if (aio_return (info->si_value.sival_ptr) == -1)
{
printf ("readSigHandler: aio_return for aiocb_read failed\n");
}
else
{
printf ("aioExample: message = %s\n", buffer);
}
}
7
7.8 Devices in VxWorks
The VxWorks I/O system is flexible, allowing different device drivers to handle
the seven basic I/O functions. All VxWorks device drivers follow the basic
conventions outlined previously, but differ in specifics; this section describes those
specifics.
Table 7-7
Devices Provided with VxWorks
Device
Driver Description
tty
Terminal device
pty
Pseudo-terminal device
pipe
Pipe device
mem
Pseudo memory device
nfs
NFS client device
net
Network device for remote file access
null
Null device
ram
RAM device for creating a RAM disk
scsi
SCSI interface
romfs
ROMFS device
–
Other hardware-specific device
391
VxWorks
Kernel Programmer's Guide, 6.6
!
WARNING: Devices should not be given the same name, or they will overwrite
each other in core I/O.
NOTE: Only VxBus-compatible drivers can be used with the symmetric
multiprocessing (SMP) configuration of VxWorks. For general information about
VxWorks SMP and about migration, see 15. VxWorks SMP and 15.15 Migrating
Code to VxWorks SMP, p.702.
7.8.1 Serial I/O Devices: Terminal and Pseudo-Terminal Devices
VxWorks provides terminal and pseudo-terminal devices (tty and pty). The tty
device is for actual terminals; the pty device is for processes that simulate
terminals. These pseudo terminals are useful in applications such as remote login
facilities.
VxWorks serial I/O devices are buffered serial byte streams. Each device has a ring
buffer (circular buffer) for both input and output. Reading from a tty device
extracts bytes from the input ring. Writing to a tty device adds bytes to the output
ring. The size of each ring buffer is specified when the device is created during
system initialization.
NOTE: For the remainder of this section, the term tty is used to indicate both tty
and pty devices
tty Options
The tty devices have a full range of options that affect the behavior of the device.
These options are selected by setting bits in the device option word using the
ioctl( ) routine with the FIOSETOPTIONS function. For example, to set all the tty
options except OPT_MON_TRAP:
status = ioctl (fd, FIOSETOPTIONS, OPT_TERMINAL & ~OPT_MON_TRAP);
For more information, see I/O Control Functions, p.396.
Table 7-8 is a summary of the available options. The listed names are defined in the
header file ioLib.h. For more detailed information, see the API reference entry for
tyLib.
392
7 I/O System
7.8 Devices in VxWorks
Table 7-8
Tty Options
Library
Description
OPT_LINE
Selects line mode. (See Raw Mode and Line Mode, p.393.)
OPT_ECHO
Echoes input characters to the output of the same channel.
OPT_CRMOD
Translates input RETURN characters into NEWLINE (\n);
translates output NEWLINE into RETURN-LINEFEED.
OPT_TANDEM
Responds to software flow control characters CTRL+Q and
CTRL+S (XON and XOFF).
OPT_7_BIT
Strips the most significant bit from all input bytes.
OPT_MON_TRAP Enables the special ROM monitor trap character, CTRL+X by
default.
OPT_ABORT
Enables the special kernel shell abort character, CTRL+C by
default. (Only useful if the kernel shell is configured into the
system)
OPT_TERMINAL
Sets all of the above option bits.
OPT_RAW
Sets none of the above option bits.
Raw Mode and Line Mode
A tty device operates in one of two modes: raw mode (unbuffered) or line mode. Raw
mode is the default. Line mode is selected by the OPT_LINE bit of the device option
word (see tty Options, p.392).
In raw mode, each input character is available to readers as soon as it is input from
the device. Reading from a tty device in raw mode causes as many characters as
possible to be extracted from the input ring, up to the limit of the user’s read buffer.
Input cannot be modified except as directed by other tty option bits.
In line mode, all input characters are saved until a NEWLINE character is input; then
the entire line of characters, including the NEWLINE, is made available in the ring
at one time. Reading from a tty device in line mode causes characters up to the end
of the next line to be extracted from the input ring, up to the limit of the user’s read
buffer. Input can be modified by the special characters CTRL+H (backspace),
393
7
VxWorks
Kernel Programmer's Guide, 6.6
CTRL+U (line-delete), and CTRL+D (end-of-file), which are discussed in tty Special
Characters, p.394.
tty Special Characters
The following special characters are enabled if the tty device operates in line mode,
that is, with the OPT_LINE bit set:
■
The backspace character, by default CTRL+H, causes successive previous
characters to be deleted from the current line, up to the start of the line. It does
this by echoing a backspace followed by a space, and then another backspace.
■
The line-delete character, by default CTRL+U, deletes all the characters of the
current line.
■
The end-of-file (EOF) character, by default CTRL+D, causes the current line to
become available in the input ring without a NEWLINE and without entering
the EOF character itself. Thus if the EOF character is the first character typed
on a line, reading that line returns a zero byte count, which is the usual
indication of end-of-file.
The following characters have special effects if the tty device is operating with the
corresponding option bit set:
■
The software flow control characters CTRL+Q and CTRL+S (XON and XOFF).
Receipt of a CTRL+S input character suspends output to that channel.
Subsequent receipt of a CTRL+Q resumes the output. Conversely, when the
VxWorks input buffer is almost full, a CTRL+S is output to signal the other side
to suspend transmission. When the input buffer is empty enough, a CTRL+Q
is output to signal the other side to resume transmission. The software flow
control characters are enabled by OPT_TANDEM.
■
The ROM monitor trap character, by default CTRL+X. This character traps to the
ROM-resident monitor program. Note that this is drastic. All normal VxWorks
functioning is suspended, and the computer system is controlled entirely by
the monitor. Depending on the particular monitor, it may or may not be
possible to restart VxWorks from the point of interruption.1 The monitor trap
character is enabled by OPT_MON_TRAP.
■
The special kernel shell abort character, by default CTRL+C. This character
restarts the kernel shell if it gets stuck in an unfriendly routine, such as one that
1. It will not be possible to restart VxWorks if un-handled external interrupts occur during the
boot countdown.
394
7 I/O System
7.8 Devices in VxWorks
has taken an unavailable semaphore or is caught in an infinite loop. The kernel
shell abort character is enabled by OPT_ABORT.
The characters for most of these functions can be changed using the tyLib routines
shown in Table 7-9.
Table 7-9
Tty Special Characters
Character
Description
Modifier
CTRL+H
backspace (character delete)
tyBackspaceSet( )
CTRL+U
line delete
tyDeleteLineSet( )
CTRL+D
EOF (end of file)
tyEOFSet( )
CTRL+C
kernel shell abort
tyAbortSet( )
CTRL+X
trap to boot ROMs
tyMonitorTrapSet( )
CTRL+S
output suspend
N/A
CTRL+Q
output resume
N/A
395
7
VxWorks
Kernel Programmer's Guide, 6.6
I/O Control Functions
The tty devices respond to the ioctl( ) functions in Table 7-10, defined in ioLib.h.
For more information, see the reference entries for tyLib, ttyDrv, and ioctl( ).
Table 7-10
I/O Control Functions Supported by tyLib
Function
Description
FIOBAUDRATE
Sets the baud rate to the specified argument.
FIOCANCEL
Cancels a read or write.
FIOFLUSH
Discards all bytes in the input and output buffers.
FIOGETNAME
Gets the filename of the file descriptor.
FIOGETOPTIONS Returns the current device option word.
!
FIONREAD
Gets the number of unread bytes in the input buffer.
FIONWRITE
Gets the number of bytes in the output buffer.
FIOSETOPTIONS
Sets the device option word.
CAUTION: To change the driver’s hardware options (for example, the number of
stop bits or parity bits), use the ioctl( ) function SIO_HW_OPTS_SET. Because this
command is not implemented in most drivers, you may need to add it to your BSP
serial driver, which resides in installDir/vxworks-6.x/target/src/drv/sio. The details
of how to implement this command depend on your board’s serial chip. The
constants defined in the header file installDir/vxworks-6.x/target/h/sioLib.h
provide the POSIX definitions for setting the hardware options.
7.8.2 Pipe Devices
Pipes are virtual devices by which tasks communicate with each other through the
I/O system. Tasks write messages to pipes; these messages can then be read by
other tasks. Pipe devices are managed by pipeDrv and use the kernel message
queue facility to bear the actual message traffic.
396
7 I/O System
7.8 Devices in VxWorks
Creating Pipes
Pipes are created by calling the pipe create routine:
status = pipeDevCreate ("/pipe/name", maxMsgs, maxLength);
The new pipe can have at most maxMsgs messages queued at a time. Tasks that
write to a pipe that already has the maximum number of messages queued are
blocked until a message is dequeued. Each message in the pipe can be at most
maxLength bytes long; attempts to write longer messages result in an error.
7
Writing to Pipes from ISRs
VxWorks pipes are designed to allow ISRs to write to pipes in the same way as
task-level code. Many VxWorks facilities cannot be used from ISRs, including
output to devices other than pipes. However, ISRs can use pipes to communicate
with tasks, which can then invoke such facilities. ISRs write to a pipe using the
write( ) call. Tasks and ISRs can write to the same pipes. However, if the pipe is
full, the message is discarded because the ISRs cannot pend. ISRs must not invoke
any I/O function on pipes other than write( ). For more information ISRs, see
4.20 Interrupt Service Routines, p.241.
I/O Control Functions
Pipe devices respond to the ioctl( ) functions summarized in Table 7-11. The
functions listed are defined in the header file ioLib.h. For more information, see
the reference entries for pipeDrv and for ioctl( ) in ioLib.
Table 7-11
I/O Control Functions Supported by pipeDrv
Function
Description
FIOFLUSH
Discards all messages in the pipe.
FIOGETNAME
Gets the pipe name of the file descriptor.
FIONMSGS
Gets the number of messages remaining in the pipe.
FIONREAD
Gets the size in bytes of the first message in the pipe.
397
VxWorks
Kernel Programmer's Guide, 6.6
7.8.3 Pseudo I/O Device
The memDrv device allows the I/O system to access memory directly as a
pseudo-I/O device. Memory location and size are specified when the device is
created. The device provides a high-level means for reading and writing bytes in
absolute memory locations through I/O calls. It is useful when data must be
preserved between boots of VxWorks or when sharing data between CPUs.
The memDrv driver is initialized automatically by the system with memDrv( )
when the INCLUDE_USR_MEMDRV component is included in VxWorks. The call
for device creation must be made from the kernel:
STATUS memDevCreate
(char * name, char * base, int length)
Memory for the device is an absolute memory location beginning at base. The
length parameter indicates the size of the memory.
For additional information on the memory driver, see the memDrv( ),
memDevCreate( ), and memDevCreateDir( ) entries in the VxWorks API
reference, as well as the entry for memdrvbuild in the online Wind River Host
Utilities API Reference.
For information about creating a RAM disk, which provides support for file
systems,.
I/O Control Functions
The memory device responds to the ioctl( ) functions summarized in Table 7-12.
The functions listed are defined in the header file ioLib.h.
Table 7-12
I/O Control Functions Supported by memDrv
Function
Description
FIOSEEK
Sets the current byte offset in the file.
FIOWHERE
Returns the current byte position in the file.
For more information, see the reference entries for memDrv, ioLib, and ioctl( ).
398
7 I/O System
7.8 Devices in VxWorks
7.8.4 Network File System (NFS) Devices
Network File System (NFS) devices allow files on remote hosts to be accessed with
the NFS protocol. The NFS protocol specifies both client software, to read files from
remote machines, and server software, to export files to remote machines.
The driver nfsDrv acts as a VxWorks NFS client to access files on any NFS server
on the network. VxWorks also allows you to run an NFS server to export files to
other systems.
Using NFS devices, you can create, open, and access remote files exactly as though
they were on a file system on a local disk. This is called network transparency.
For detailed information about the VxWorks implementation of NFS, see
9. Network File System: NFS.
Mounting a Remote NFS File System from VxWorks
Access to a remote NFS file system is established by mounting that file system
locally and creating an I/O device for it using nfsMount( ). Its arguments are
(1) the host name of the NFS server, (2) the name of the host file system, and (3) the
local name for the file system.
For example, the following call mounts /usr of the host mars as /vxusr locally:
nfsMount ("mars", "/usr", "/vxusr");
This creates a VxWorks I/O device with the specified local name (/vxusr, in this
example). If the local name is specified as NULL, the local name is the same as the
remote name.
After a remote file system is mounted, the files are accessed as though the file
system were local. Thus, after the previous example, opening the file /vxusr/foo
opens the file /usr/foo on the host mars.
The remote file system must be exported by the system on which it actually resides.
However, NFS servers can export only local file systems. Use the appropriate
command on the server to see which file systems are local. NFS requires
authentication parameters to identify the user making the remote access. To set
these parameters, use the routines nfsAuthUnixSet( ) and nfsAuthUnixPrompt( ).
To include NFS client support, use the INCLUDE_NFS component.
The subject of exporting and mounting NFS file systems and authenticating access
permissions is discussed in more detail in 9. Network File System: NFS. See also the
399
7
VxWorks
Kernel Programmer's Guide, 6.6
reference entries nfsLib and nfsDrv, and the NFS documentation from Sun
Microsystems.
I/O Control Functions for NFS Clients
NFS client devices respond to the ioctl( ) functions summarized in Table 7-13. The
functions listed are defined in ioLib.h. For more information, see the reference
entries for nfsDrv, ioLib, and ioctl( ).
Table 7-13
I/O Control Functions Supported by nfsDrv
Function
Description
FIOFSTATGET
Gets file status information (directory entry data).
FIOGETNAME
Gets the filename of the file descriptor.
FIONREAD
Gets the number of unread bytes in the file.
FIOREADDIR
Reads the next directory entry.
FIOSEEK
Sets the current byte offset in the file.
FIOSYNC
Flushes data to a remote NFS file.
FIOWHERE
Returns the current byte position in the file.
7.8.5 Non-NFS Network Devices
VxWorks also supports network access to files on a remote host through the
Remote Shell protocol (RSH) or the File Transfer Protocol (FTP).
These implementations of network devices use the driver netDrv, which is
included in the Wind River Network Stack. Using this driver, you can open, read,
write, and close files located on remote systems without needing to manage the
details of the underlying protocol used to effect the transfer of information. (For
more information, see the Wind River Network Stack for VxWorks 6 Programmer’s
Guide.)
When a remote file is opened using RSH or FTP, the entire file is copied into local
memory. As a result, the largest file that can be opened is restricted by the available
memory. Read and write operations are performed on the memory-resident copy
400
7 I/O System
7.8 Devices in VxWorks
of the file. When closed, the file is copied back to the original remote file if it was
modified.
In general, NFS devices are preferable to RSH and FTP devices for performance
and flexibility, because NFS does not copy the entire file into local memory.
However, NFS is not supported by all host systems.
Creating Network Devices
To access files on a remote host using either RSH or FTP, a network device must
first be created by calling the routine netDevCreate( ). The arguments to
netDevCreate( ) are (1) the name of the device, (2) the name of the host the device
accesses, and (3) which protocol to use: 0 (RSH) or 1 (FTP).
For example, the following call creates an RSH device called mars: that accesses the
host mars. By convention, the name for a network device is the remote machine’s
name followed by a colon (:).
netDevCreate ("mars:", "mars", 0);
Files on a network device can be created, opened, and manipulated as if on a local
disk. Thus, opening the file mars:/usr/foo actually opens /usr/foo on host mars.
Note that creating a network device allows access to any file or device on the
remote system, while mounting an NFS file system allows access only to a
specified file system.
For the files of a remote host to be accessible with RSH or FTP, permissions and
user identification must be established on both the remote and local systems.
Creating and configuring network devices is discussed in detail in Wind River
Network Stack for VxWorks 6 Programmer’s Guide: File Access Applications and in the
API reference entry for netDrv.
I/O Control Functions
RSH and FTP devices respond to the same ioctl( ) functions as NFS devices except
for FIOSYNC and FIOREADDIR. The functions are defined in the header file
ioLib.h. For more information, see the API reference entries for netDrv and ioctl( ).
401
7
VxWorks
Kernel Programmer's Guide, 6.6
7.8.6 Null Devices
VxWorks provides both /null and /dev/null for null devices. The /null device is the
traditional VxWorks null device, which is provided by default for backward
compatibility. The /dev/null device is provided by the
BUNDLE_RTP_POSIX_PSE52 component bundle, and is required for conformance
with the POSIX PSE52 profile.
Note that the devs shell command lists /null and /dev/null with other devices, but
the ls command does not list /dev/null under the VRFS root directory (because the
name violates the VRFS naming scheme). Applications can, in any case, use /null
or /dev/null as required.
For information about POSIX PSE52, see the VxWorks Kernel Programmer’s Guide:
POSIX Facilities. For information about VRFS, see VxWorks Kernel Programmer’s
Guide: Local File Systems.
7.8.7 Sockets
In VxWorks, the underlying basis of network communications is sockets. A socket
is an endpoint for communication between tasks; data is sent from one socket to
another. Sockets are not created or opened using the standard I/O functions.
Instead, they are created by calling socket( ), and connected and accessed using
other routines in sockLib. However, after a stream socket (using TCP) is created
and connected, it can be accessed as a standard I/O device, using read( ), write( ),
ioctl( ), and close( ). The value returned by socket( ) as the socket handle is in fact
an I/O system file descriptor.
VxWorks socket routines are source-compatible with the BSD 4.4 UNIX socket
functions and the Windows Sockets (Winsock 1.1) networking standard. Use of
these routines is discussed in Wind River Network Stack for VxWorks 6 Programmer’s
Guide.
7.8.8 Extended Block Device Facility: XBD
The extended block device (XBD) facility mediates I/O activity between file
systems and block devices. It provides a standard interface between file systems
on the one hand, and block drivers on the other.
The XBD facility also provides support for removable file systems, automatic file
system detection, and multiple file systems. For more information on these
features, see 8. Local File Systems.
402
7 I/O System
7.8 Devices in VxWorks
NOTE: The XBD facility is required for some file systems (such as HRFS, dosFs,
cdromFs, and rawFs), but not others (such as ROMFS).
For detailed information on developing XBD-compatible device drivers, see the
VxWorks Device Driver Developer’s Guide.
The basic XBD facility is proved with the INCLUDE_XBD component, which also
provides generic service for following optional components:
INCLUDE_XBD_RAMDRV
Provides support for RAM disks. See XBD RAM Disk, p.411.
INCLUDE_XBD_PART_LIB
Provides disk partitioning facilities. See XBD Disk Partition Manager, p.403.
INCLUDE_XBD_BLK_DEV
Provides support for legacy block device drivers that were designed to work
with the predecessor to XBD—the cache block I/O (CBIO) facility. These
devices include floppy drives, SCSI, and TrueFFS (the disk-access emulator for
flash). See XBD Block Device Wrapper, p.404.
INCLUDE_XBD_TRANS
Provides a transaction-based file system (TRFS) facility, which can be used
with dosFs. It provides fault-tolerant file system consistency and fast recovery
in response to power loss. See 7.8.9 Transaction-Based Reliable File System
Facility: TRFS, p.405.
XBD Disk Partition Manager
VxWorks provides support for PC-style disk partitioning with the
INCLUDE_XBD_PART_LIB component, which facilitates sharing fixed disks and
removable cartridges between VxWorks target systems and PCs running
Windows. This component includes two modules: xbdPartition and partLib.
xbdPartition Module
The xbdPartition facility creates a device for each partition that is detected on
media. Each partition that is detected is probed by the file system monitor and an
I/O device is added to VxWorks. This device is an instantiation of the file system
found by the file system monitor (or rawFs if the file system is not recognized or
detected). If no partitions are detected, a single device is created to represent the
entire media. There can be up to four partitions on a single media. For information
about the file system monitor, see 8.2 File System Monitor, p.455.
403
7
VxWorks
Kernel Programmer's Guide, 6.6
The partition facility also names the partitions. The names are derived from the
base device name and the partition number. The base device name is derived from
the device driver name (for more information in this regard, see the VxWorks
Device Driver Developer’s Guide). For example, the XBD-compatible device for an
ATA hard disk would have a base name of /ata00. If it had four partitions, they
would be named as follows:
/ata00:1
/ata00:2
/ata00:3
/ata00:4
If there were no partitions, the name would be /ata00:0. For an example of how the
facility is used, see Example 8-1.
partLib Library
The partLib library provides facilities for creating PC-style partitions on media. It
can create up to four primary partitions. Note that when partitions are created, any
existing information on the media is lost. For more information see the VxWorks
API reference for xbdCreatePartition( ).
XBD Block Device Wrapper
The INCLUDE_XBD_BLKDEV component provides support for legacy block
devices that were designed to work the predecessor to XBD—the cache block I/O
(CBIO) facility. It provides a wrapper XBD facility that converts the block I/O
driver interface based on the BLK_DEV logical block device structure into an XBD
API-compliant interface.
NOTE: The Wind River devices that require the INCLUDE_XBD_BLKDEV
component in addition to INCLUDE_XBD are floppy, SCSI, and TrueFFS (the
disk-access emulator for flash) drivers. Any third-party device drivers based on
the BLK_DEV interface also require INCLUDE_XBD_BLKDEV.
The Wind River drivers that do not require the INCLUDE_XBD_BLK_DEV
component are USB block storage, ATA, and the XBD RAM disk.
!
CAUTION: Depending on the implementation of the driver, the
INCLUDE_XBD_BLK_DEV component may not properly detect media insertion
and removal. It may, therefore remove the file system when the media is removed,
or not instantiate a file system when media is inserted.
404
7 I/O System
7.8 Devices in VxWorks
XBD TRFS Component
The INCLUDE_XBD_TRANS component is an XBD-compatible transaction-based
reliable file system (TRFS) facility. TRFS an I/O facility that provides fault-tolerant
file system layer for the dosFs file system. See 7.8.9 Transaction-Based Reliable File
System Facility: TRFS, p.405 for more information.
7.8.9 Transaction-Based Reliable File System Facility: TRFS
The transaction-based reliable file system (TRFS) facility provides a fault-tolerant
file system I/O layer for the dosFs file system. It is provided with the
INCLUDE_XBD_TRANS component.
TRFS provides both file system consistency and fast recovery for the dosFs file
system—DOS-compatible file systems are themselves neither reliable nor
transaction-based. It is designed to operate with XBD-compliant device drivers for
hard disks, floppy disks, compact flash media, TrueFFS flash devices, and so on. It
can also be used with the XBD wrapper component for device drivers that are not
XBD-compliant.
TRFS provides reliability in resistance to sudden power loss: files and data that are
already written to media are unaffected, they will not be deleted or corrupted
because data is always written either in its entirety or not at all.
TRFS provides additional guarantees in its transactional feature: data is always
maintained intact up to a given commit transaction. User applications set
transaction points on the file system. If there is an unexpected failure of the system,
the file system is returned to the state it was in at the last transaction point. That is,
if data has changed on the media after a commit transaction but prior to a power
loss, it is automatically restored to the its state at the last commit transaction to
further ensure data integrity. On mounting the file system, TRFS detects any
failure and rolls back data to the last secure transaction.
Unlike some facilities that provide data integrity on a file-by-file basis, TRFS
protects the medium as a whole. It is transactional for a file system, which means
that setting transaction points will commit all files, not just the one used to set the
transaction point.
405
7
VxWorks
Kernel Programmer's Guide, 6.6
NOTE: While TRFS is a I/O layer added to dosFs, it uses a modified on-media
format that is not compatible with other FAT-based file systems, including
Microsoft Windows and the VxWorks dosFs file system without the TRFS layer. It
should not, therefore, be used when compatibility with other systems is a
requirement
For information about dosFs, see 8.5 MS-DOS-Compatible File System: dosFs, p.478.
Configuring VxWorks With TRFS
Configure VxWorks with the INCLUDE_XBD_TRANS component to provide TRFS
functionality for your dosFs file system.
Automatic Instantiation of TRFS
TRFS is automatically detected and instantiated if the media has already been
formatted for use with TRFS, in a manner very similar to the instantiation of the
dosFs or HRFS file system. The primary difference is that when TRFS is detected
by the file system monitor, it calls the TRFS creation function, and the creation
function then creates another XBD instance and generates an insertion event for it.
The monitor then detects the new XBD and begins probing. In this case, however,
the monitor does not examine the media directly—all commands are routed
through TRFS, which performs the appropriate translations. If a file system is
detected, such as dosFs, the dosFs creation function is called by the monitor and
dosFs is instantiated. If not, rawfs is instantiated.
For information about how file systems are automatically instantiated, see 8.2 File
System Monitor, p.455.
Formatting a Device for TRFS
TRFS low-level formatting is accomplished with the call:
usrFormatTrans(device, overHead, type);
The arguments are:
device
The volume name to format. For example, "/ata".
406
7 I/O System
7.8 Devices in VxWorks
overHead
An integer that identifies the portion of the disk to use as transactional
workspace in parts-per-thousand of the disk.
type
An integer with the values of either FORMAT_REGULAR (0), which does not
reserve any blocks from the disk; or FORMAT_TFFS (1), which reserves the first
block.
Once the TRFS format is complete, a dosFs file system can be created by calling the
dosFs formatter on the same volume.
When a FAT file system is created using the function dosFsVolFormat( ) in
conjunction with TRFS, a transaction point is automatically inserted following the
format. One cannot, therefore, unformat by rolling back a transaction point.
Example 7-4
Formatting a Device for TRFS
/* Create a RAM disk with 512 byte sized sectors and 1024 sectors.*/
if (xbdRamDiskDevCreate (512, 1024 * 512, 0, "/trfs") == NULL)
{
printf ("Could not create RAM disk\n");
return;
}
/* Put TRFS on the RAM disk */
if (usrFormatTrans ("/trfs", 100, 0) != OK)
{
printf ("Could not format\n");
return;
}
/* Now put dosFs on TRFS */
if (dosFsVolFormat ("/trfs", DOS_OPT_BLANK, 0) != OK)
{
printf ("Could not format for dos\n");
return;
}
/* Create a file on the TRFS/DosFS volume */
fd = open ("/trfs/myfile", O_CREAT | O_RDWR, 0666);
if (fd < 0)
{
printf ("Couldn't create file\n");
return;
}
/* Commit the file creation to media */
ioctl (fd, CBIO_TRANS_COMMIT, 0);
407
7
VxWorks
Kernel Programmer's Guide, 6.6
Using TRFS in Applications
Once TRFS and dosFs are created, the dosFs file system may be used with the
ordinary file creation and manipulation commands. No changes to the file system
become permanent, however, until TRFS is used to commit them.
It is important to note that the entire dosFs file system—and not individual files—
are committed. The entire disk state must therefore be consistent before executing
a commit; that is, there must not be a file system operation in progress (by another
task, for example) when the file system is committed. If multiple tasks update the
file system, care must be taken to ensure the file data is in a known state before
setting a transaction point.
There are two ways to commit the file system:
■
Using the volume name of the device formatted for TRFS.
■
Using a file descriptor which is open on TRFS.
The function usrTransCommit( ) takes the volume name of the TRFS device and
causes it to commit. The function usrTransCommitFd( ) takes a file descriptor
open on TRFS and causes a commit of the entire file system.
TRFS Code Examples
The following code examples illustrate creating a file system with TRFS and setting
a transaction point. The first routine creates a new TRFS layer and dosFs file
system; and the second sets a transaction point.
void createTrfs
(
void
)
{
/* Create an XBD RAM disk with 512 byte sized sectors and 1024 sectors.*/
if (xbdRamDiskDevCreate (512, 1024 * 512, 0, "/trfs") == NULL)
{
printf ("Could not create XBD RAM disk\n");
return;
}
/* Put TRFS on the RAM disk */
/* Use 10% of the disk as overhead */
if (usrFormatTrans ("/trfs", 100, 0) != OK)
{
printf ("Could not format for TRFS\n");
return;
}
/* Now put dosFs on TRFS */
408
7 I/O System
7.8 Devices in VxWorks
if (dosFsVolFormat ("/trfs", DOS_OPT_BLANK, 0) != OK)
{
printf ("Could not format for dos\n");
return;
}
void transTrfs
(
void
)
{
/* This assumes a TRFS with DosFs on "/trfs" */
7
... /* Perform file operations here */
usrTransCommit ("/trfs");
... /* Perform more file operations here */
usrTransCommit ("/trfs");
}
409
VxWorks
Kernel Programmer's Guide, 6.6
7.8.10 Block Devices
A physical block device is a device that is organized as a sequence of individually
accessible blocks of data. The most common type of block device is a disk. In
VxWorks, the term block refers to the smallest addressable unit on the device. For
most disk devices, a VxWorks block corresponds to a sector, although terminology
varies.
Block devices in VxWorks have a slightly different interface than other I/O
devices. Rather than interacting directly with the I/O system, the I/O activity of
block device drivers is mediated by the extended block device (XBD) facility and a
file system. The XBD facility provides a standard interface for block device drivers
on the one hand, and for file systems on the other.
Figure 7-2 shows a layered model of I/O for both block and non-block (character)
devices. This architecture allows the same block device driver to be used with
different file systems, and reduces the number of I/O functions that must be
supported in the driver.
For information about the XBD facility, see 7.8.8 Extended Block Device Facility:
XBD, p.402.
For information about the file systems that can be used with block devices, see
8Local File Systems, p.451.
For information about information about block device drivers and how to develop
them, see the VxWorks Device Driver Developer’s Guide.
410
7 I/O System
7.8 Devices in VxWorks
Figure 7-2
Non-Block Devices and Block Devices
Application
I/O System
7
driver table
File System
XBD
Non-Block
Device Driver
Block
Device Driver
Device(s)
Device(s)
XBD RAM Disk
A RAM driver emulates a disk device, but keeps all data in memory. The
INCLUDE_XBD_RAMDRV component allows the use of a file system to access data
411
VxWorks
Kernel Programmer's Guide, 6.6
stored in RAM memory. RAM disks can be created using volatile as well a
non-volatile RAM. A RAM disk can be used with the HRFS, dosFs, and rawFs file
systems. The RAM disk links into the file system monitor and event framework.
For more about information about RAM disks, see the API reference for
xbdRamDisk, as well as Example 7-4, Example 8-3, and Example 8-7. For
information about compatible file systems, see 8.4 Highly Reliable File System:
HRFS, p.459, 8.5 MS-DOS-Compatible File System: dosFs, p.478, and 8.6 Raw File
System: rawFs, p.505.
Note that the XBD-compatible RAM disk facility supersedes the ramDrv facility.
SCSI Drivers
SCSI is a standard peripheral interface that allows connection with a wide variety
of hard disks, optical disks, floppy disks, tape drives, and CD-ROM devices. SCSI
block drivers are compatible with the dosFs libraries, and offer several advantages
for target configurations. They provide:
■
■
local mass storage in non-networked environments
faster I/O throughput than Ethernet networks
The SCSI-2 support in VxWorks supersedes previous SCSI support, although it
offers the option of configuring the original SCSI functionality, now known as
SCSI-1. With SCSI-2 enabled, the VxWorks environment can still handle SCSI-1
applications, such as file systems created under SCSI-1. However, applications that
directly used SCSI-1 data structures defined in scsiLib.h may require
modifications and recompilation for SCSI-2 compatibility.
The VxWorks SCSI implementation consists of two modules, one for the
device-independent SCSI interface and one to support a specific SCSI controller.
The scsiLib library provides routines that support the device-independent
interface; device-specific libraries provide configuration routines that support
specific controllers. There are also additional support routines for individual
targets in sysLib.c.
Configuring SCSI Drivers
Components associated with SCSI drivers are listed in Table 7-14.
412
7 I/O System
7.8 Devices in VxWorks
Table 7-14
SCSI and Related Components
Component
Description
INCLUDE_SCSI
Includes SCSI interface.
INCLUDE_SCSI2
Includes SCSI-2 extensions.
INCLUDE_SCSI_DMA
Enables DMA for SCSI.
INCLUDE_SCSI_BOOT
Allows booting from a SCSI device.
SCSI_AUTO_CONFIG
Auto-configures and locates all targets on a SCSI bus.
INCLUDE_DOSFS
Includes the dosFs file system.
INCLUDE_HRFS
Includes the HRFS file system.
INCLUDE_CDROMFS
Includes CD-ROM file system support.
7
To include SCSI-1 functionality in VxWorks, use the INCLUDE_SCSI component.
To include SCSI-2 functionality, you must use INCLUDE_SCSI2 in addition to
INCLUDE_SCSI.
Auto-configuration, DMA, and booting from a SCSI device are defined
appropriately for each BSP. If you must change these settings, see the VxWorks
API reference for sysScsiConfig( ) and the source file
installDir/vxworks-6.x/target/src/config/usrScsi.c.
!
CAUTION: Including SCSI-2 in your VxWorks image can significantly increase the
image size.
Configuring the SCSI Bus ID
Each board in a SCSI-2 environment must define a unique SCSI bus ID for the SCSI
initiator. SCSI-1 drivers, which support only a single initiator at a time, assume an
initiator SCSI bus ID of 7. However, SCSI-2 supports multiple initiators, up to eight
initiators and targets at one time. Therefore, to ensure a unique ID, choose a value
in the range 0-7 to be passed as a parameter to the driver’s initialization routine (for
example, ncr710CtrlInitScsi2( )) by the sysScsiInit( ) routine in sysScsi.c. For
more information, see the reference entry for the relevant driver initialization
routine. If there are multiple boards on one SCSI bus, and all of these boards use
the same BSP, then different versions of the BSP must be compiled for each board
by assigning unique SCSI bus IDs.
413
VxWorks
Kernel Programmer's Guide, 6.6
ROM Size Adjustment for SCSI Boot
If the INCLUDE_SCSI_BOOT component is included, larger ROMs may be required
for some boards.
Structure of the SCSI Subsystem
The SCSI subsystem supports libraries and drivers for both SCSI-1 and SCSI-2. It
consists of the following six libraries which are independent of any SCSI controller:
scsiLib
routines that provide the mechanism for switching SCSI requests to either
the SCSI-1 library (scsi1Lib) or the SCSI-2 library (scsi2Lib), as configured
by the board support package (BSP).
scsi1Lib
SCSI-1 library routines and interface, used when only INCLUDE_SCSI is
used (see Configuring SCSI Drivers, p.412).
scsi2Lib
SCSI-2 library routines and all physical device creation and deletion
routines.
scsiCommonLib
commands common to all types of SCSI devices.
scsiDirectLib
routines and commands for direct access devices (disks).
scsiSeqLib
routines and commands for sequential access block devices (tapes).
Controller-independent support for the SCSI-2 functionality is divided into
scsi2Lib, scsiCommonLib, scsiDirectLib, and scsiSeqLib. The interface to any of
these SCSI-2 libraries can be accessed directly. However, scsiSeqLib is designed to
be used in conjunction with tapeFs, while scsiDirectLib works with dosFs and
rawFs. Applications written for SCSI-1 can be used with SCSI-2; however, SCSI-1
device drivers cannot.
VxWorks targets using SCSI interface controllers require a controller-specific
device driver. These device drivers work in conjunction with the
controller-independent SCSI libraries, and they provide controller configuration
and initialization routines contained in controller-specific libraries. For example,
the Western Digital WD33C93 SCSI controller is supported by the device driver
libraries wd33c93Lib, wd33c93Lib1, and wd33c93Lib2. Routines tied to SCSI-1
(such as wd33c93CtrlCreate( )) and SCSI-2 (such as wd33c93CtrlCreateScsi2( ))
414
7 I/O System
7.8 Devices in VxWorks
are segregated into separate libraries to simplify configuration. There are also
additional support routines for individual targets in sysLib.c.
Booting and Initialization
When VxWorks is built with the INCLUDE_SCSI component, the system startup
code initializes the SCSI interface by executing sysScsiInit( ) and usrScsiConfig( ).
The call to sysScsiInit( ) initializes the SCSI controller and sets up interrupt
handling. The physical device configuration is specified in usrScsiConfig( ),
which is in installDir/vxworks-6.x/target/src/config/usrScsi.c. The routine contains
an example of the calling sequence to declare a hypothetical configuration,
including:
■
definition of physical devices with scsiPhysDevCreate( )
■
creation of logical partitions with scsiBlkDevCreate( )
■
creation of an XBD block wrapper driver with xbdBlkDevCreate( ).
If a recognized file system exists on the SCSI media, it is instantiated automatically
when xbdBlkDevCreate( ) returns. If not, the file system formatter must be called
to create the file system. See the dosFsVolFormat( ) API reference for information
about creating a dosFs file system; see the hrfsFormat( ) API reference for creating
an HRFS file system.
If you are not using SCSI_AUTO_CONFIG, modify usrScsiConfig( ) to reflect your
actual configuration. For more information on the calls used in this routine, see the
reference entries for scsiPhysDevCreate( ), scsiBlkDevCreate( ), and
xbdBlkDevCreate( ).
Device-Specific Configuration Options
The SCSI libraries have the following default behaviors enabled:
■
■
■
■
■
SCSI messages
disconnects
minimum period and maximum REQ/ACK offset
tagged command queuing
wide data transfer
Device-specific options do not need to be set if the device shares this default
behavior. However, if you must configure a device that diverges from these
default characteristics, use scsiTargetOptionsSet( ) to modify option values.
These options are fields in the SCSI_OPTIONS structure, shown below.
SCSI_OPTIONS is declared in scsi2Lib.h. You can choose to set some or all of these
option values to suit your particular SCSI device and application.
415
7
VxWorks
Kernel Programmer's Guide, 6.6
typedef struct
{
UINT
selTimeOut;
BOOL
messages;
BOOL
disconnect;
UINT8
maxOffset;
UINT8
minPeriod;
SCSI_TAG_TYPE tagType;
UINT
maxTags;
UINT8
xferWidth;
} SCSI_OPTIONS;
/* SCSI_OPTIONS - programmable options */
/*
/*
/*
/*
/*
/*
/*
/*
device selection time-out (us)
FALSE => do not use SCSI messages
FALSE => do not use disconnect
max sync xfer offset (0 => async.)
min sync xfer period (x 4 ns)
default tag type
max cmd tags available (0 => untag
wide data trnsfr width in SCSI units
*/
*/
*/
*/
*/
*/
*/
*/
There are numerous types of SCSI devices, each supporting its own mix of SCSI-2
features. To set device-specific options, define a SCSI_OPTIONS structure and
assign the desired values to the structure’s fields. After setting the appropriate
fields, call scsiTargetOptionsSet( ) to effect your selections. Example 7-6
illustrates one possible device configuration using SCSI_OPTIONS.
Call scsiTargetOptionsSet( ) after initializing the SCSI subsystem, but before
initializing the SCSI physical device. For more information about setting and
implementing options, see the reference entry for scsiTargetOptionsSet( ).
!
WARNING: Calling scsiTargetOptionsSet( ) after the physical device has been
initialized may lead to undefined behavior.
The SCSI subsystem performs each SCSI command request as a SCSI transaction.
This requires the SCSI subsystem to select a device. Different SCSI devices require
different amounts of time to respond to a selection; in some cases, the selTimeOut
field may need to be altered from the default.
If a device does not support SCSI messages, the boolean field messages can be set
to FALSE. Similarly, if a device does not support disconnect/reconnect, the
boolean field disconnect can be set to FALSE.
The SCSI subsystem automatically tries to negotiate synchronous data transfer
parameters. However, if a SCSI device does not support synchronous data
transfer, set the maxOffset field to 0. By default, the SCSI subsystem tries to
negotiate the maximum possible REQ/ACK offset and the minimum possible data
transfer period supported by the SCSI controller on the VxWorks target. This is
done to maximize the speed of transfers between two devices. However, speed
depends upon electrical characteristics, like cable length, cable quality, and device
termination; therefore, it may be necessary to reduce the values of maxOffset or
minPeriod for fast transfers.
416
7 I/O System
7.8 Devices in VxWorks
The tagType field defines the type of tagged command queuing desired, using one
of the following macros:
■
■
■
■
SCSI_TAG_UNTAGGED
SCSI_TAG_SIMPLE
SCSI_TAG_ORDERED
SCSI_TAG_HEAD_OF_QUEUE
For more information about the types of tagged command queuing available, see
the ANSI X3T9-I/O Interface Specification Small Computer System Interface
(SCSI-2).
The maxTags field sets the maximum number of command tags available for a
particular SCSI device.
Wide data transfers with a SCSI target device are automatically negotiated upon
initialization by the SCSI subsystem. Wide data transfer parameters are always
negotiated before synchronous data transfer parameters, as specified by the SCSI
ANSI specification, because a wide negotiation resets any prior negotiation of
synchronous parameters. However, if a SCSI device does not support wide
parameters and there are problems initializing that device, you must set the
xferWidth field to 0. By default, the SCSI subsystem tries to negotiate the
maximum possible transfer width supported by the SCSI controller on the
VxWorks target in order to maximize the default transfer speed between the two
devices. For more information on the actual routine call, see the reference entry for
scsiTargetOptionsSet( ).
SCSI Configuration Examples
The following examples show some possible configurations for different SCSI
devices. Example 7-5 is a simple block device configuration setup. Example 7-6
involves selecting special options and demonstrates the use of
scsiTargetOptionsSet( ). Example 7-7 configures a SCSI device for synchronous
data transfer. Example 7-8 shows how to configure the SCSI bus ID. These
examples can be embedded either in the usrScsiConfig( ) routine or in a
user-defined SCSI configuration function.
Example 7-5
Configuring SCSI Drivers
In the following example, usrScsiConfig( ) was modified to reflect a new system
configuration. The new configuration has a SCSI disk with a bus ID of 4 and a
Logical Unit Number (LUN) of 0 (zero). The disk is configured with a dosFs file
system (with a total size of 0x20000 blocks) and a rawFs file system (spanning the
remainder of the disk).
417
7
VxWorks
Kernel Programmer's Guide, 6.6
The following usrScsiConfig( ) code reflects this modification.
/* configure Winchester at busId = 4, LUN = 0 */
if ((pSpd40 = scsiPhysDevCreate (pSysScsiCtrl, 4, 0, 0, NONE, 0, 0, 0))
== (SCSI_PHYS_DEV *) NULL)
{
SCSI_DEBUG_MSG ("usrScsiConfig: scsiPhysDevCreate failed.\n");
}
else
{
/* create block devices - one for dosFs and one for rawFs */
if (((pSbd0 = scsiBlkDevCreate (pSpd40, 0x20000, 0)) == NULL) ||
((pSbd1 = scsiBlkDevCreate (pSpd40, 0, 0x20000)) == NULL))
{
return (ERROR);
}
/* initialize both dosFs and rawFs file systems */
if ((xbdBlkDevCreate (pSbd0, "/sd0") == NULL) ||
(xbdBlkDevCreate (pSbd1,"/sd1") == NULL)
{
return (ERROR);
}
}
If problems with your configuration occur, insert the following lines at the
beginning of usrScsiConfig( ) to obtain further information on SCSI bus activity.
#if FALSE
scsiDebug = TRUE;
scsiIntsDebug = TRUE;
#endif
Do not declare the global variables scsiDebug and scsiIntsDebug locally. They
can be set or reset from the shell.
418
7 I/O System
7.8 Devices in VxWorks
Example 7-6
Configuring a SCSI Disk Drive with Asynchronous Data Transfer and No Tagged Command
Queuing
In this example, a SCSI disk device is configured without support for synchronous
data transfer and tagged command queuing. The scsiTargetOptionsSet( ) routine
is used to turn off these features. The SCSI ID of this disk device is 2, and the LUN
is 0:
int
SCSI_OPTIONS
int
which;
option;
devBusId;
devBusId = 2;
which = SCSI_SET_OPT_XFER_PARAMS | SCSI_SET_OPT_TAG_PARAMS;
option.maxOffset = SCSI_SYNC_XFER_ASYNC_OFFSET;
/* => 0 defined in scsi2Lib.h */
option.minPeriod = SCSI_SYNC_XFER_MIN_PERIOD; /* defined in scsi2Lib.h */
option.tagType = SCSI_TAG_UNTAGGED;
/* defined in scsi2Lib.h */
option.maxTag = SCSI_MAX_TAGS;
if (scsiTargetOptionsSet (pSysScsiCtrl, devBusId, &option, which) == ERROR)
{
SCSI_DEBUG_MSG ("usrScsiConfig: could not set options\n", 0, 0, 0, 0,
0, 0);
return (ERROR);
}
/* configure SCSI disk drive at busId = devBusId, LUN = 0 */
if ((pSpd20 = scsiPhysDevCreate (pSysScsiCtrl, devBusId, 0, 0, NONE, 0, 0,
0)) == (SCSI_PHYS_DEV *) NULL)
{
SCSI_DEBUG_MSG ("usrScsiConfig: scsiPhysDevCreate failed.\n");
return (ERROR);
}
419
7
VxWorks
Kernel Programmer's Guide, 6.6
Example 7-7
Configuring a SCSI Disk for Synchronous Data Transfer with Non-Default Offset and Period
Values
In this example, a SCSI disk drive is configured with support for synchronous data
transfer. The offset and period values are user-defined and differ from the driver
default values. The chosen period is 25, defined in SCSI units of 4 ns. Thus, the
period is actually 4 * 25 = 100 ns. The synchronous offset is chosen to be 2. Note
that you may need to adjust the values depending on your hardware environment.
int
SCSI_OPTIONS
int
which;
option;
devBusId;
devBusId = 2;
which = SCSI_SET_IPT_XFER_PARAMS;
option.maxOffset = 2;
option.minPeriod = 25;
if (scsiTargetOptionsSet (pSysScsiCtrl, devBusId &option, which) ==
ERROR)
{
SCSI_DEBUG_MSG ("usrScsiConfig: could not set options\n",
0, 0, 0, 0, 0, 0)
return (ERROR);
}
/* configure SCSI disk drive at busId = devBusId, LUN = 0 */
if ((pSpd20 = scsiPhysDevCreate (pSysScsiCtrl, devBusId, 0, 0, NONE,
0, 0, 0)) == (SCSI_PHYS_DEV *) NULL)
{
SCSI_DEBUG_MSG ("usrScsiConfig: scsiPhysDevCreate failed.\n")
return (ERROR);
}
Example 7-8
Changing the Bus ID of the SCSI Controller
To change the bus ID of the SCSI controller, modify sysScsiInit( ) in sysScsi.c. Set
the SCSI bus ID to a value between 0 and 7 in the call to xxxCtrlInitScsi2( ), where
xxx is the controller name. The default bus ID for the SCSI controller is 7.
Troubleshooting
■
Incompatibilities Between SCSI-1 and SCSI-2
Applications written for SCSI-1 may not execute for SCSI-2 because data
structures in scsi2Lib.h, such as SCSI_TRANSACTION and SCSI_PHYS_DEV,
have changed. This applies only if the application used these structures
directly.
420
7 I/O System
7.8 Devices in VxWorks
If this is the case, you can choose to configure only the SCSI-1 level of support,
or you can modify your application according to the data structures in
scsi2Lib.h. In order to set new fields in the modified structure, some
applications may simply need to be recompiled, and some applications will
have to be modified and then recompiled.
■
SCSI Bus Failure
If your SCSI bus hangs, it could be for a variety of reasons. Some of the more
common are:
–
Your cable has a defect. This is the most common cause of failure.
–
The cable exceeds the cumulative maximum length of 6 meters specified
in the SCSI-2 standard, thus changing the electrical characteristics of the
SCSI signals.
–
The bus is not terminated correctly. Consider providing termination
power at both ends of the cable, as defined in the SCSI-2 ANSI
specification.
–
The minimum transfer period is insufficient or the REQ/ACK offset is too
great. Use scsiTargetOptionsSet( ) to set appropriate values for these
options.
–
The driver is trying to negotiate wide data transfers on a device that does
not support them. In rejecting wide transfers, the device-specific driver
cannot handle this phase mismatch. Use scsiTargetOptionsSet( ) to set the
appropriate value for the xferWidth field for that particular SCSI device.
421
7
VxWorks
Kernel Programmer's Guide, 6.6
7.9 Differences Between VxWorks and Host System I/O
Most commonplace uses of I/O in VxWorks are completely source-compatible
with I/O in UNIX and Windows. However, note the following differences:
■
Device Configuration
In VxWorks, device drivers can be installed and removed dynamically. But
only in the kernel space.
■
File Descriptors
In VxWorks, file descriptors are unique to the kernel and to each process—as
in UNIX and Windows. The kernel and each process has its own universe of
file descriptors, distinct from each other. When the process is created, its
universe of file descriptors is initially populated by duplicating the file
descriptors of its creator. (This applies only when the creator is a process. If the
creator is a kernel task, only the three standard I/O descriptors 0, 1 and 2 are
duplicated.) Thereafter, all open, close, or dup activities affect only that
process’ universe of descriptors.
In kernel and in each process, file descriptors are global to that entity, meaning
that they are accessible by any task running in it.
In the kernel, however, standard input, standard output, and standard error
(0, 1, and 2) can be made task specific.
For more information see 7.4.1 File Descriptors, p.365 and 7.4.3 Standard I/O
Redirection, p.367.
■
I/O Control
The specific parameters passed to ioctl( ) functions can differ between UNIX
and VxWorks.
■
Driver Routines
In UNIX, device drivers execute in system mode and cannot be preempted. In
VxWorks, driver routines can be preempted because they execute within the
context of the task that invoked them.
422
7 I/O System
7.10 Internal I/O System Structure
7.10 Internal I/O System Structure
The VxWorks I/O system differs from most I/O systems in the way that the work
of performing user I/O requests is distributed between the device-independent
I/O system and the device drivers themselves.
In many systems, the device driver supplies a few routines to perform low-level
I/O functions such as reading a sequence of bytes from, or writing them to,
character-oriented devices. The higher-level protocols, such as communications
protocols on character-oriented devices, are implemented in the
device-independent part of the I/O system. The user requests are heavily
processed by the I/O system before the driver routines get control.
While this approach is designed to make it easy to implement drivers and to
ensure that devices behave as much alike as possible, it has several drawbacks. The
driver writer is often seriously hampered in implementing alternative protocols
that are not provided by the existing I/O system. In a real-time system, it is
sometimes desirable to bypass the standard protocols altogether for certain
devices where throughput is critical, or where the device does not fit the standard
model.
In the VxWorks I/O system, minimal processing is done on user I/O requests
before control is given to the device driver. The VxWorks I/O system acts as a
switch to route user requests to appropriate driver-supplied routines. Each driver
can then process the raw user requests as appropriate to its devices. In addition,
however, several high-level subroutine libraries are available to driver writers that
implement standard protocols for both character- and block-oriented devices.
Thus the VxWorks I/O system provides the best of both worlds: while it is easy to
write a standard driver for most devices with only a few pages of device-specific
code, driver writers are free to execute the user requests in nonstandard ways
where appropriate.
There are two fundamental types of device: block and character (or non-block; see
Figure 7-2). Block devices are used for storing file systems. They are random access
devices where data is transferred in blocks. Examples of block devices include
hard and floppy disks. Character devices are typically of the tty/sio type.
As discussed in earlier sections, the three main elements of the VxWorks I/O
system are drivers, devices, and files. The following sections describe these
elements in detail. The discussion focuses on character drivers; however, much of
it is applicable to block devices. Because block drivers must interact with VxWorks
file systems, they use a slightly different organization.
423
7
VxWorks
Kernel Programmer's Guide, 6.6
NOTE: This discussion is designed to clarify the structure of VxWorks I/O
facilities and to highlight some considerations relevant to writing I/O drivers for
VxWorks. For detailed information about writing device drivers, see the VxWorks
Device Driver Developer’s Guide.
Example 7-9 shows the abbreviated code for a hypothetical driver that is used as
an example throughout the following discussions. This example driver is typical
of drivers for character-oriented devices.
In VxWorks, each driver has a short, unique abbreviation, such as net or tty, which
is used as a prefix for each of its routines. The abbreviation for the example driver
is xx.
Example 7-9
Hypothetical Driver
/*
* xxDrv - driver initialization routine
* xxDrv() init’s the driver. It installs the driver via iosDrvInstall.
* It may allocate data structures, connect ISRs, and initialize hardware
*/
STATUS xxDrv ()
{
xxDrvNum = iosDrvInstall (xxCreat, 0, xxOpen, 0, xxRead, xxWrite, xxIoctl)
;
(void) intConnect (intvec, xxInterrupt, ...);
...
}
/************************************************************************
* xxDevCreate - device creation routine
*
* Called to add a device called to be svced by this driver. Other
* driver-dependent arguments may include buffer sizes, device addresses.
* The routine adds the device to the I/O system by calling iosDevAdd.
* It may also allocate and initialize data structures for the device,
* initialize semaphores, initialize device hardware, and so on.
*/
STATUS xxDevCreate (name, ...)
char * name;
...
{
status = iosDevAdd (xxDev, name, xxDrvNum);
...
}
424
7 I/O System
7.10 Internal I/O System Structure
/*
*
* The following routines implement the basic I/O functions.
* The xxOpen() return value is meaningful only to this driver,
* and is passed back as an argument to the other I/O routines.
*/
int xxOpen (xxDev, remainder, mode)
XXDEV * xxDev;
char * remainder;
int mode;
{
/* serial devices should have no file name part */
7
if (remainder[0] != 0)
return (ERROR);
else
return ((int) xxDev);
}
int xxRead (xxDev, buffer, nBytes)
XXDEV * xxDev;
char * buffer;
int nBytes;
...
int xxWrite (xxDev, buffer, nBytes)
...
int xxIoctl (xxDev, requestCode, arg)
...
/*
*
*
*
*
*
*
*/
xxInterrupt - interrupt service routine
Most drivers have routines that handle interrupts from the devices
serviced by the driver. These routines are connected to the interrupts
by calling intConnect (usually in xxDrv above). They can receive a
single argument, specified in the call to intConnect (see intLib).
VOID xxInterrupt (arg)
...
425
VxWorks
Kernel Programmer's Guide, 6.6
7.10.1 Drivers
A driver for a non-block device generally implements the seven basic I/O
functions—creat( ), remove( ), open( ), close( ), read( ), write( ), and ioctl( )—for a
particular kind of device. The driver implements these general functions with
corresponding device-specific routines that are installed with iosDrvInstall( ).
NOTE: Only VxBus-compatible drivers can be used with the symmetric
multiprocessing (SMP) configuration of VxWorks. For general information about
VxWorks SMP and about migration, see 15. VxWorks SMP and 15.15 Migrating
Code to VxWorks SMP, p.702.
Not all of the general I/O functions are implemented if they are not supported by
a particular device. For example, remove( ) is usually not supported for devices
that are not used with file systems.
If any of the seven basic I/O routines are not implemented by a driver, a null
function pointer should be used for the corresponding iosDrvInstall( ) parameter
when the driver is installed. Any call to a routine that is not supported will then
fail and return an ENOTSUP error.
Drivers may (optionally) allow tasks to wait for activity on multiple file
descriptors. This functionality is implemented with the driver’s ioctl( ) routine; see
Implementing select( ), p.441.
A driver for a block device interfaces with a file system, rather than directly with
the I/O system. The file system in turn implements most I/O functions. The driver
need only supply routines to read and write blocks, reset the device, perform I/O
control, and check device status.
When an application invokes one of the basic I/O functions, the I/O system routes
the request to the appropriate routine of a specific driver, as described in the
following sections. The driver’s routine runs in the calling task’s context, as though
it were called directly from the application. Thus, the driver is free to use any
facilities normally available to tasks, including I/O to other devices. This means
that most drivers have to use some mechanism to provide mutual exclusion to
critical regions of code. The usual mechanism is the semaphore facility provided
in semLib.
In addition to the routines that implement the seven basic I/O functions, drivers
also have three other routines:
■
426
An initialization routine that installs the driver in the I/O system, connects to
any interrupts used by the devices serviced by the driver, and performs any
necessary hardware initialization. This routine is typically named xxDrv( ).
7 I/O System
7.10 Internal I/O System Structure
■
A routine to add devices that are to be serviced by the driver to the I/O system.
This routine is typically named xxDevCreate( ).
■
Interrupt-level routines that are connected to the interrupts of the devices
serviced by the driver.
The Driver Table and Installing Drivers
The function of the I/O system is to route user I/O requests to the appropriate
routine of the appropriate driver. The I/O system does this by maintaining a table
that contains the address of each routine for each driver. Drivers are installed
dynamically by calling the I/O system internal routine iosDrvInstall( ). The
arguments to this routine are the addresses of the seven I/O routines for the new
driver. The iosDrvInstall( ) routine enters these addresses in a free slot in the
driver table and returns the index of this slot. This index is known as the driver
number and is used subsequently to associate particular devices with the driver.
Null (0) addresses can be specified for any of the seven basic I/O routines that are
not supported by a device. For example, remove( ) is usually not supported for
non-file-system devices, and a null is specified for the driver’s remove function.
When a user I/O call matches a null driver routine, the call fails and an ENOTSUP
error is returned.
VxWorks file systems (such as dosFsLib) contain their own entries in the driver
table, which are created when the file system library is initialized.
427
7
VxWorks
Kernel Programmer's Guide, 6.6
Figure 7-3
Example – Driver Initialization for Non-Block Devices
DRIVER CALL:
drvnum = iosDrvInstall (xxCreat, 0, xxOpen, 0, xxRead, xxWrite, xxIoctl);
[1] Driver’s install routine specifies driver
routines for seven I/O functions.
[2] I/O system locates next
available slot in driver table.
[4] I/O system returns
driver number
(drvnum = 2).
create
0
1
2 xxCreat
3
4
DRIVER TABLE:
0
open
close
xxOpen
0
read
write
ioctl
xxReadxxWrite xxIoctl
delete
[3] I/O system enters driver
routines in driver table.
Example of Installing a Driver
Figure 7-3 shows the actions taken by the example driver and by the I/O system
when the initialization routine xxDrv( ) runs.
The driver calls iosDrvInstall( ), specifying the addresses of the driver’s routines
for the seven basic I/O functions. Then, the I/O system:
1.
Locates the next available slot in the driver table, in this case slot 2.
2.
Enters the addresses of the driver routines in the driver table.
3.
Returns the slot number as the driver number of the newly installed driver.
428
7 I/O System
7.10 Internal I/O System Structure
7.10.2 Devices
Some drivers are capable of servicing many instances of a particular kind of device.
For example, a single driver for a serial communications device can often handle
many separate channels that differ only in a few parameters, such as device
address.
In the VxWorks I/O system, devices are defined by a data structure called a device
header (DEV_HDR). This data structure contains the device name string and the
driver number for the driver that services this device. The device headers for all
the devices in the system are kept in a memory-resident linked list called the device
list. The device header is the initial part of a larger structure determined by the
individual drivers. This larger structure, called a device descriptor, contains
additional device-specific data such as device addresses, buffers, and semaphores.
The Device List and Adding Devices
Non-block devices are added to the I/O system dynamically by calling the internal
I/O routine iosDevAdd( ). The arguments to iosDevAdd( ) are the address of the
device descriptor for the new device, the device’s name, and the driver number of
the driver that services the device. The device descriptor specified by the driver
can contain any necessary device-dependent information, as long as it begins with
a device header. The driver does not need to fill in the device header, only the
device-dependent information. The iosDevAdd( ) routine enters the specified
device name and the driver number in the device header and adds it to the system
device list.
To add a block device to the I/O system, call the device initialization routine for
the file system required on that device—for example, dosFsDevCreate( ). The
device initialization routine then calls iosDevAdd( ) automatically.
The routine iosDevFind( ) can be used to locate the device structure (by obtaining
a pointer to the DEV_HDR, which is the first member of that structure) and to
verify that a device name exists in the table.
The following is an example using iosDevFind( ):
429
7
VxWorks
Kernel Programmer's Guide, 6.6
char * pTail;
/* pointer to tail of devName */
char devName[6] = "DEV1:";
/* name of device */
DOS_VOLUME_DESC * pDosVolDesc;
/* first member is DEV_HDR */
...
pDosVolDesc = iosDevFind(devName, (char**)&pTail);
if (NULL == pDosVolDesc)
{
/* ERROR: device name does not exist and no default device */
}
else
{
/*
* pDosVolDesc is a valid DEV_HDR pointer
* and pTail points to beginning of devName.
* Check devName against pTail to determine if it is
* the default name or the specified devName.
*/
}
Example of Adding Devices
In Figure 7-4, the example driver’s device creation routine xxDevCreate( ) adds
devices to the I/O system by calling iosDevAdd( ).
430
7 I/O System
7.10 Internal I/O System Structure
Figure 7-4
Example – Addition of Devices to I/O System
DRIVER CALLS:
status = iosDevAdd (dev0, "/xx0", drvnum);
status = iosDevAdd (dev1, "/xx1", drvnum);
I/O system adds device descriptors
to device list. Each descriptor contains
device name and driver number (in this
case 2) and any device-specific data.
7
DEVICE LIST:
"/dk0/"
1
DRIVER TABLE:
"/xx0"
2
"/xx1"
2
devicedependent
data
devicedependent
data
create delete
open
close
read
write
ioctl
0
1
2
3
4
Deleting Devices
A device can be deleted with iosDevDelete( ) and the associated driver removed
with iosDrvRemove( ).
Note that a device-deletion operation causes the file descriptors that are open on
the device to be invalidated, but not closed. The file descriptors can only be closed
by an explicit act on the part of an application. If this were not the case, and file
descriptors were closed automatically by the I/O system, the descriptors could be
reassigned to new files while they were still being used by an application that was
unaware of the deletion of the device. The new files could then be accessed
unintentionally by an application attempting to use the files associated with the
431
VxWorks
Kernel Programmer's Guide, 6.6
deleted device, as well as by an application that was correctly using the new files.
This would result in I/O errors and possible device data corruption.
Because the file descriptors of a device that has been deleted are invalid, any
subsequent I/O calls that use them—except close( )—will fail. The behavior of the
I/O routines in this regard is as follows:
■
close( ) releases the file descriptor at I/O system level and the driver close
routine is not called.
■
read( ), write( ), and ioctl( ) fail with error ENXIO (no such device or address).
■
While open( ), remove( ), and create( ) do not take an open file descriptor as
input, they fail because the device name is no longer in the device list.
Note that even if a device is deleted and immediately added again with the same
device name, the file descriptors that were invalidated with the deletion are not
restored to valid status. The behavior of the I/O calls on the associated file
descriptors is the same as if the device had not been added again.
Applications that are likely to encounter device deletion should be sure to check
for ENXIO errors from read( ), write( ), and ioctl( ) calls, and to then close the
relevant file descriptors.
Using Callback Routines to Manage Device Deletion
For situations in which devices are dynamically installed and deleted, the
iosDevDelCallback( ) routine provides the means for calling a post-deletion
handler after all driver invocations are exited.
A common use of a device deletion callback is to prevent a race condition that
would result from a device descriptor being deleted in one thread of execution
while it was still being used in another.
A device descriptor belongs to an application, and the I/O system cannot control
its creation and release. It is a user data structure with DEV_HDR data structure
embedded at the front of it, followed by any specific member of the device. Its
pointer can be used to pass into any iosDevXyz( ) routine as a DEV_HDR pointer,
or used as the device descriptor for user device handling.
When a device is deleted, an application should not immediately release the device
descriptor memory after iosDevDelete( ) and iosDrvRemove( ) calls because a
driver invocation of the deleted device might still be in process. If the device
descriptor is deleted while it is still in use by a driver routine, serious errors could
occur.
432
7 I/O System
7.10 Internal I/O System Structure
For example, the following would produce a race condition: task A invokes the
driver routine xyzOpen( ) by a call to open( ) and the xyzOpen( ) call does not
return before task B deletes the device and releases the device descriptor.
However, if descriptor release is not performed by task B, but by a callback
function installed with iosDevDelCallback( ), then the release occurs only after
task A’s invocation of the driver routine has finished.
A device callback routine is called immediately when a device is deleted with
iosDevDelete( ) or iosDrvRemove( ) as long as no invocations of the associated
driver are operative (that is, the device driver reference counter is zero).
Otherwise, the callback routine is not executed until the last driver call exits (and
the device driver reference counter reaches zero).
A device deletion callback routine should be called with only one parameter, the
pointer to the DEV_HDR data structure of the device in question. For example:
devDeleteCallback(pDevHdr)
The callback should be installed with iosDevDelCallback( ) after the
iosDevAdd( ) call.
The following code fragments illustrate callback use. The file system device
descriptor pVolDesc is installed into the I/O device list. Its device deletion
callback, fsVolDescRelease( ) performs the post-deletion processing, including
releasing memory allocated for the device volume descriptor.
void fsVolDescRelease
(
FS_VOLUME_DESC * pVolDesc
)
{
. . . . . .
free (pVolDesc->pFsemList);
free (pVolDesc->pFhdlList);
free (pVolDesc->pFdList);
. . . . . .
}
STATUS fsDevCreate
(
char * pDevName,
/* device name */
device_t device,
/* underlying block device */
u_int
maxFiles,
/* max no. of simultaneously open files */
u_int
devCreateOptions /* write option & volume integrity */
)
{
FS_VOLUME_DESC *pVolDesc = NULL; /* volume descriptor ptr */
. . . . . .
pVolDesc = (FS_VOLUME_DESC *) malloc (sizeof (*pVolDesc));
pVolDesc->device = device;
433
7
VxWorks
Kernel Programmer's Guide, 6.6
. . . . . .
if (iosDevAdd((void *)pVolDesc, pDevName, fsDrvNum ) == ERROR)
{
pVolDesc->magic = NONE;
goto error_iosadd;
}
/* Device deletion callback installed to release memory resource. */
iosDevDelCallback((DEV_HDR *) pVolDesc, (FUNCPTR) fsVolDescRelease);
. . . . . .
}
STATUS fsDevDelete
(
FS_VOLUME_DESC *pVolDesc
/* pointer to volume descriptor */
)
{
. . . . . .
/*
* Delete the file system device from I/O device list. Callback
* fsVolDescRelease will be called from now on at a
* safe time by I/O system.
*/
iosDevDelete((DEV_HDR *) pVolDesc);
. . . . . .
}
The application should check the error returned by a deleted device, as follows:
if (write (fd, (char *)buffer, nbytes) == ERROR)
{
if (errno == ENXIO)
{
/* Device is deleted. fd must be closed by application. */
close(fd);
}
else
{
/* write failure due to other reason. Do some error dealing. */
. . . . . .
}
}
434
7 I/O System
7.10 Internal I/O System Structure
7.10.3 File Descriptors
Several file descriptors can be open to a single device at one time. A device driver
can maintain additional information associated with a file descriptor beyond the
I/O system’s device information. In particular, devices on which multiple files can
be open at one time have file-specific information (for example, file offset)
associated with each file descriptor. You can also have several file descriptors open
to a non-block device, such as a tty; typically there is no additional information,
and thus writing on any of the file descriptors produces identical results.
7
File Descriptor Table
Files are opened with open( ) or creat( ). The I/O system searches the device list
for a device name that matches the filename (or an initial substring) specified by
the caller. If a match is found, the I/O system uses the driver number contained in
the corresponding device header to locate and call the driver’s open routine in the
driver table.
The I/O system must establish an association between the file descriptor used by
the caller in subsequent I/O calls, and the driver that services it. Additionally, the
driver must associate some data structure per descriptor. In the case of non-block
devices, this is usually the device descriptor that was located by the I/O system.
The I/O system maintains these associations in a table called the file descriptor table.
This table contains the driver number and an additional driver-determined 4-byte
value. The driver value is the internal descriptor returned by the driver’s open
routine, and can be any value the driver requires to identify the file. In subsequent
calls to the driver’s other I/O functions (read( ), write( ), ioctl( ), and close( )), this
value is supplied to the driver in place of the file descriptor in the application-level
I/O call.
435
VxWorks
Kernel Programmer's Guide, 6.6
Example of Opening a File
In Figure 7-5 and Figure 7-6, a user calls open( ) to open the file /xx0. The I/O
system takes the following series of actions:
1.
It searches the device list for a device name that matches the specified filename
(or an initial substring). In this case, a complete device name matches.
2.
It reserves a slot in the file descriptor table and creates a new file descriptor
object, which is used if the open is successful.
3.
It then looks up the address of the driver’s open routine, xxOpen( ), and calls
that routine. Note that the arguments to xxOpen( ) are transformed by the I/O
system from the user’s original arguments to open( ). The first argument to
xxOpen( ) is a pointer to the device descriptor the I/O system located in the
full filename search. The next parameter is the remainder of the filename
specified by the user, after removing the initial substring that matched the
device name. In this case, because the device name matched the entire
filename, the remainder passed to the driver is a null string. The driver is free
to interpret this remainder in any way it wants. In the case of block devices,
this remainder is the name of a file on the device. In the case of non-block
devices like this one, it is usually an error for the remainder to be anything but
the null string. The third parameter is the file access flag, in this case
O_RDONLY; that is, the file is opened for reading only. The last parameter is
the mode, as passed to the original open( ) routine.
4.
It executes xxOpen( ), which returns a value that subsequently identifies the
newly opened file. In this case, the value is the pointer to the device descriptor.
This value is supplied to the driver in subsequent I/O calls that refer to the file
being opened. Note that if the driver returns only the device descriptor, the
driver cannot distinguish multiple files opened to the same device. In the case
of non-block device drivers, this is usually appropriate.
5.
The I/O system then enters the driver number and the value returned by
xxOpen( ) in the new file descriptor object.
Again, the value entered in the file descriptor object has meaning only for the
driver, and is arbitrary as far as the I/O system is concerned.
6.
436
Finally, it returns to the user the index of the slot in the file descriptor table, in
this case 3.
7 I/O System
7.10 Internal I/O System Structure
Figure 7-5
Example: Call to I/O Routine open( ) [Part 1]
USER CALL:
DRIVER CALL:
fd = open ("/xx0", O_RDONLY, 0);
xxdev = xxOpen (xxdev, "", O_RDONLY, 0);
[1] I/O system finds
name in device list.
[2] I/O system reserves
[3] I/O system calls
driver’s open routine
a slot in the file descriptor
with pointer to
table.
device descriptor.
pDevHdr value
FILE DESCRIPTOR TABLE:
0
1
2
3
4
DEVICE LIST:
"/dk0/"
1
"/xx0"
2
"/xx1"
2
devicedependent
data
DRIVER TABLE:
create delete
0
1
2
3
4
open
close
read
write
ioctl
xxOpen
437
7
VxWorks
Kernel Programmer's Guide, 6.6
Figure 7-6
Example: Call to I/O Routine open( ) [Part 2]
USER CALL:
DRIVER CALL:
fd = open ("/xx0", O_RDONLY, 0);
xxdev = xxOpen (xxdev, "", O_RDONLY, 0);
[6] I/O system returns
index in table of
new open file (fd = 3).
[5] I/O system enters
driver number and
identifying value in
reserved table slot.
FILE DESCRIPTOR TABLE:
[4] Driver returns any
identifying value, in
this case the pointer to
the device descriptor.
drvnum value
0
1
2
3
4
2
xxdev
DEVICE LIST:
"/dk0/"
1
"/xx0"
2
"/xx1"
2
devicedependent
data
create delete
DRIVER TABLE:
0
1
2
3
4
438
open
close
read
write
ioctl
7 I/O System
7.10 Internal I/O System Structure
Example of Reading Data from the File
In Figure 7-7, the user calls read( ) to obtain input data from the file. The specified
file descriptor is the index into the file descriptor table for this file. The I/O system
uses the driver number contained in the table to locate the driver’s read routine,
xxRead( ). The I/O system calls xxRead( ), passing it the identifying value in the
file descriptor table that was returned by the driver’s open routine, xxOpen( ).
Again, in this case the value is the pointer to the device descriptor. The driver’s
read routine then does whatever is necessary to read data from the device. The
process for user calls to write( ) and ioctl( ) follow the same procedure.
439
7
VxWorks
Kernel Programmer's Guide, 6.6
Figure 7-7
Example: Call to I/O Routine read( )
USER CALL:
DRIVER CALL:
n = read (fd, buf, len);
n = xxRead (xxdev, buf, len);
I/O system transforms the user’s I/O
routine calls into driver routine calls
replacing the fd with the value returned
by the driver’s open routine, xxOpen( ).
drvnum value
FILE DESCRIPTOR TABLE:
0
1
2
3
4
2
xxdev
DEVICE LIST:
"/dk0/"
1
"/xx0"
2
"/xx1"
2
devicedependent
data
DRIVER TABLE:
440
create remove open
0
1
2
3
4
close
read
xxRead
write
ioctl
7 I/O System
7.10 Internal I/O System Structure
Example of Closing a File
The user terminates the use of a file by calling close( ). As in the case of read( ), the
I/O system uses the driver number contained in the file descriptor table to locate
the driver’s close routine. In the example driver, no close routine is specified; thus
no driver routines are called. Instead, the I/O system marks the slot in the file
descriptor table as being available. Any subsequent references to that file
descriptor cause an error. Subsequent calls to open( ) can reuse that slot.
7
Implementing select( )
Supporting select( ) in your driver allows tasks to wait for input from multiple
devices or to specify a maximum time to wait for the device to become ready for
I/O. Writing a driver that supports select( ) is simple, because most of the
functionality is provided in selectLib. You might want your driver to support
select( ) if any of the following is appropriate for the device:
■
The tasks want to specify a timeout to wait for I/O from the device. For
example, a task might want to time out on a UDP socket if the packet never
arrives.
■
The driver supports multiple devices, and the tasks want to wait
simultaneously for any number of them. For example, multiple pipes might be
used for different data priorities.
■
The tasks want to wait for I/O from the device while also waiting for I/O from
another device. For example, a server task might use both pipes and sockets.
To implement select( ), the driver must keep a list of tasks waiting for device
activity. When the device becomes ready, the driver unblocks all the tasks waiting
on the device.
For a device driver to support select( ), it must declare a SEL_WAKEUP_LIST
structure (typically declared as part of the device descriptor structure) and
initialize it by calling selWakeupListInit( ). This is done in the driver’s
xxDevCreate( ) routine. When a task calls select( ), selectLib calls the driver’s
ioctl( ) routine with the function FIOSELECT or FIOUNSELECT. If ioctl( ) is called
with FIOSELECT, the driver must do the following:
441
VxWorks
Kernel Programmer's Guide, 6.6
1.
Add the SEL_WAKEUP_NODE (provided as the third argument of ioctl( )) to
the SEL_WAKEUP_LIST by calling selNodeAdd( ).
2.
Use the routine selWakeupType( ) to check whether the task is waiting for
data to read from the device (SELREAD) or if the device is ready to be written
(SELWRITE).
3.
If the device is ready (for reading or writing as determined by
selWakeupType( )), the driver calls the routine selWakeup( ) to make sure
that the select( ) call in the task does not pend. This avoids the situation where
the task is blocked but the device is ready.
If ioctl( ) is called with FIOUNSELECT, the driver calls selNodeDelete( ) to remove
the provided SEL_WAKEUP_NODE from the wakeup list.
When the device becomes available, selWakeupAll( ) is used to unblock all the
tasks waiting on this device. Although this typically occurs in the driver’s ISR, it
can also occur elsewhere. For example, a pipe driver might call selWakeupAll( )
from its xxRead( ) routine to unblock all the tasks waiting to write, now that there
is room in the pipe to store the data. Similarly the pipe’s xxWrite( ) routine might
call selWakeupAll( ) to unblock all the tasks waiting to read, now that there is data
in the pipe.
442
7 I/O System
7.10 Internal I/O System Structure
Example 7-10
Driver Code Using the Select Facility
/* This code fragment shows how a driver might support select(). In this
* example, the driver unblocks tasks waiting for the device to become ready
* in its interrupt service routine.
*/
/* myDrvLib.h - header file for driver */
typedef struct
/* MY_DEV */
{
DEV_HDR
devHdr;
BOOL
myDrvDataAvailable;
BOOL
myDrvRdyForWriting;
SEL_WAKEUP_LIST selWakeupList;
} MY_DEV;
/*
/*
/*
/*
device header */
data is available to read */
device is ready to write */
list of tasks pended in select */
/* myDrv.c - code fragments for supporting select() in a driver */
#include
#include
/* First create and initialize the device */
STATUS myDrvDevCreate
(
char * name,
)
{
MY_DEV * pMyDrvDev;
... additional driver code ...
/* name of device to create */
/* pointer to device descriptor*/
/* allocate memory for MY_DEV */
pMyDrvDev = (MY_DEV *) malloc (sizeof MY_DEV);
... additional driver code ...
/* initialize MY_DEV */
pMyDrvDev->myDrvDataAvailable=FALSE
pMyDrvDev->myDrvRdyForWriting=FALSE
/* initialize wakeup list */
selWakeupListInit (&pMyDrvDev->selWakeupList);
... additional driver code ...
}
/* ioctl function to request reading or writing */
STATUS myDrvIoctl
(
MY_DEV * pMyDrvDev,
/* pointer to device descriptor */
443
7
VxWorks
Kernel Programmer's Guide, 6.6
int
request,
int
arg
)
{
... additional driver code ...
/* ioctl function */
/* where to send answer */
switch (request)
{
... additional driver code ...
case FIOSELECT:
/* add node to wakeup list */
selNodeAdd (&pMyDrvDev->selWakeupList, (SEL_WAKEUP_NODE *) arg);
if (selWakeupType ((SEL_WAKEUP_NODE *) arg) == SELREAD
&& pMyDrvDev->myDrvDataAvailable)
{
/* data available, make sure task does not pend */
selWakeup ((SEL_WAKEUP_NODE *) arg);
}
if (selWakeupType ((SEL_WAKEUP_NODE *) arg) == SELWRITE
&& pMyDrvDev->myDrvRdyForWriting)
{
/* device ready for writing, make sure task does not pend */
selWakeup ((SEL_WAKEUP_NODE *) arg);
}
break;
case FIOUNSELECT:
/* delete node from wakeup list */
selNodeDelete (&pMyDrvDev->selWakeupList, (SEL_WAKEUP_NODE *) arg);
break;
... additional driver code ...
}
}
/* code that actually uses the select() function to read or write */
void myDrvIsr
(
MY_DEV * pMyDrvDev;
)
{
... additional driver code ...
/* if there is data available to read, wake up all pending tasks */
if (pMyDrvDev->myDrvDataAvailable)
selWakeupAll (&pMyDrvDev->selWakeupList, SELREAD);
444
7 I/O System
7.10 Internal I/O System Structure
/* if the device is ready to write, wake up all pending tasks */
if (pMyDrvDev->myDrvRdyForWriting)
selWakeupAll (&pMyDrvDev->selWakeupList, SELWRITE);
}
Cache Coherency
NOTE: The cache facilities described in this section are provided for the
uniprocessor (UP) configuration of VxWorks, some of which are not appropriate—
and not provided—for the symmetric multiprocessor (SMP) configuration. For
more information in this regard, see cacheLib Restrictions, p.708. For general
information about VxWorks SMP and about migration, see 15. VxWorks SMP and
15.15 Migrating Code to VxWorks SMP, p.702.
Drivers written for boards with caches must guarantee cache coherency. Cache
coherency means data in the cache must be in sync, or coherent, with data in RAM.
The data cache and RAM can get out of sync any time there is asynchronous access
to RAM (for example, DMA device access or VMEbus access). Data caches are used
to increase performance by reducing the number of memory accesses. Figure 7-8
shows the relationships between the CPU, data cache, RAM, and a DMA device.
Data caches can operate in one of two modes: writethrough and copyback.
Write-through mode writes data to both the cache and RAM; this guarantees cache
coherency on output but not input. Copyback mode writes the data only to the
cache; this makes cache coherency an issue for both input and output of data.
Figure 7-8
Cache Coherency
CPU
Data Cache
RAM
DMA
Device
445
7
VxWorks
Kernel Programmer's Guide, 6.6
If a CPU writes data to RAM that is destined for a DMA device, the data can first
be written to the data cache. When the DMA device transfers the data from RAM,
there is no guarantee that the data in RAM was updated with the data in the cache.
Thus, the data output to the device may not be the most recent—the new data may
still be sitting in the cache. This data incoherence can be solved by making sure the
data cache is flushed to RAM before the data is transferred to the DMA device.
If a CPU reads data from RAM that originated from a DMA device, the data read
can be from the cache buffer (if the cache buffer for this data is not marked invalid)
and not the data just transferred from the device to RAM. The solution to this data
incoherence is to make sure that the cache buffer is marked invalid so that the data
is read from RAM and not from the cache.
Drivers can solve the cache coherency problem either by allocating cache-safe
buffers (buffers that are marked non-cacheable) or flushing and invalidating cache
entries any time the data is written to or read from the device. Allocating
cache-safe buffers is useful for static buffers; however, this typically requires MMU
support. Non-cacheable buffers that are allocated and freed frequently (dynamic
buffers) can result in large amounts of memory being marked non-cacheable. An
alternative to using non-cacheable buffers is to flush and invalidate cache entries
manually; this allows dynamic buffers to be kept coherent.
The routines cacheFlush( ) and cacheInvalidate( ) are used to manually flush and
invalidate cache buffers. Before a device reads the data, flush the data from the
cache to RAM using cacheFlush( ) to ensure the device reads current data. After
the device has written the data into RAM, invalidate the cache entry with
cacheInvalidate( ). This guarantees that when the data is read by the CPU, the
cache is updated with the new data in RAM.
446
7 I/O System
7.10 Internal I/O System Structure
Example 7-11
DMA Transfer Routine
/* This a sample DMA transfer routine. Before programming the device
* to output the data to the device, it flushes the cache by calling
* cacheFlush(). On a read, after the device has transferred the data,
* the cache entry must be invalidated using cacheInvalidate().
*/
#include
#include
#include
#include "example.h"
void exampleDmaTransfer
/* 1 = READ, 0 = WRITE */
(
UINT8 *pExampleBuf,
int exampleBufLen,
int xferDirection
)
{
if (xferDirection == 1)
{
myDevToBuf (pExampleBuf);
cacheInvalidate (DATA_CACHE, pExampleBuf, exampleBufLen);
}
else
{
cacheFlush (DATA_CACHE, pExampleBuf, exampleBufLen);
myBufToDev (pExampleBuf);
}
}
447
7
VxWorks
Kernel Programmer's Guide, 6.6
It is possible to make a driver more efficient by combining cache-safe buffer
allocation and cache-entry flushing or invalidation. The idea is to flush or
invalidate a cache entry only when absolutely necessary. To address issues of
cache coherency for static buffers, use cacheDmaMalloc( ). This routine initializes
a CACHE_FUNCS structure (defined in cacheLib.h) to point to flush and invalidate
routines that can be used to keep the cache coherent.
The macros CACHE_DMA_FLUSH and CACHE_DMA_INVALIDATE use this
structure to optimize the calling of the flush and invalidate routines. If the
corresponding function pointer in the CACHE_FUNCS structure is NULL, no
unnecessary flush/invalidate routines are called because it is assumed that the
buffer is cache coherent (hence it is not necessary to flush/invalidate the cache
entry manually).
The driver code uses a virtual address and the device uses a physical address.
Whenever a device is given an address, it must be a physical address. Whenever
the driver accesses the memory, it must use the virtual address.
The device driver should use CACHE_DMA_VIRT_TO_PHYS to translate a virtual
address to a physical address before passing it to the device. It may also use
CACHE_DMA_PHYS_TO_VIRT to translate a physical address to a virtual one, but
this process is time-consuming and non-deterministic, and should be avoided
whenever possible.
448
7 I/O System
7.10 Internal I/O System Structure
Example 7-12
Address-Translation Driver
/* The following code is an example of a driver that performs address
* translations. It attempts to allocate a cache-safe buffer, fill it, and
* then write it out to the device. It uses CACHE_DMA_FLUSH to make sure
* the data is current. The driver then reads in new data and uses
* CACHE_DMA_INVALIDATE to guarantee cache coherency.
*/
#include
#include
#include "myExample.h"
STATUS myDmaExample (void)
{
void * pMyBuf;
void * pPhysAddr;
7
/* allocate cache safe buffers if possible */
if ((pMyBuf = cacheDmaMalloc (MY_BUF_SIZE)) == NULL)
return (ERROR);
… fill buffer with useful information …
/* flush cache entry before data is written to device */
CACHE_DMA_FLUSH (pMyBuf, MY_BUF_SIZE);
/* convert virtual address to physical */
pPhysAddr = CACHE_DMA_VIRT_TO_PHYS (pMyBuf);
/* program device to read data from RAM */
myBufToDev (pPhysAddr);
… wait for DMA to complete …
… ready to read new data …
/* program device to write data to RAM */
myDevToBuf (pPhysAddr);
… wait for transfer to complete …
/* convert physical to virtual address */
pMyBuf = CACHE_DMA_PHYS_TO_VIRT (pPhysAddr);
/* invalidate buffer */
CACHE_DMA_INVALIDATE (pMyBuf, MY_BUF_SIZE);
… use data …
/* when done free memory */
if (cacheDmaFree (pMyBuf) == ERROR)
return (ERROR);
return (OK);
}
449
VxWorks
Kernel Programmer's Guide, 6.6
7.11 PCMCIA
A PCMCIA card can be plugged into notebook computers to connect devices such
as modems and external hard drives.2 VxWorks provides PCMCIA facilities for
pcPentium, pcPentium2, and pcPentium3 BSPs and PCMCIA drivers that allow
VxWorks running on these targets to support PCMCIA hardware.
PCMCIA support is at the PCMCIA Release 2.1 level. It does not include socket
services or card services, which are not required by VxWorks. It does include chip
drivers and libraries. The PCMCIA libraries and drivers are also available in
source code form for VxWorks systems based on CPU architectures other than
Intel Pentium.
To include PCMCIA support in your system, configure VxWorks with the
INCLUDE_PCMCIA component. For information about PCMCIA facilities, see the
API references for pcmciaLib and pcmciaShow.
7.12 Peripheral Component Interconnect: PCI
Peripheral Component Interconnect (PCI) is a bus standard for connecting
peripherals to a PC, and is used in Pentium systems, among others. PCI includes
buffers that de-couple the CPU from relatively slow peripherals, allowing them to
operate asynchronously.
For information about PCI facilities, see the API references for pciAutoConfigLib,
pciConfigLib, pciInitLib, and pciConfigShow.
2. PCMCIA stands for Personal Computer Memory Card International Association, and refers
to both the association and the standards that it has developed.
450
8
Local File Systems
8.1 Introduction 452
8.2 File System Monitor 455
8.3 Virtual Root File System: VRFS 457
8.4 Highly Reliable File System: HRFS 459
8.5 MS-DOS-Compatible File System: dosFs 478
8.6 Raw File System: rawFs 505
8.7 CD-ROM File System: cdromFs 510
8.8 Read-Only Memory File System: ROMFS 516
8.9 Target Server File System: TSFS 518
451
VxWorks
Kernel Programmer's Guide, 6.6
8.1 Introduction
VxWorks provides a variety of file systems that are suitable for different types of
applications. The file systems can be used simultaneously, and in most cases in
multiple instances, for a single VxWorks system.
Most VxWorks file systems rely on the extended block device (XBD) facility for a
a standard I/O interface between the file system and device drivers. This standard
interface allows you to write your own file system for VxWorks, and freely mix file
systems and device drivers.
File systems used for removable devices make use of the file system monitor for
automatic detection of device insertion and instantiation of the appropriate file
system on the device.
The relationship between applications, file systems, I/O facilities, device drivers
and hardware devices is illustrated in Figure 8-1. Note that this illustration is
relevant for the HRFS, dosFs, rawFs, and cdromFs file systems. The dotted line
indicates the elements that must be configured and instantiated to create a specific,
functional run-time file system.
452
8 Local File Systems
8.1 Introduction
Figure 8-1
File Systems in a VxWorks System
Application
I/O System
8
File System
HRFS, dosFs, rawFs, cdromFs
XBD Facility
Block Device
SCSI, ATA, RAM disk, Floppy, TrueFFS, and so on
Hardware
For information about the XBD facility, see 7.8.8 Extended Block Device Facility:
XBD, p.402.
453
VxWorks
Kernel Programmer's Guide, 6.6
This chapter discusses the file system monitor and the following VxWorks file
systems, describing how they are organized, configured, and used:
■
VRFS
A virtual root file system for use with applications that require a POSIX root
file system. The VRFS is simply a root directory from which other file systems
and devices can be accessed. See 8.3 Virtual Root File System: VRFS, p.457.
■
HRFS
A POSIX-compliant transactional file system designed for real-time use of
block devices (disks). Can be used on flash memory in conjunction with
TrueFFS and the XBD block wrapper component. See 8.4 Highly Reliable File
System: HRFS, p.459.
■
dosFs
An MS-DOS compatible file system designed for real-time use of block
devices. Can be used with flash memory in conjunction with the TrueFFS and
the XBD block wrapper component. Can also be used with the
transaction-based reliable file system (TRFS) facility. See
8.5 MS-DOS-Compatible File System: dosFs, p.478.
■
rawFS
Provides a simple raw file system that treats an entire disk as a single large file.
See 8.6 Raw File System: rawFs, p.505.
■
cdromFs
Allows applications to read data from CD-ROMs formatted according to the
ISO 9660 standard file system. See 8.7 CD-ROM File System: cdromFs, p.510.
■
ROMFS
Designed for bundling applications and other files with a VxWorks system
image. No storage media is required beyond that used for the VxWorks boot
image. See 8.8 Read-Only Memory File System: ROMFS, p.516.
■
TSFS
Uses the host target server to provide the target with access to files on the host
system. See 8.9 Target Server File System: TSFS, p.518.
For information about the XBD facility, see 7.8.8 Extended Block Device Facility:
XBD, p.402).
454
8 Local File Systems
8.2 File System Monitor
File Systems and Flash Memory
VxWorks can be configured with file-system support for flash memory devices
using TrueFFS and the HRFS or dosFs file system. For more information, see
8.5 MS-DOS-Compatible File System: dosFs, p.478 and 10. Flash File System Support:
TrueFFS.
NOTE: This chapter provides information about facilities available in the VxWorks
kernel. For information about facilities available to real-time processes, see the
VxWorks Application Programmer’s Guide: Local File Systems.
8
8.2 File System Monitor
The file system monitor provides for automatic detection of device insertion, and
instantiation of the appropriate file system on the device. The monitor is required
for all file systems that are used with the extended block device (XBD) facility. It is
provided with the INCLUDE_FS_MONITOR component.
The file systems that require both the XBD and the file system monitor components
are HRFS, dosFs, rawFs, and cdromFs.
The process by which devices are detected, and file systems created, is as follows:
1.
When file systems are initialized (at boot time), they register probe routines
and instantiation routines with the file system monitor.
2.
When a device is detected or inserted (for example, when a driver is initialized,
or media is inserted into an existing device—such as a floppy disk into a
floppy drive) the block device associated with it generates a primary insertion
event. (See Device Insertion Events, p.456.)
3.
In response to the primary insertion event, the file system monitor creates an
XBD partition manager if the device can support partitions. (For information
about the partition manager, see XBD Disk Partition Manager, p.403.)
4.
If the partition manager finds partitions on the physical device, it creates a
device for each partition; and whether or not partitions are found, the manager
generates a secondary insertion event.
5.
When the file system monitor receives a secondary event, all the registered file
system probe functions are run.
455
VxWorks
Kernel Programmer's Guide, 6.6
6.
When a file system's probe routine returns success, that file system's
instantiation routine is executed. If none of the probes are successful, or if the
file system instantiation routine fails, a rawFs file system is created on the
device by default.
When a device is removed, the following occurs:
1.
The block device detects the removal of the hardware device associated with
it and generates a removal event.
2.
The block device removes itself, freeing all its resources.
3.
The file system associated with the block device removes itself from core I/O,
invalidating its file handles.
4.
The file system removes itself, freeing all its resources.
Device Insertion Events
The types of device insertion events to which the file system monitor responds are
described in more detail below.
XBD Primary Insertion Event
An XBD-compliant block device generates a primary insertion event when
media that can support partitions is inserted (that is, if a partition table is
found). In response, the file system monitor creates a partition manager, which
in turn generates secondary insertion events for each partition that it finds on
the media (see below).
Note that for block devices used with the XBD wrapper component
(INCLUDE_XBD_BLK_DEV), a primary insertion event is always generated,
regardless of the media. The wrapper element is essentially hardware
agnostic; it cannot know if the device might include partitions. For example,
the device could be a hard disk—for which partitions are expected—or it could
be a floppy device.
Note also that a RAM disk device can generate a primary insertion event,
depending on the parameters used when it was created (see XBD RAM Disk,
p.411 and the API reference for XbdRamDisk).
XBD Secondary Insertion Event
A secondary insertion event is generated by either by a block device whose
media does not support partitions, or by an XBD partition manager. The
secondary event signals the file system manager to run the probe routines that
identify the file system on the device. If a probe routine returns OK, the
456
8 Local File Systems
8.3 Virtual Root File System: VRFS
associated file system creation routine is executed. If none of the probe
routines identifies a file system, or if a file system creation routine fails, the
rawFs file system is created by default.
XBD Soft Insert Event
Unlike the other events, an XBD soft insert event is produced by application
directive rather than by physical media being swapped. When ioctl( ) is called
with the XBD_SOFT_EJECT control function it tells the file system manager that
the current file system has been removed, and that a rawFs file system should
be created. This call causes the system to bypass the usual file system detection
operations, and ensures that rawFs is instantiated in place of the current file
system.
XBD Name Mapping Facility
The file system monitor name mapping facility allows XBD names to be mapped
to a more suitable name. It's primary use is for the partition manager which
appends :x to the base xbd name when it detects a partition. By using the fsm name
facility you can map the partition names to something more useful. For example,
the floppy drive configlette uses the name component to map the supplied floppy
name plus the :0 the partition manager will add to /fdx. Where x represents the
floppy drive number. If this was not done one would see the default device names
in the list generated by the devs shell command. For more information see the API
references for fsmNameInstall( ), fsmNameMap( ), and fsmNameUninstall( );
also see Example 8-2.
8.3 Virtual Root File System: VRFS
VxWorks provides a virtual root file system (VRFS) for use with applications that
require a POSIX root file system. The VRFS is simply a “/” or root directory from
which other file systems and devices can be accessed. VRFS is not a true file
system, as files and directories cannot be created with the sorts of commands that
are ordinarily associated with file systems, and it is read-only.
Only devices whose names begin with a single leading forward slash—and which
do not contain any other forward slash characters—are recognized by the VRFS.
457
8
VxWorks
Kernel Programmer's Guide, 6.6
To include the VRFS in VxWorks, configure the kernel with the INCLUDE_VRFS
component. The VRFS is created and mounted automatically if the component is
included in VxWorks.
This shell session illustrates the relationship between device names and access to
devices and file systems with the VRFS.
-> devs
drv name
0 /null
1 /tyCo/0
1 /tyCo/1
2 /aioPipe/0x1817040
6 /romfs
7 /
9 yow-build02-lx:
10 /vio
11 /shm
12 /ram0
value = 25 = 0x19
-> cd "/"
value = 0 =
-> ll
?--------drwxrwxr-x
?--------drwxrwxrwx
drwxrwxrwx
value = 0 =
0x0
0 0
0 15179
0 0
1 0
1 0
0x0
0
100
0
0
0
0
20
0
0
2048
Jan 1 00:00 null
Jan 23 2098 romfs/
Jan 1 00:00 vio
Jan 1 00:00 shm/
Jan 1 00:00 ram0/
Note that /tyCo/0, /tyCo/1, /aioPipe/0x1817040 and yow-build02-lx do not show
up in the directory listing of the root directory as they do not follow the naming
convention required by the VRFS. The first three include forward slashes in the
body of the device name and the fourth does not have a leading forward slash in
its name.
Also note that the listings of file systems have a trailing forward slash character.
Other devices do not, and they have a question mark in the permissions (or
attributes) column of the listing because they do not have recognizable file
permissions.
NOTE: Configuring VxWorks with support for POSIX PSE52 conformance (using
BUNDLE_RTP_POSIX_PSE52) provides the /dev/null device. Note that the devs
shell command lists /dev/null with other devices, but the ls command does not list
/dev/null under the VRFS root directory (because the name violates the VRFS
naming scheme). Applications can, in any case, use /dev/null as required. For
information about null devices, see 7.8.6 Null Devices, p.402. For information about
POSIX PSE52, see the VxWorks Application Programmer’s Guide: POSIX Facilities.
458
8 Local File Systems
8.4 Highly Reliable File System: HRFS
!
CAUTION: VRFS alters the behavior of other file systems because it provides a root
directory on VxWorks. Changing directory to an absolute path on a host file
system will not work when VRFS is installed without preceding the absolute path
with the VxWorks device name. For example, if the current working directory is
hostname, changing directory to /home/panloki will not work— it must be named
hostname:/home/panloki.
8
8.4 Highly Reliable File System: HRFS
The Highly Reliable File System (HRFS) is a transactional file system for real-time
systems. The primary features of the file system are:
■
Fault tolerance. The file system is never in an inconsistent state, and is
therefore able to recover quickly from unexpected loses of power.
■
Configurable commit policies.
■
Hierarchical file and directory system, allowing for efficient organization of
files on a volume.
■
Compatibility with a widely available storage devices.
■
POSIX compliance.
For more information about the HRFS libraries see the VxWorks API references for
hrfsFormatLib, hrFsLib, and hrfsChkDskLib.
HRFS and Flash Memory
For information about using HRFS with flash memory, see 10. Flash File System
Support: TrueFFS.
8.4.1 Configuring VxWorks for HRFS
To include HRFS support in VxWorks, configure the kernel with the appropriate
required and optional components.
459
VxWorks
Kernel Programmer's Guide, 6.6
Required Components
The components required for HRFS are the following:
■
Either the INCLUDE_HRFS or the INCLUDE_HRFS_READONLY component is
required. As its name indicates, the latter is a read-only version of the main
HRFS component. The libraries it provides are smaller as it provides no
facilities for disk modifications.
■
The appropriate component for your block device; for example,
INCLUDE_ATA.
■
If you are using a device driver that is not designed for use with the XBD
facility, you must use the INCLUDE_XBD_BLK_DEV wrapper component in
addition to INCLUDE_XBD.
See XBD Block Device Wrapper, p.404 for more information.
Optional HRFS Components
The INCLUDE_HRFS_FORMAT (formatter), INCLUDE_HRFS_CHKDSK
(consistency checker), and INCLUDE_HRFS_ACCESS_TIMESTAMP (access file
stamp) components are optional. Note that the
INCLUDE_HRFS_ACCESS_TIMESTAMP component is included in the
BUNDLE_RTP_POSIX_PSE52 component bundle.
Optional XBD Components
The INCLUDE_XBD_PART_LIB (disk partitioning) and INCLUDE_XBD_RAMDRV
(RAM disk) components are optional.
For information about the XBD facility, see 7.8.8 Extended Block Device Facility:
XBD, p.402).
8.4.2 Configuring HRFS
HRFS provides the following component configuration parameters:
HRFS_DEFAULT_MAX_BUFFERS
Defines how many buffers HRFS uses for its caching mechanism. HRFS needs
a minimum of 6 buffers. The default setting is 16. This parameter applies to all
HRFS volumes. Note that while increasing the number of buffers can increase
performance, it does so at the expense of heap memory.
460
8 Local File Systems
8.4 Highly Reliable File System: HRFS
HRFS_DEFAULT_MAX_FILES
Defines how many files can be simultaneously open on an HRFS volume. The
minimum is 1. The default setting is 10. Note that is not the same as the
maximum number of file descriptors.
HRFS_DEFAULT_COMMIT_POLICY
Defines the default commit policy for an HRFS volume, which is
FS_COMMIT_AUTO. Commit policies can also be changed at runtime. For
more information see 8.4.5 Transactional Operations and Commit Policies, p.469
and 8.4.6 Configuring Transaction Points at Runtime, p.471.
HRFS_DEFAULT_COMMIT_PERIOD
Defines the initial commit period of an HRFS volume if it has been configured
for periodic commits. This parameter is measured in milliseconds. The default
value is 5000 milliseconds (5 seconds). The commit period can also be changed
at runtime. For more information see 8.4.5 Transactional Operations and Commit
Policies, p.469 and 8.4.6 Configuring Transaction Points at Runtime, p.471.
8.4.3 Creating an HRFS File System
This section describes the process of creating an HRFS file system. It first provides
a summary overview and then a detailed description of each step. See 8.4.4 HRFS,
ATA, and RAM Disk Examples, p.463 for examples of the steps and code examples.
Overview of HRFS File System Creation
For information operating system configuration, see 8.4.1 Configuring VxWorks for
HRFS, p.459. Note that the file system is initialized automatically at boot time.
The steps involved in creating an HRFS file system are as follows:
1.
If you are using a custom driver, create the appropriate block device. See
Step 1:Create a Block Device, p.462.
If you are using a standard VxWorks component for the device, it is created
automatically.
2.
If you are using a device driver that is not XBD-compliant, create an XBD
device wrapper. See Step 2:Create an XBD Device Wrapper, p.462 (Also see XBD
Block Device Wrapper, p.404.).
3.
Optionally, create and mount partitions. See Step 3:Create Partitions, p.463.
461
8
VxWorks
Kernel Programmer's Guide, 6.6
4.
If you are not using pre-formatted disks, format the volumes. See
Step 4:Formatting the Volume, p.463.
HRFS File System Creation Steps
Before any other operations can be performed, the HRFS file system library,
hrFsLib, must be initialized. This happens automatically at boot time, triggered by
the required HRFS components that were included in the system.
Initializing HRFS involves the creation of a vnode layer. HRFS installs an number
of internal vnode operators into the this layer. The vnode layer invokes
iosDrvInstall( ) when media is detected, which adds the driver to the I/O driver
table. The driver number assigned to vnodes—and therefore HRFS—is recorded
in a global variable, vnodeAffDriverNumber. The table specifies entry points for
the vnode file operations that are accessing devices using HRFS.
Step 1:
Create a Block Device
If you are using a standard VxWorks component for the device, it is created
automatically.
If you are using a custom driver, create the appropriate block device by calling the
creation routine for the device driver. The format for this routine is
xxxDevCreate( ) where xxx represents the device driver type; for example,
ataDevCreate( ).
The driver routine returns a pointer to a block device descriptor structure,
BLK_DEV. This structure describes the physical attributes of the device and
specifies the routines that the device driver provides. The pointer returned is used
to create an XBD device wrapper in the next step.
Step 2:
Create an XBD Device Wrapper
If you are using a device driver that is not XBD-compliant, it requires an XBD
device wrapper.
The wrapper is created automatically if you have configured VxWorks with the
INCLUDE_XBD_BLK_DEV wrapper component (See XBD Block Device Wrapper,
p.404). Otherwise, create a wrapper for each block device using
xbdBlkDevCreate( ).
After the XBD device wrapper is created, the physical device is automatically
probed for a file system and partitions. If a disk is already formatted, the disk is
462
8 Local File Systems
8.4 Highly Reliable File System: HRFS
mounted. If a file system is found, it is mounted as well. If file system is not HRFS,
it must be formatted (see below).
Step 3:
Create Partitions
If you have included the INCLUDE_XBD_PART_LIB component in your system,
you can create partitions on a disk and mount volumes atop the partitions. Use the
xbdCreatePartition( ) routine to create partitions.
This step should only be performed once, when the disk is first initialized. If
partitions are already written to the disk, this step should not be performed as it
destroys data.
8
Step 4:
Formatting the Volume
If you are using unformatted disk or wish to replace the current file system on the
disk, format the disk by calling hrFsFormat( ). For more information, see the
VxWorks API reference for this routine.
!
CAUTION: Reformatting a disk destroys any data that may be on it.
8.4.4 HRFS, ATA, and RAM Disk Examples
This section provides examples of the steps discussed in the preceding section.
They are meant to be relatively generic, and illustrate the following:
!
■
Creating and working with an HRFS file system on an ATA disk with
commands from the shell.
■
Code that creates and formats partitions.
■
Code that creates and formats a RAM disk volume.
CAUTION: Because device names are recognized by the I/O system using simple
substring matching, file systems should not use a slash (/) alone as a name;
unexpected results may otherwise occur.
Example 8-1
Create HRFS in Partitions on an ATA Disk
This example demonstrates how to initialize an ATA disk with HRFS on two
partitions from the shell. While these steps use an ATA device, they are applicable
to other block devices.
463
VxWorks
Kernel Programmer's Guide, 6.6
1.
If you are using a custom driver, create an ATA block device that controls the
master ATA hard disk (drive zero) on the primary ATA controller (controller
zero). This device uses the entire disk.
-> xbd = ataXbdDevCreate(0,0,0,0,"/ata")
New symbol "xbd" added to kernel symbol table.
Instantiating /ata:0 as rawFs
xbd = 0xca4fe0: value = 262145 = 0x40001
The xbd variable is of type device_t. A value of zero would indicate an error
in the ataXbdDevCreate( ) call, which usually indicates a BSP configuration or
hardware configuration error.
If you are using the standard INCLUDE_ATA device component, the block
device is created automatically. Note that in this case the default device name
(provided by the component) is /ata0a.
2.
Display information about devices.
-> devs
drv name
0 /null
1 /tyCo/0
1 /tyCo/1
8 yow-grand:
9 /vio
4 /ata:0
value = 25 = 0x19
3.
The new ata driver /ata:0 is listed. The zero in the name indicates that no
partitions were detected. Note that if no file system is detected on the device,
the rawFs file system is instantiated automatically and appears in the device
list. Prepare the disk for first use. Create two partitions on this disk device,
specifying 50% of the disk space for the second partition, leaving 50% for the
first partition. This step should only be performed once, when the disk is first
initialized. If partitions are already written to the disk, this step should not be
performed since it destroys any data on the disk.
-> xbdCreatePartition ("/ata:0", 2, 50, 0, 0)
value = 0 = 0x0
The four arguments are to xbdCreatePartition( ) are:
■
■
■
■
■
the drive name
the number of partitions
the percent of disk to use for the second partition
the percent of disk to use for the third partition
the percent of disk to use for the fourth partition
The remainder of the disk is used for the first partition.
464
8 Local File Systems
8.4 Highly Reliable File System: HRFS
4.
Then list the devices to display information about the new partitions.
-> devs
drv name
0 /null
1 /tyCo/0
1 /tyCo/1
8 yow-grand:
9 /vio
3 /ata:1
3 /ata:2
Note that /ata:0 does not appear in this list, and two new devices, /ata:1 and
/ata:2, have been added to represent the new partitions. Each volume has
rawfs instantiated in it as they are new and unformatted.
5.
Format the volumes for HRFS. This step need only be done once, when the
volumes are first created. If the volumes have already been formatted, then
omit this step. This example formats the file system volumes with default
options.
-> hrfsFormat ("/ata:1", 0ll, 0, 0)
Formatting /ata:1 for HRFS
Instantiating /ata:1 as rawFs
Formatting...OK.
value = 0 = 0x0
-> hrfsFormat ("/ata:2", 0ll, 0, 0)
Formatting /ata:2 for HRFS
Instantiating /ata:2 as rawFs
Formatting...OK.
value = 0 = 0x0
Note that in the hrfsFormat( ) call, the ll (two lower-case L letters) used with
the second parameter is required to indicate to the shell that the data type is
long long.
For more information, see the API reference for hrFsFormatLib.
6.
Display information about the HRFS volumes.
-> ll "/ata:1"
Listing Directory /ata:1:
drwxrwxrwx 1 0
0
drwxrwxrwx 1 0
0
value = 0 = 0x0
8192 Jan
8192 Jan
1 00:13 ./
1 00:13 ../
8192 Jan
8192 Jan
1 00:13 ./
1 00:13 ../
-> ll "/ata:2"
Listing Directory /ata:2:
drwxrwxrwx 1 0
0
drwxrwxrwx 1 0
0
value = 0 = 0x0
465
8
VxWorks
Kernel Programmer's Guide, 6.6
If you are working with an ATA hard disk or a CD-ROM file system from an
ATAPI CD-ROM drive, you can, alternatively, use usrAtaConfig( ). This
routine processes several steps at once. For more information, see the API
reference.
Example 8-2
Creating and Partitioning a Disk and Creating Volumes
This code takes the name of a block device that you have already instantiated,
creates three partitions, creates the partition handler for these partitions, and
creates the HRFS device handler for them. Then it formats the partitions using
hrfsFormat( ).
STATUS usrPartDiskFsInit
(
char * xbdName /* device name used during creation of XBD */
)
{
const char * devNames[] = { "/sd0a", "/sd0b", "/sd0c" };
devname_t xbdPartName;
int i;
/* Map partition names */
for (i = 1; i <= 3; i++)
{
sprintf (xbdPartName, "%s:d", devNames[i],i);
fsmNameInstall (devNames[i-1], xbdPartName);
}
/* create partitions */
if((xbdCreatePartition (xbdName,3,50,45)) == ERROR)
return ERROR;
/* Formatting the first partition */
if(hrfsFormat (devNames[0], 0ll,0, 0) == ERROR)
return ERROR;
/* Formatting the second partition */
if(hrfsFormat (devNames[1], 0ll, 0, 0) == ERROR)
return ERROR;
/* Formatting the third partition */
if(hrfsFormat (devNames[2], 0ll, 0, 0) == ERROR)
return ERROR;
return OK;
}
466
8 Local File Systems
8.4 Highly Reliable File System: HRFS
Note that in most cases you would be likely to format the different partitions for
different file systems.
Example 8-3
Creating and Formatting a RAM Disk Volume and Performing File I/O
The following code creates a RAM disk, formats it for use with the HRFS file
system, and performs file system operations.
#include
#include
#include
#include
#include
#include
#include
#define
#define
#define
DEVNAME "/myram" /* name of the RAM disk */
BLOCKSIZE 512
DISKSIZE (BLOCKSIZE * 2000)
8
STATUS hrfsSetup
(
void
)
{
STATUS error;
device_t xbd;
/* Create a RAM disk. Don’t support partitions */
xbd = xbdRamDiskDevCreate (BLOCKSIZE, DISKSIZE, 0, DEVNAME);
if (xbd == NULLDEV)
{
printf("Failed to create RAM disk. errno = 0x%x\n", errno);
return (ERROR);
}
/*
* Format the RAM disk for HRFS. Allow for upto a 1000 files/directories
* and let HRFS determine the logical block size.
*/
error = hrfsFormat (DEVNAME, 0ll, 0, 1000);
if (error != OK)
{
printf("Failed to format RAM disk. errno = 0x%x\n", errno);
return (ERROR);
}
printf ("%s now ready for use.\n", DEVNAME);
return (OK);
}
467
VxWorks
Kernel Programmer's Guide, 6.6
STATUS hrfsFileExample
(
void
)
{
int fd;
char path[PATH_MAX];
char *testString = "hello world";
int size = strlen (testString) + 1; /* size of test string including EOS */
int len;
/* Create a file under the root directory */
/* First build the path */
sprintf (path, "%s/myfile", DEVNAME);
fd = open (path, O_RDWR | O_CREAT, 0777);
if (fd < 0)
{
printf ("Couldn’t create file %s. errno = 0x%x\n", path, errno);
return (ERROR);
}
/* Write to the file */
printf("Writing %d bytes to file.\n", size);
len = write (fd, testString, size);
if (len != size)
{
printf ("Couldn’t write to file %s. errno = 0x%x\n", path, errno);
close (fd);
return (ERROR);
}
/* Close and re-open file */
close (fd);
fd = open (path, O_RDWR, 0777);
if (fd < 0)
{
printf ("Couldn’t re-open file %s. errno = 0x%x\n", path, errno);
return (ERROR);
}
/* Now read back what we wrote */
printf("Reading %d bytes from file.\n", size);
len = read (fd, path, size);
if (len != 12)
{
printf ("Couldn’t read from file %s. errno = 0x%x\n", path, errno);
close (fd);
return (ERROR);
}
468
8 Local File Systems
8.4 Highly Reliable File System: HRFS
/* Make sure we read what we wrote */
if ((len = strcmp (path, testString)) != 0)
{
printf ("Read data different from written data. errno = 0x%x, %d\n", errno, len);
close (fd);
return (ERROR);
}
close (fd);
return (OK);
}
Note that to use this code, you must configure VxWorks with the
INCLUDE_HRFS_FORMAT, INCLUDE_XBD_RAMDRV and
INCLUDE_XBD_PART_LIB components.
8
The following illustrates running the example from the shell.
-> hrfsSetup
Instantiating /myram as rawFs
Formatting /myram for HRFS
Instantiating /myram as rawFs
Formatting...OK.
/myram now ready for use.
value = 0 = 0x0
-> hrfsFileExample
Writing 12 bytes to file.
Reading 12 bytes from file.
value = 0 = 0x0
-> ll "/myram"
Listing Directory /myram:
drwxrwxrwx 1 0
0
drwxrwxrwx 1 0
0
-rwxrwxrwx 1 0
0
value = 0 = 0x0
->
2048 Jan
2048 Jan
12 Jan
1 00:00 ./
1 00:00 ../
1 00:00 myfile
8.4.5 Transactional Operations and Commit Policies
HRFS is a transactional file system. That is, transaction or commit points are set to
make disk changes permanent. Commit points can be configured to be set under
different conditions, which are referred to as policies. Some disk operations trigger
commits regardless of the policy. Under certain circumstances, HRFS rollbacks
undo disk changes since the last commit, in order to protect the integrity of the file
system.
For information about static and dynamic configuration of commit policies, see
8.4.2 Configuring HRFS, p.460 and 8.4.6 Configuring Transaction Points at Runtime,
p.471.
469
VxWorks
Kernel Programmer's Guide, 6.6
Commit Policies
The HRFS commit policies are as follows:
Automatic
Any operation that changes data on the disk results in a transaction point
being set. This is the safest policy in terms of the potential for data loss. It is
also the slowest in terms of performance, as every write to disk cause a
commit. This is the default policy. There is no need for explicit action on the
part of an application to commit a change. The following routines, for
example, cause modifications to disk and result in a commit when the
automatic policy is in force:
■
■
■
■
■
■
■
■
■
■
write( )
remove( )
delete( )
mkdir( )
rmdir( )
link( )
unlink( )
truncate( )
ftruncate( )
ioctl( ) when used with a control function that requires modifying the
disk.
Manual
The application decides when a commit is to be performed. The user explicitly
sets transaction points. This is the fastest policy in terms of performance but
obviously has the potential for greater data loss. The application can, however,
decide when critical data has been written and needs to be committed. The
commit( ) routine is used with this policy.
Periodic
Transaction points are set automatically at periodic intervals. This policy is in
between automatic and manual in terms of performance and potential data
loss.
470
8 Local File Systems
8.4 Highly Reliable File System: HRFS
Mandatory Commits
For both manual and periodic commit policies there are circumstances under
which a commit is always performed. Mandatory commits occur under the
following circumstances:
■
Creation of a file or directory
■
Deletion of a file or directory.
■
Renaming/moving a file or directory.
■
Space in the inode journal is exhausted.
■
Commit policy is changed at runtime.
8
Note that mandatory commits are a subset of automatic commits—they do not, for
example, include write( ) and truncate( ).
Rollback
A rollback undoes any disk changes since the last commit. Rollbacks usually occur
when the system is unexpectedly powered down or reset. Rollbacks can also occur
when the file system encounters errors; for example, the lack of disk space to
complete a write( ), or an error is reported by the underlying device driver.
Rollbacks of this nature only happen on operations that modify the media. Errors
on read operations do not force a rollback.
A rollback involves HRFS returning to the state of the disk at the last transaction
point, which thereby preserves the integrity of the file system, but at the expense
of losing file data that has changed since the last transaction point. If the manual
or periodic commit policy is specified, there is the potential for losing a lot of
data—although the integrity of the file system is preserved.
8.4.6 Configuring Transaction Points at Runtime
The Highly Reliable File System (HRFS) provides configurable transaction points,
which allow for finer control of how and when transaction points are set.
The HRFS_DEFAULT_COMMIT_POLICY and HRFS_DEFAULT_COMMIT_PERIOD
component configuration parameters are used to statically define the default
commit policy and period (for more information see 8.4.2 Configuring HRFS,
p.460).
471
VxWorks
Kernel Programmer's Guide, 6.6
Both kernel and RTP applications can change commit policies at runtime. The
following ioctl( ) functions are used to get and set commit policies:
■
■
■
■
FIOCOMMITPOLICYGETFS
FIOCOMMITPOLICYSETFS
FIOCOMMITPERIODGETFS
FIOCOMMITPERIODSETFS
The commit policy for each volume can be changed using the ioctl( ) function
FIOCOMMITPOLICYSETFS as the second parameter.
The third parameter then specifies the actual commit policy:
FS_COMMIT_POLICY_AUTO, FS_COMMIT_POLICY_MANUAL, or
FS_COMMIT_POLICY_PERIODIC.
If an HRFS volume has been configured for periodic commits, the commit period
can be changed with ioctl( ) function FIOCOMMITPERIODSETFS. The third
parameter is used to specify the commit period in milliseconds. If 0 is specified
then the default commit period is used.
The commit( ) routine can be used to commit programmatically. The routine is
provided by the INCLUDE_DISK_UTILS component.
Example 8-4
Setting an HRFS Commit Policy at Runtime
/* open root directory of an HRFS volume */
fd = open ("/hrfs", O_READONLY, 0666);
if (fd < 0)
return ERROR;
/* Set commit policy to manual */
if (ioctl (fd, FIOCOMMITPOLICYSETFS, (void *)FS_COMMIT_POLICY_MANUAL)) ==
ERROR)
return ERROR;
/* Policy set to manual. Change policy to periodic */
if (ioctl (fd, FIOCOMMITPOLICYSETFS, (void *)FS_COMMIT_POLICY_PERIODIC)) ==
ERROR)
return ERROR;
/* Policy set to periodic. Change commit period to 10 seconds*/
if (ioctl (fd, FIOCOMMITPERIODSETFS, (void *)10000)) == ERROR)
return ERROR;
/* Period set to 10 seconds. Change commit period back to default */
if (ioctl (fd, FIOCOMMITPERIODSETFS, (void *)0)) == ERROR)
return ERROR;
/* Period reset. Change policy back to automatic */
if (ioctl (fd, FIOCOMMITPOLICYSETFS, (void *)FS_COMMIT_POLICY_AUTO)) ==
ERROR)
return ERROR;
472
8 Local File Systems
8.4 Highly Reliable File System: HRFS
8.4.7 File Access Time Stamps
Access time stamps can be enabled by configuring VxWorks with the
INCLUDE_HRFS_ACCESS_TIMESTAMP component. The component is include in
the BUNDLE_RTP_POSIX_PSE52 component bundle.
For access time stamps to be saved to disk, the volume must be formatted with
HRFS on-disk format 1.2 or greater. Version 1.2 is the default version for VxWorks
6.3. See API references for hrfsAdvFormat( ) and hrfsAdvFormatFd( ) for more
information.
When the access timestamp component is included, and the appropriate disk
format version is used, reading from a file or directory causes its access time stamp
to be updated. This can cause significant performance loss, as a write to disk occurs
even on a read operation and a transaction point is set. Only use access time stamps
if the application requires it for POSIX compliance.
8.4.8 Maximum Number of Files and Directories
HRFS files and directories are stored on disk in data structures called inodes.
During formatting the maximum number of inodes is specified as a parameter to
hrfsFormat( ). The total number of files and directories can never exceed the
number inodes. Attempting to create a file or directory when all inodes are in use
generates an error. Deleting a file or directory frees the corresponding inode.
8.4.9 Working with Directories
This section discusses creating and removing directories, and reading directory
entries.
Creating Subdirectories
You can create as many subdirectories as there are inodes. Subdirectories can be
created in the following ways:
■
With open( ). To create a directory, the O_CREAT option must be set in the
flags parameter and the S_IFDIR or FSTAT_DIR option must be set in the mode
parameter. The open( ) calls returns a file descriptor that describes the new
directory. The file descriptor can only be used for reading only and should be
closed when it no longer needed.
473
8
VxWorks
Kernel Programmer's Guide, 6.6
■
With mkdir( ) from usrFsLib.
When creating a directory using either of the above methods, the new directory
name must be specified. This name can be either a full pathname or a pathname
relative to the current working directory.
Removing Subdirectories
A directory that is to be deleted must be empty (except for the “.” and “..” entries).
The root directory can never be deleted. Subdirectories can be removed in the
following ways:
■
Using ioctl( ) with the FIORMDIR function and specifying the name of the
directory. The file descriptor used can refer to any file or directory on the
volume, or to the entire volume itself.
■
Using the remove( ), specifying the name of the directory.
■
Use rmdir( ) from usrFsLib.
Reading Directory Entries
You can programmatically search directories on HRFS volumes using the
opendir( ), readdir( ), rewinddir( ), and closedir( ) routines.
To obtain more detailed information about a specific file, use the fstat( ) or stat( )
routine. Along with standard file information, the structure used by these routines
also provides the file-attribute byte from a directory entry.
For more information, see the API reference for dirLib.
8.4.10 Working with Files
This section discusses file I/O and file attributes.
File I/O Routines
Files on an HRFS file system device are created, deleted, written, and read using
the standard VxWorks I/O routines: creat( ), remove( ), write( ), and read( ). For
more information, see 7.4 Basic I/O, p.365, and the ioLib API references.
474
8 Local File Systems
8.4 Highly Reliable File System: HRFS
Note that and remove( ) is synonymous with unlink( ) for HRFS.
File Linking and Unlinking
When a link is created an inode is not used. Another directory entry is created at
the location specified by the parameter to link( ). In addition, a reference count to
the linked file is stored in the file's corresponding inode. When unlinking a file, this
reference count is decremented. If the reference count is zero when unlink( ) is
called, the file is deleted except if there are open file descriptors open on the file. In
this case the directory entry is removed but the file still exists on the disk. This
prevents tasks and processes (RTPs) from opening the file. When the final open file
descriptor is closed the file is fully deleted freeing its inode.
Note that you cannot create a link to a subdirectory only to a regular file.
File Permissions
HRFS files have POSIX-style permission bits (unlike dosFs files, which have
attributes). The bits can be changed using the chmod( ) and fchmod( ) routines. See
the API references for more information.
8.4.11 Crash Recovery and Volume Consistency
HRFS is a transactional based file system that is designed to be consistent at all
times.
Crash Recovery
If a system unexpectedly loses power or crashes, HRFS rolls back to the last
transaction point when the system reboots. The rollback occurs automatically
when the file system is mounted. Any changes made after the last complete
transaction are lost, but the disk remains in a consistent state.
475
8
VxWorks
Kernel Programmer's Guide, 6.6
Consistency Checking
An HRFS file system remains in a consistent state for most media (such as hard
drives) as long as the underlying hardware is working correctly and never writes
an incomplete sector or physical block.
This is necessarily true for RAM disks, however, because sector writing is simply
a copy of one memory location to another. The write operation may be interrupted
before completion if the system loses power or crashes.
The hrfsChkDsk( ) routine can, however, be used to check for inconsistencies in
the file system. The execution of the disk checker is not automatic; it must be done
programmatically. The hrfsChkDsk( ) routine is also useful in relation to
hardware problems, and is usually run at boot time for this purpose.
8.4.12 I/O Control Functions Supported by HRFS
The HRFS file system supports the ioctl( ) functions. These functions are defined
in the header file ioLib.h along with their associated constants; and they are listed
in Table 8-1.
Table 8-1
I/O Control Functions Supported by HRFS
Function
Decimal
Value Description
FIODISKCHANGE
13
Announces a media change.
FIODISKFORMAT
5
Formats the disk (device driver function).
FIODISKINIT
6
Initializes a file system on a disk volume.
FIOFLUSH
2
Flushes the file output buffer.
FIOFSTATGET
38
Gets file status information (directory entry data).
FIOGETNAME
18
Gets the filename of the fd.
FIOMOVE
47
Moves a file (does not rename the file).
FIONFREE
30
Gets the number of free bytes on the volume.
FIONREAD
1
Gets the number of unread bytes in a file.
FIOREADDIR
37
Reads the next directory entry.
476
8 Local File Systems
8.4 Highly Reliable File System: HRFS
Table 8-1
I/O Control Functions Supported by HRFS (cont’d)
Function
Decimal
Value Description
FIORENAME
10
Renames a file or directory.
FIORMDIR
32
Removes a directory.
FIOSEEK
7
Sets the current byte offset in a file.
FIOSYNC
21
Same as FIOFLUSH, but also re-reads buffered file
data.
FIOTRUNC
42
Truncates a file to a specified length.
FIOUNMOUNT
39
Un-mounts a disk volume.
FIOWHERE
8
Returns the current byte position in a file.
FIONCONTIG64
50
Gets the maximum contiguous disk space into a
64-bit integer.
FIONFREE64
51
Gets the number of free bytes into a 64-bit integer.
FIONREAD64
52
Gets the number of unread bytes in a file into a 64-bit
integer.
FIOSEEK64
53
Sets the current byte offset in a file from a 64-bit
integer.
FIOWHERE64
54
Gets the current byte position in a file into a 64-bit
integer.
FIOTRUNC64
55
Set the file's size from a 64-bit integer.
8
For more information, see the API reference for ioctl( ) in ioLib.
477
VxWorks
Kernel Programmer's Guide, 6.6
8.5 MS-DOS-Compatible File System: dosFs
The dosFs file system is an MS-DOS-compatible file system that offers
considerable flexibility appropriate to the multiple demands of real-time
applications. The primary features are:
■
Hierarchical files and directories, allowing efficient organization and an
arbitrary number of files to be created on a volume.
■
A choice of contiguous or non-contiguous files on a per-file basis.
■
Compatible with widely available storage and retrieval media (diskettes, hard
drives, and so on).
■
The ability to boot VxWorks from a dosFs file system.
■
Support for VFAT (Microsoft VFAT long file names)
■
Support for FAT12, FAT16, and FAT32 file allocation table types.
For information about dosFs libraries, see the VxWorks API references for
dosFsLib and dosFsFmtLib.
For information about the MS-DOS file system, please see the Microsoft
documentation.
dosFs and Flash Memory
For information about using dosFs with flash memory, see 10. Flash File System
Support: TrueFFS.
dosFs and the Transaction-Based Reliable File System Facility
The dosFs file system can be used with the transaction-based reliable file system
(TRFS) facility; see 7.8.9 Transaction-Based Reliable File System Facility: TRFS, p.405.
8.5.1 Configuring VxWorks for dosFs
To include dosFs support in VxWorks, configure the kernel with the appropriate
required and optional components.
478
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
Required Components
The following components are required:
INCLUDE_DOSFS_MAIN
INCLUDE_DOSFS_FAT
INCLUDE_XBD
dosFsLib
dosFs FAT12/16/32 FAT handler
XBD component
And, either one or both of the following components are required:
INCLUDE_DOSFS_DIR_VFAT Microsoft VFAT direct handler
INCLUDE_DOSFS_DIR_FIXED Strict 8.3 & VxLongNames directory handler
In addition, you must include the appropriate component for your block device;
for example, INCLUDE_ATA.
If you are using a device driver that is not designed for use with the XBD facility,
you must use the INCLUDE_XBD_BLK_DEV wrapper component in addition to
INCLUDE_XBD. See XBD Block Device Wrapper, p.404 for more information.
Note that you can use INCLUDE_DOSFS to automatically include the following
components:
■
■
■
■
■
■
INCLUDE_DOSFS_MAIN
INCLUDE_DOSFS_DIR_VFAT
INCLUDE_DOSFS_DIR_FIXED
INCLUDE_DOSFS_FAT
INCLUDE_DOSFS_CHKDSK
INCLUDE_DOSFS_FMT
Optional dosFs Components
The optional dosFs components are:
INCLUDE_DOSFS_CACHE
INCLUDE_DOSFS_FMT
INCLUDE_DOSFS_CHKDSK
INCLUDE_DISK_UTIL
INCLUDE_TAR
disk cache facility
dosFs file system formatting module
file system integrity checking
standard file system operations, such as ls, cd,
mkdir, xcopy, and so on
the tar utility
479
8
VxWorks
Kernel Programmer's Guide, 6.6
Optional XBD Components
Optional XBD components are:
INCLUDE_XBD_PART_LIB
INCLUDE_XBD_TRANS
INCLUDE_XBD_RAMDRV
disk partitioning facilities
TRFS support facility
RAM disk facility
For information about the XBD facility, see 7.8.8 Extended Block Device Facility:
XBD, p.402).
8.5.2 Configuring dosFs
Several dosFs component configuration parameters can be used to define how the
file system behaves when a dosfs volume is mounted. These parameters are as
follows:
DOSFS_CHK_ONLY
When a dosfs volume is mounted, the media is analyzed for errors, but no
repairs are made.
DOSFS_CHK_REPAIR
Similar to DOSFS_CHK_ONLY, but an attempt to repair the media is made if
errors are found.
DOSFS_CHK_NONE
Media is not checked for errors on mount.
DOSFS_CHK_FORCE
Used in conjunction with DOSFS_CHK_ONLY and DOSFS_CHK_REPAIR to
force a consistency check even if the disk has been marked clean.
DOS_CHK_VERB_SILENT or DOS_CHK_VERB_0
dosFs does not to produce any output to the terminal when mounting.
DOS_CHK_VERB_1
dosFs produces a minimal amount of output to the terminal when mounting.
DOS_CHK_VERB_2
dosFs to produces maximum amount output to the terminal when mounting.
Other parameters can be used to configure physical attributes of the file system.
They are as follows:
480
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
DOSFS_DEFAULT_CREATE_OPTIONS
The default parameter for the dosFsLib component. It specifies the action to be
taken when a dosFs file system is instantiated. Its default is
DOSFS_CHK_NONE.
DOSFS_DEFAULT_MAX_FILES
The maximum number of files. The default is 20.
DOSFS_DEFAULT_DATA_CACHE_SIZE
The size of the data cache. The default is 128 KB.
DOSFS_DEFAULT_FAT_CACHE_SIZE
The size of the FAT cache. The default 16 KB.
8
DOSFS_DEFAULT_DIR_CACHE_SIZE
The directory cache size. The default is 64 KB.
Caches can be tuned dynamically for individual instances of the file system using
the dosFsCacheInfo( ) and dosFsCacheTune( ) routines.
The routines dosFsCacheDelete( ) and dosFsCacheCreate( ) can be used to delete
and changes the size of caches. To change the size, first delete, and then create.
481
VxWorks
Kernel Programmer's Guide, 6.6
8.5.3 Creating a dosFs File System
This section describes the process of creating a dosFs file system. It first provides
a summary overview and then a detailed description of each step. See 8.5.4 dosFs,
ATA Disk, and RAM Disk Examples, p.486 for examples of the steps and code
examples.
Overview of dosFs File System Creation
For information operating system configuration, see 8.5.1 Configuring VxWorks for
dosFs, p.478. Note that The file system is initialized automatically at boot time.
The steps involved in creating a dosFs file system are as follows:
1.
If you are using a custom driver, create the appropriate block device. See
Step 1:Create a Block Device, p.462.
If you are using a standard VxWorks component for the device, it is created
automatically.
2.
If you are using a device driver that is not XBD-compliant, create an XBD
device wrapper. See Step 2:Create an XBD Device Wrapper, p.483. (Also see XBD
Block Device Wrapper, p.404.)
3.
Optionally, create and mount partitions. See Step 3:Create Partitions, p.463.
4.
If you are not using pre-formatted disks, format the volumes. See
Step 4:Formatting the Volume, p.484.
5.
Optionally, change the size of the disk cache. See Step 5:Change the Disk Cache
Size, p.485.
6.
Optionally, check the disk for volume integrity. See Step 6:Check Disk Volume
Integrity, p.485.
dosFs File System Creation Steps
Before any other operations can be performed, the dosFs file system library,
dosFsLib, must be initialized. This happens automatically at boot time, triggered
by the required dosFs components that were included in the system.
Initializing the file system invokes iosDrvInstall( ), which adds the driver to the
I/O system driver table. The driver number assigned to the dosFs file system is
482
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
recorded in a global variable, dosFsDrvNum. The table specifies the entry points
for the dosFs file operations that are accessed by the devices using dosFs.
Step 1:
Create a Block Device
If you are using a standard VxWorks component for the device, it is created
automatically.
If you are using a custom driver, create the appropriate block device by calling the
creation routine for the device driver. The format for this routine is
xxxDevCreate( ) where xxx represents the device driver type; for example,
scsiBlkDevCreate( ) or ataDevCreate( ).
The driver routine returns a pointer to a block device descriptor structure,
BLK_DEV. This structure describes the physical attributes of the device and
specifies the routines that the device driver provides. The pointer returned is used
to create an XBD block device wrapper. For more information on block devices, see
TRFS Code Examples, p.408.
Step 2:
Create an XBD Device Wrapper
If you are using a device driver that is not XBD-compliant, it requires an XBD
device wrapper.
The wrapper is created automatically if you have configured VxWorks with the
INCLUDE_XBD_BLK_DEV wrapper component (See XBD Block Device Wrapper,
p.404). Otherwise, create a wrapper for each block device using
xbdBlkDevCreate( ).
After the XBD device wrapper is created the physical device is automatically
probed for a file system and partitions. If a disk is already formatted, the disk is
mounted. If a a file system is found, it is mounted. If the file system is not dosFs, it
must be formatted (see below).
Step 3:
Create Partitions
If you have included the INCLUDE_XBD_PART_LIB component in your system,
you can create partitions on a disk and mount volumes atop the partitions. Use the
xbdCreatePartition( ) routine to create partitions.
This step should only be performed once, when the disk is first initialized. If
partitions are already written to the disk, this step should not be performed as it
destroys data.
483
8
VxWorks
Kernel Programmer's Guide, 6.6
Step 4:
Formatting the Volume
If you are using unformatted disk or wish to replace the current file system on the
disk, format the disk by calling dosFsVolFormat( ). For more information, see the
VxWorks API reference for this routine.
The MS-DOS and dosFs file systems provide options for the format of the File
Allocation Table (FAT) and the format of the directory. These options, described
below, are completely independent.
!
CAUTION: Reformatting a disk destroys any data that may be on it.
File Allocation Table (FAT) Formats
A volume FAT format is set during disk formatting, according to either the volume
size (by default), or the per-user defined settings passed to dosFsVolFormat( ).
FAT options are summarized in Table 8-2:
Table 8-2
FAT Formats
Format
FAT Table Entry SIze
Usage
Size
FAT12
12 bits per cluster
number
Appropriate for very small devices Typically, each cluster is two
with up to 4,084 KB clusters.
sectors large.
FAT16
16 bits per cluster
number
Appropriate for small disks of up Typically, used for volumes up
to 65,524 KB clusters.
to 2 GB; can support up to 8 GB.
FAT32
32 bits (only 28 used) Appropriate for medium and
per cluster number larger disk drives.
By convention, used for
volumes larger than 2 GB.
Directory Formats
The options for the directory format are:
■
MSFT Long Names (VFAT)
Uses case-insensitive long filenames, with up to 254 characters. This format
accepts disks created with short names. MSFT Long Names1 is the default
directory format.
1. The MSFT Long Names (VFAT) format supports 32-bit file size fields, limiting the file size
to a 4 GB maximum.
484
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
■
Short Names (8.3)
Case-insensitive MS-DOS-style filenames (8.3), with eight uppercase
characters for the name itself and three for the extension.
Step 5:
Change the Disk Cache Size
If you have included the INCLUDE_DOSFS_CACHE component, disk caches are
automatically created. Three parameters define the size of the data, directory
entry, and FAT caches: DOSFS_DEFAULT_DATA_CACHE_SIZE (default 128 KB),
DOSFS_DEFAULT_FAT_CACHE_SIZE (default 16 KB), and
DOSFS_DEFAULT_DIR_CACHE_SIZE (default 64 KB).
In addition, the caches can be tuned dynamically using the dosFsCacheInfo( ) and
dosFsCacheTune( ) routines.
You can change the size of the cache for a particular instantiation of the file system
by first destroying the cache with dosFsCacheDelete( ) and then re-creating the
cache with dosFsCacheCreate( ). For more information see the VxWorks API
references for these routines.
A disk cache is intended to reduce the to reduce the number of accesses to the
media. It is not intended for use with RAM disks or TrueFFS. If the cache
component is included with TrueFFS, the cache should be removed
programmatically with dosFsCacheDelete( ).
Step 6:
Check Disk Volume Integrity
Optionally, check the disk for volume integrity using dosFsChkDsk( ). Disk
checking large disks can be time-consuming.
The parameter DOSFS_DEFAULT_CREATE_OPTIONS (of the
INCLUDE_DOSFS_MAIN component) provides an option for checking the disk,
which takes place automatically when the dosfs file system is mounted. This can,
however, be a time consuming process and makes the file system inaccessible until
the check is complete. Alternatively, the check can be performed
programmatically by calling dosFsChkDsk( ).
485
8
VxWorks
Kernel Programmer's Guide, 6.6
8.5.4 dosFs, ATA Disk, and RAM Disk Examples
This section provides examples of the steps discussed in the preceding section.
These examples use a variety of configurations and device types. They are meant
to be relatively generic and applicable to most block devices. The examples
illustrate the following:
■
Creating and working with a dosFs file system on an ATA disk with
commands from the shell.
■
Code that creates and formats partitions.
■
Code that creates and formats a RAM disk volume.
The examples in this section require that VxWorks be configured with the
INCLUDE_DOSFS_FMT component. One example also relies on the
INCLUDE_DOSFS_CACHE component.
!
CAUTION: Because device names are recognized by the I/O system using simple
substring matching, file systems should not use a slash (/) alone as a name;
unexpected results may occur.
Example 8-5
Create dosFs for an ATA Disk
This example demonstrates how to initialize an ATA disk with dosFs from the
shell. While these steps use an XBD-compatible ATA block device, they are
applicable to any XBD-compatible block device.
1.
If you are using a custom driver, create an ATA block device that controls the
master ATA hard disk (drive zero) on the primary ATA controller (controller
zero). This device uses the entire disk.
-> xbd = ataXbdDevCreate(0,0,0,0,"/ata")
New symbol "xbd" added to kernel symbol table.
Instantiating /ata:0 as rawFs
xbd = 0xca4fe0: value = 262145 = 0x40001
The xbd variable is of type device_t. A value of zero would indicate an error
in ataXbdDevCreate( ). Such an error usually indicates a BSP configuration or
hardware configuration error.
If you are using the standard INCLUDE_ATA device component, the block
device is created automatically. Note that in this case the default device name
(provided by the component) is /ata0a.
486
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
2.
Display information about devices.
-> devs
drv name
0 /null
1 /tyCo/0
1 /tyCo/1
8 yow-grand:
9 /vio
4 /ata:0
value = 25 = 0x19
The new ata driver /ata:0 is listed. The zero in the name indicates that no
partitions were detected. Note that if no file system is detected on the device,
the rawFs file system is instantiated automatically and appears the device list.
3.
Create two partitions on this disk device, specifying 50% of the disk space for
the second partition, leaving 50% for the first partition. This step should only
be performed once, when the disk is first initialized. If partitions are already
written to the disk, this step should not be performed since it destroys data.
-> xbdCreatePartition ("/ata:0", 2, 50, 0, 0)
value = 0 = 0x0
4.
Then list the devices to display information about the new partitions.
-> devs
drv name
0 /null
1 /tyCo/0
1 /tyCo/1
8 yow-grand:
9 /vio
3 /ata:1
3 /ata:2
Note that /ata:0 does not appear in this list, and two new devices, /ata:1 and
/ata:2, have been added to represent the new partitions. Each volume has
rawfs instantiated in it as they are new and unformatted.
487
8
VxWorks
Kernel Programmer's Guide, 6.6
5.
Format the volumes for dosFs. This step need only be done once, when the
volumes are first initialized. If the volumes have already been initialized
(formatted), then omit this step. This example formats the file system volumes
with default options.
-> dosFsVolFormat ("/ata:1", 0, 0)
Formatting /ata:1 for DOSFS
Instantiating /ata:1 as rawFs
Formatting...Retrieved old volume params with %100 confidence:
Volume Parameters: FAT type: FAT32, sectors per cluster 8
2 FAT copies, 0 clusters, 38425 sectors per FAT
Sectors reserved 32, hidden 0, FAT sectors 76850
Root dir entries 0, sysId (null) , serial number 3a80000
Label:"
" ...
Disk with 40149184 sectors of 512 bytes will be formatted with:
Volume Parameters: FAT type: FAT32, sectors per cluster 8
2 FAT copies, 5008841 clusters, 39209 sectors per FAT
Sectors reserved 32, hidden 0, FAT sectors 78418
Root dir entries 0, sysId VX5DOS32, serial number 3a80000
Label:"
" ...
OK.
value = 0 = 0x0
-> dosFsVolFormat ("/ata:2", 0, 0)
Formatting /ata:2 for DOSFS
Instantiating /ata:2 as rawFs
Formatting...Retrieved old volume params with %100 confidence:
Volume Parameters: FAT type: FAT32, sectors per cluster 8
2 FAT copies, 0 clusters, 19602 sectors per FAT
Sectors reserved 32, hidden 0, FAT sectors 39204
Root dir entries 0, sysId (null) , serial number c78ff000
Label:"
" ...
Disk with 40144000 sectors of 512 bytes will be formatted with:
Volume Parameters: FAT type: FAT32, sectors per cluster 8
2 FAT copies, 5008195 clusters, 39204 sectors per FAT
Sectors reserved 32, hidden 0, FAT sectors 78408
Root dir entries 0, sysId VX5DOS32, serial number c78ff000
Label:"
" ...
OK.
value = 0 = 0x0
For more information, see the API reference for dosFsFmtLib.
6.
If the INCLUDE_DOSFS_CACHE component is included in VxWorks, 128 KB
data, 16 KB directory and a 64 KB FAT cache are created by default. The size
of these caches can be changed by removing them and creating a new one. The
following example deletes the default caches and creates new ones at twice
their size.
-> dosFsCacheDelete "/ata:1"
value = 0 = 0x0
-> dosFsCacheCreate "/ata:1", 0, 256 * 1024, 0, 32 * 1024, 0, 128 * 1024;
value = 0 = 0x0
488
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
7.
Display information about the dosFs volumes.
-> ll "/ata:1"
Listing Directory /ata:1:
value = 0 = 0x0
-> ll "/ata:2"
Listing Directory /ata:2:
value = 0 = 0x0
-> dosFsShow "/ata:2"
volume descriptor ptr (pVolDesc):
0xc7c358
XBD device block I/O handle: 0x60001
auto disk check on mount:
NOT ENABLED
volume write mode:
copyback (DOS_WRITE)
max # of simultaneously open files:
22
file descriptors in use:
0
# of different files in use:
0
# of descriptors for deleted files:
0
# of obsolete descriptors:
0
8
current volume configuration:
- volume label:
NO LABEL ; (in boot sector:
- volume Id:
0xc78ff000
- total number of sectors:
40,144,000
- bytes per sector:
512
- # of sectors per cluster: 8
- # of reserved sectors:
32
- FAT entry size:
FAT32
- # of sectors per FAT copy:
39,204
- # of FAT table copies:
2
- # of hidden sectors:
0
- first cluster is in sector # 78,440
- Update last access date for open-read-close = FALSE
- directory structure:
VFAT
- file name format:
8-bit (extended-ASCII)
- root dir start cluster:
2
FAT handler information:
------------------------ allocation group size:
- free space on volume:
value = 0 = 0x0
)
501 clusters
20,513,562,620 bytes
Above, we can see the Volume parameters for the /ata:2 volume. The file
system volumes are now mounted and ready to be used.
If you are working with an ATA hard disk or a CD-ROM file system from an
ATAPI CD-ROM drive, you can, alternatively, use usrAtaConfig( ). This
routine processes several steps at once. For more information, see the API
reference.
489
VxWorks
Kernel Programmer's Guide, 6.6
Example 8-6
Creating and Partitioning a Disk and Creating Volumes
This code example takes a pointer to a block device, creates three partitions, creates
the partition handler for these partitions, and creates the dosFs device handler for
them. Then, it formats the partitions using dosFsVolFormat( ).
STATUS usrPartDiskFsInit
(
char * xbdName /* device name used during creation of XBD */
)
{
const char * devNames[] = { "/sd0a", "/sd0b", "/sd0c" };
devname_t xbdPartName;
int newDataCacheSize = 0x40000 /* 256 KB data cache */
int newFatCacheSize = 0x20000 /* 128 KB FAT cache */
int newDirCacheSize = 0x8000 /* 32 KB Dir cache */
DOSFS_CACHE_INFO cacheParams;
int i;
/* Map partition names */
for (i = 1; i <= 3; i++)
{
sprintf (xbdPartName, "%s:d", devNames[i-1],i);
fsmNameInstall (devNames[i], xbdPartName);
}
/* create partitions */
if((xbdCreatePartition (xbdName,3,50,45)) == ERROR)
return ERROR;
/* Formatting the first partition */
if(dosFsVolFormat (devNames[0], 2,0) == ERROR)
return ERROR;
/* Re-configure the cache for the first partition */
if (dosFsCacheCreate (devNames[0], NULL, newDataCacheSize, NULL,
newDirCacheSize, NULL, newFatCacheSize) == ERROR)
return ERROR;
/* Retrieve the current data cache tuning parameters and double them */
if (dosFsCacheInfoGet (devNames[0], DOS_DATA_CACHE, &cacheParams) == ERROR)
return ERROR;
cacheParams.bypass = cacheParams.bypass * 2;
cacheParams.readAhead = cacheParams.readAhead * 2;
if (dosFsCacheTune (devNames[0], DOS_DATA_CACHE, &cacheParams) == ERROR)
return ERROR;
490
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
/* Formatting the second partition */
if(dosFsVolFormat (devNames[1], 2,0) == ERROR)
return ERROR;
/* Formatting the third partition */
if(dosFsVolFormat (devNames[2], 2,0) == ERROR)
return ERROR;
return OK;
}
Note that in most cases you would be likely to format the different partitions for
different file systems.
Example 8-7
Creating and Formatting a RAM Disk Volume
The following code creates a RAM disk and formats it for use with the dosFs file
system.
STATUS usrRamDiskInit
(
void
)
{
int ramDiskSize = 512 * 1024 ;
char *ramDiskDevName = "/ram0" ;
device_t xbd;
/* no argument */
/* 512KB, 512 bytes per sector */
/* 512 byte/sec, no partition support */
xbd = xbdRamDiskDevCreate (512, ramDiskSize, 0, ramDiskDevName);
if( xbd == NULL )
return ERROR ;
/* format the RAM disk, ignore memory contents */
dosFsVolFormat( ramDiskDevName, DOS_OPT_BLANK | DOS_OPT_QUIET, NULL );
return OK;
}
8.5.5 Working with Volumes and Disks
This section discusses accessing volume configuration information and
synchronizing volumes. For information about ioctl( ) support functions, see
8.5.10 I/O Control Functions Supported by dosFsLib, p.500.
491
8
VxWorks
Kernel Programmer's Guide, 6.6
Accessing Volume Configuration Information
The dosFsShow( ) routine can be used to display volume configuration
information from the shell. The dosFsVolDescGet( ) routine can be used
programmatically obtain or verify a pointer to the DOS_VOLUME_DESC structure.
For more information, see the API references for these routines.
Synchronizing Volumes
When a disk is synchronized, all modified buffered data is physically written to the
disk, so that the disk is up to date. This includes data written to files, updated
directory information, and the FAT. To avoid loss of data, a disk should be
synchronized before it is removed. For more information, see the API references
for close( ) and dosFsVolUnmount( ).
8.5.6 Working with Directories
This section discusses creating and removing directories, and reading directory
entries.
Creating Subdirectories
For FAT32, subdirectories can be created in any directory at any time. For FAT12
and FAT16, subdirectories can be created in any directory at any time, except in the
root directory once it reaches its maximum entry count. Subdirectories can be
created in the following ways:
■
Using ioctl( ) with the FIOMKDIR function. The name of the directory to be
created is passed as a parameter to ioctl( ).
■
Using open( ). To create a directory, the O_CREAT option must be set in the
flags parameter and the FSTAT_DIR option must be set in the mode parameter.
The open( ) call returns a file descriptor that describes the new directory. Use
this file descriptor for reading only, and close it when it is no longer needed.
■
Use mkdir( ) from usrFsLib.
When creating a directory using any of the above methods, the new directory
name must be specified. This name can be either a full pathname or a pathname
relative to the current working directory.
492
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
Removing Subdirectories
A directory that is to be deleted must be empty (except for the “.” and “..” entries).
The root directory can never be deleted. Subdirectories can be removed in the
following ways:
■
Using ioctl( ) with the FIORMDIR function, specifying the name of the
directory. The file descriptor used can refer to any file or directory on the
volume, or to the entire volume itself.
■
Using the remove( ) function, specifying the name of the directory.
■
Use rmdir( ) from usrFsLib.
8
Reading Directory Entries
You can programmatically search directories on dosFs volumes using the
opendir( ), readdir( ), rewinddir( ), and closedir( ) routines.
To obtain more detailed information about a specific file, use the fstat( ) or stat( )
routine. Along with standard file information, the structure used by these routines
also returns the file-attribute byte from a directory entry. For more information,
see the API reference for dirLib.
8.5.7 Working with Files
This section discusses file I/O and file attributes.
File I/O Routines
Files on a dosFs file system device are created, deleted, written, and read using the
standard VxWorks I/O routines: creat( ), remove( ), write( ), and read( ). For more
information, see 7.4 Basic I/O, p.365, and the ioLib API references.
File Attributes
The file-attribute byte in a dosFs directory entry consists of a set of flag bits, each
indicating a particular file characteristic. The characteristics described by the
file-attribute byte are shown in Table 8-3.
493
VxWorks
Kernel Programmer's Guide, 6.6
Table 8-3
Flags in the File-Attribute Byte
VxWorks Flag Name
Hex Value
Description
DOS_ATTR_RDONLY
0x01
read-only file
DOS_ATTR_HIDDEN
0x02
hidden file
DOS_ATTR_SYSTEM
0x04
system file
DOS_ATTR_VOL_LABEL
0x08
volume label
DOS_ATTR_DIRECTORY
0x10
subdirectory
DOS_ATTR_ARCHIVE
0x20
file is subject to archiving
DOS_ATTR_RDONLY
If this flag is set, files accessed with open( ) cannot be written to. If the
O_WRONLY or O_RDWR flags are set, open( ) returns ERROR, setting errno to
S_dosFsLib_READ_ONLY.
DOS_ATTR_HIDDEN
This flag is ignored by dosFsLib and produces no special handling. For
example, entries with this flag are reported when searching directories.
DOS_ATTR_SYSTEM
This flag is ignored by dosFsLib and produces no special handling. For
example, entries with this flag are reported when searching directories.
DOS_ATTR_VOL_LABEL
This is a volume label flag, which indicates that a directory entry contains the
dosFs volume label for the disk. A label is not required. If used, there can be
only one volume label entry per volume, in the root directory. The volume
label entry is not reported when reading the contents of a directory (using
readdir( )). It can only be determined using the ioctl( ) function FIOLABELGET.
The volume label can be set (or reset) to any string of 11 or fewer characters,
using the ioctl( ) function FIOLABELSET. Any file descriptor open to the
volume can be used during these ioctl( ) calls.
DOS_ATTR_DIRECTORY
This is a directory flag, which indicates that this entry is a subdirectory, and
not a regular file.
DOS_ATTR_ARCHIVE
This is an archive flag, which is set when a file is created or modified. This flag
is intended for use by other programs that search a volume for modified files
494
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
and selectively archive them. Such a program must clear the archive flag, since
VxWorks does not.
All the flags in the attribute byte, except the directory and volume label flags, can
be set or cleared using the ioctl( ) function FIOATTRIBSET. This function is called
after the opening of the specific file with the attributes to be changed. The
attribute-byte value specified in the FIOATTRIBSET call is copied directly; to
preserve existing flag settings, determine the current attributes using stat( ) or
fstat( ), then change them using bitwise AND and OR operators.
Example 8-8
Setting DosFs File Attributes
This example makes a dosFs file read-only, and leaves other attributes intact.
STATUS changeAttributes
(
void
)
{
int
fd;
struct stat
statStruct;
/* open file */
if ((fd = open ("file", O_RDONLY, 0)) == ERROR)
return (ERROR);
/* get directory entry data */
if (fstat (fd, &statStruct) == ERROR)
return (ERROR);
/* set read-only flag on file */
if (ioctl (fd, FIOATTRIBSET, (statStruct.st_attrib | DOS_ATTR_RDONLY))
== ERROR)
return (ERROR);
/* close file */
close (fd);
return (OK);
}
NOTE: You can also use the attrib( ) routine to change file attributes. For more
information, see the entry in usrFsLib.
495
8
VxWorks
Kernel Programmer's Guide, 6.6
8.5.8 Disk Space Allocation Options
The dosFs file system allocates disk space using one of the following methods. The
first two methods are selected based upon the size of the write operation. The last
method must be manually specified.
■
single cluster allocation
Single cluster allocation uses a single cluster, which is the minimum allocation
unit. This method is automatically used when the write operation is smaller
than the size of a single cluster.
■
cluster group allocation (nearly contiguous)
Cluster group allocation uses adjacent (contiguous) groups of clusters, called
extents. Cluster group allocation is nearly contiguous allocation and is the
default method used when files are written in units larger than the size of a
disk’s cluster.
■
absolutely contiguous allocation
Absolutely contiguous allocation uses only absolutely contiguous clusters.
Because this type of allocation is dependent upon the existence of such space,
it is specified under only two conditions: immediately after a new file is
created and when reading from a file assumed to have been allocated to a
contiguous space. Using this method risks disk fragmentation.
For any allocation method, you can deallocate unused reserved bytes by using the
POSIX-compliant routine ftruncate( ) or the ioctl( ) function FIOTRUNC.
Choosing an Allocation Method
Under most circumstances, cluster group allocation is preferred to absolutely
contiguous file access. Because it is nearly contiguous file access, it achieves a
nearly optimal access speed. Cluster group allocation also significantly minimizes
the risk of fragmentation posed by absolutely contiguous allocation.
Absolutely contiguous allocation attains raw disk throughput levels, however this
speed is only slightly faster than nearly contiguous file access. Moreover,
fragmentation is likely to occur over time. This is because after a disk has been in
use for some period of time, it becomes impossible to allocate contiguous space.
Thus, there is no guarantee that new data, appended to a file created or opened
with absolutely continuous allocation, will be contiguous to the initially written
data segment.
496
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
It is recommended that for a performance-sensitive operation, the application
regulate disk space utilization, limiting it to 90% of the total disk space.
Fragmentation is unavoidable when filling in the last free space on a disk, which
has a serious impact on performance.
Using Cluster Group Allocation
The dosFs file system defines the size of a cluster group based on the media’s
physical characteristics. That size is fixed for each particular media. Since seek
operations are an overhead that reduces performance, it is desirable to arrange
files so that sequential portions of a file are located in physically contiguous disk
clusters. Cluster group allocation occurs when the cluster group size is considered
sufficiently large so that the seek time is negligible compared to the read/write
time. This technique is sometimes referred to as nearly contiguous file access
because seek time between consecutive cluster groups is significantly reduced.
Because all large files on a volume are expected to have been written as a group of
extents, removing them frees a number of extents to be used for new files
subsequently created. Therefore, as long as free space is available for subsequent
file storage, there are always extents available for use. Thus, cluster group
allocation effectively prevents fragmentation (where a file is allocated in small units
spread across distant locations on the disk). Access to fragmented files can be
extremely slow, depending upon the degree of fragmentation.
Using Absolutely Contiguous Allocation
A contiguous file is made up of a series of consecutive disk sectors. Absolutely
contiguous allocation is intended to allocate contiguous space to a specified file (or
directory) and, by so doing, optimize access to that file. You can specify absolutely
contiguous allocation either when creating a file, or when opening a file previously
created in this manner.
For more information on the ioctl( ) functions, see 8.5.10 I/O Control Functions
Supported by dosFsLib, p.500.
497
8
VxWorks
Kernel Programmer's Guide, 6.6
Allocating Contiguous Space for a File
To allocate a contiguous area to a newly created file, follow these steps:
1.
First, create the file in the normal fashion using open( ) or creat( ).
2.
Then, call ioctl( ). Use the file descriptor returned from open( ) or creat( ) as the
file descriptor argument. Specify FIOCONTIG as the function code argument
and the size of the requested contiguous area, in bytes, as the third argument.
The FAT is then searched for a suitable section of the disk. If found, this space is
assigned to the new file. The file can then be closed, or it can be used for further
I/O operations. The file descriptor used for calling ioctl( ) should be the only
descriptor open to the file. Always perform the ioctl( ) FIOCONTIG operation
before writing any data to the file.
To request the largest available contiguous space, use CONTIG_MAX for the size
of the contiguous area. For example:
status = ioctl (fd, FIOCONTIG, CONTIG_MAX);
Allocating Space for Subdirectories
Subdirectories can also be allocated a contiguous disk area in the same manner:
■
If the directory is created using the ioctl( ) function FIOMKDIR, it must be
subsequently opened to obtain a file descriptor to it.
■
If the directory is created using options to open( ), the returned file descriptor
from that call can be used.
A directory must be empty (except for the “.” and “..” entries) when it has
contiguous space allocated to it.
Opening and Using a Contiguous File
Fragmented files require following cluster chains in the FAT. However, if a file is
recognized as contiguous, the system can use an enhanced method that improves
performance. This applies to all contiguous files, whether or not they were
explicitly created using FIOCONTIG. Whenever a file is opened, it is checked for
contiguity. If it is found to be contiguous, the file system registers the necessary
information about that file to avoid the need for subsequent access to the FAT
table. This enhances performance when working with the file by eliminating seek
operations.
When you are opening a contiguous file, you can explicitly indicate that the file is
contiguous by specifying the DOS_O_CONTIG_CHK flag with open( ). This
498
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
prompts the file system to retrieve the section of contiguous space, allocated for
this file, from the FAT table.
To find the maximum contiguous area on a device, you can use the ioctl( ) function
FIONCONTIG. This information can also be displayed by dosFsConfigShow( ).
Example 8-9
Finding the Maximum Contiguous Area on a DosFs Device
In this example, the size (in bytes) of the largest contiguous area is copied to the
integer pointed to by the third parameter to ioctl( ) (count).
STATUS contigTest
(
void
)
{
int count;
int fd;
8
/* no argument */
/* size of maximum contiguous area in bytes */
/* file descriptor */
/* open device in raw mode */
if ((fd = open ("/DEV1/", O_RDONLY, 0)) == ERROR)
return (ERROR);
/* find max contiguous area */
ioctl (fd, FIONCONTIG, &count);
/* close device and display size of largest contiguous area */
close (fd);
printf ("largest contiguous area = %d\n", count);
return (OK);
}
8.5.9 Crash Recovery and Volume Consistency
The DOS file system is inherently susceptible to data structure inconsistencies that
result from interruptions during certain types of disk updates. These types of
interruptions include power failures, system crashes (for fixed disks), and the
manual removal of a disk.
NOTE: The DOS file system is not considered a fault-tolerant file system. The
VxWorks dosFs file system, however, can be used in conjunction with the
Transaction-Based Reliable File System facility; see 7.8.9 Transaction-Based Reliable
File System Facility: TRFS, p.405.
499
VxWorks
Kernel Programmer's Guide, 6.6
Inconsistencies occur because the file system data for a single file is stored in three
separate regions of the disk. The data stored in these regions are:
■
The file chain in the File Allocation Table (FAT), located in a region near the
beginning of the disk.
■
The directory entry, located in a region that could be anywhere on the disk.
■
File clusters containing file data, that could be located anywhere on the disk.
Since all three regions are not always updated before an interruption, dosFs
includes an optional integrated consistency-checking mechanism to detect and
recover from inconsistencies. For example, if a disk is removed when a file is being
deleted, a consistency check completes the file deletion operation. Or, if a file is
being created when an interruption occurs, then the file is un-created. In other
words, the consistency checker either rolls forward or rolls back the operation that
experienced the inconsistency, making whichever correction is possible.
8.5.10 I/O Control Functions Supported by dosFsLib
The dosFs file system supports the ioctl( ) functions. These functions are defined
in the header file ioLib.h along with their associated constants, and they are
described in Table 8-4.
Table 8-4
I/O Control Functions Supported by dosFsLib
Function
Decimal
Value Description
FIOATTRIBSET
35
Sets the file-attribute byte in the dosFs directory entry.
FIOCONTIG
36
Allocates contiguous disk space for a file or directory.
FIODISKCHANGE
13
Announces a media change.
FIODISKFORMAT
5
Formats the disk (device driver function).
FIODISKINIT
6
Initializes a dosFs file system on a disk volume.
FIOFLUSH
2
Flushes the file output buffer.
FIOFSTATGET
38
Gets file status information (directory entry data).
FIOGETNAME
18
Gets the filename of the fd.
FIOLABELGET
33
Gets the volume label.
500
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
Table 8-4
I/O Control Functions Supported by dosFsLib (cont’d)
Function
Decimal
Value Description
FIOLABELSET
34
Sets the volume label.
FIOMKDIR
31
Creates a new directory.
FIOMOVE
47
Moves a file (does not rename the file).
FIONCONTIG
41
Gets the size of the maximum contiguous area on a
device.
FIONFREE
30
Gets the number of free bytes on the volume.
FIONREAD
1
Gets the number of unread bytes in a file.
FIOREADDIR
37
Reads the next directory entry.
FIORENAME
10
Renames a file or directory.
FIORMDIR
32
Removes a directory.
FIOSEEK
7
Sets the current byte offset in a file.
FIOSYNC
21
Same as FIOFLUSH, but also re-reads buffered file
data.
FIOTRUNC
42
Truncates a file to a specified length.
FIOUNMOUNT
39
Un-mounts a disk volume.
FIOWHERE
8
Returns the current byte position in a file.
FIOCONTIG64
49
Allocates contiguous disk space using a 64-bit size.
FIONCONTIG64
50
Gets the maximum contiguous disk space into a
64-bit integer.
FIONFREE64
51
Gets the number of free bytes into a 64-bit integer.
FIONREAD64
52
Gets the number of unread bytes in a file into a 64-bit
integer.
FIOSEEK64
53
Sets the current byte offset in a file from a 64-bit
integer.
8
501
VxWorks
Kernel Programmer's Guide, 6.6
Table 8-4
I/O Control Functions Supported by dosFsLib (cont’d)
Function
Decimal
Value Description
FIOWHERE64
54
Gets the current byte position in a file into a 64-bit
integer.
FIOTRUNC64
55
Set the file's size from a 64-bit integer.
For more information, see the API references for dosFsLib and for ioctl( ) in ioLib.
502
8 Local File Systems
8.5 MS-DOS-Compatible File System: dosFs
8.5.11 Booting from a Local dosFs File System Using SCSI
VxWorks can be booted from a local SCSI device (such as a hard drive in the target
system). Before you can boot from SCSI, you must make a new boot loader that
contains the SCSI library. Configure VxWorks with the INCLUDE_SCSI,
INCLUDE_SCSI_BOOT, and SYS_SCSI_CONFIG components.
After creating the SCSI boot loader ROM, you can prepare the dosFs file system for
use as a boot device. The simplest way to do this is to partition the SCSI device so
that a dosFs file system starts at block 0. You can then make the new system image,
place it on your SCSI boot device, and boot the new VxWorks system. These steps
are shown in more detail below.
!
Step 1:
WARNING: For use as a boot device, the directory name for the dosFs file system
must begin and end with slashes (as with /sd0/ used in the following example).
This is an exception to the usual naming convention for dosFs file systems and is
incompatible with the NFS requirement that device names not end in a slash.
Create the SCSI Device
Create the SCSI device using scsiPhysDevCreate( ) and initialize the disk with a
dosFs file system. Modify the file installDir/vxworks-6.x/target/bspName/sysScsi.c
to reflect your SCSI configuration.
Step 2:
Rebuild Your System
Rebuild your system.
Step 3:
Copy the VxWorks Run-time Image
Copy the file vxWorks to the drive. Below, a VxWorks task spawns the copy( )
routine, passing it two arguments.
The first argument is the source file for the copy( ) command. The source file is the
VxWorks run-time image, vxWorks. The source host name is tiamat:, the source
filename is C:/vxWorks. These are passed to copy( ) in concatenated form, as the
string “tiamat:C:/vxWorks.”
The second argument is the destination file for the copy( ) command. The dosFs
file system, on the local target SCSI disk device, is named /sd0, and the target file
name is vxWorks. These are, similarly, passed to copy( ) in concatenated form, as
the string “/sd0/vxWorks.” When booting the target from the SCSI device, the boot
loader image should specify the run-time file as “/sd0/vxWorks”.
503
8
VxWorks
Kernel Programmer's Guide, 6.6
-> sp (copy, "tiamat:c:/vxWorks","/sd0/vxWorks")
task spawned: id = 0x3f2a200, name = t2
value = 66232832 = 0x3f2a200
Copy OK: 1065570 bytes copied
Step 4:
Copy the System Symbol Table
Depending upon image configuration, the vxWorks.sym file for the system
symbol table may also be needed. Therefore, in similar fashion, copy the
vxWorks.sym file. The run-time image, vxWorks, downloads the vxWorks.sym
file from the same location.
-> sp (copy, "tiamat:c:/vxWorks.sym","/sd0/vxWorks.sym")
task spawned: id = 0x3f2a1bc, name = t3
value = 66232764 = 0x3f2a1bc
Copy OK: 147698 bytes copied
Step 5:
Test the Copying
Now, list the files to ensure that the files were correctly copied.
-> sp (ll, "/sd0")
task spawned: id = 0x3f2a1a8, name = t4
value = 66232744 = 0x3f2a1a8
->
Listing Directory /sd0:
-rwxrwxrwx 1 0
0
-rwxrwxrwx 1 0
0
Step 6:
1065570 Oct 26 2001 vxWorks
147698 Oct 26 2001 vxWorks.sym
Reboot and Change Parameters
Reboot the system, and then change the boot loader parameters. Boot device
parameters for SCSI devices follow this format:
scsi=id,lun
where id is the SCSI ID of the boot device, and lun is its Logical Unit Number
(LUN). To enable use of the network, include the on-board Ethernet device (for
example, ln for LANCE) in the other field.
The following example boots from a SCSI device with a SCSI ID of 2 and a LUN of
0.
504
8 Local File Systems
8.6 Raw File System: rawFs
boot device
processor number
host name
file name
inet on ethernet (e)
host inet (h)
user (u)
flags (f)
target name (tn)
other
:
:
:
:
:
:
:
:
:
:
scsi=2,0
0
host
/sd0/vxWorks
147.11.1.222:ffffff00
147.11.1.3
jane
0x0
t222
ln
8
8.6 Raw File System: rawFs
VxWorks provides a raw file system (rawFs) for use in systems that require only the
most basic disk I/O functions. The rawFs file system, implemented with
rawFsLib, treats the entire disk volume much like a single large file.
Although the dosFs file system provides this ability to varying degrees, the rawFs
file system offers advantages in size and performance if more complex functions
are not required.
The rawFs file system imposes no organization of the data on the disk. It maintains
no directory information; and there is therefore no division of the disk area into
specific files. All open( ) operations on rawFs devices specify only the device
name; no additional filenames are possible.
The entire disk area is treated as a single file and is available to any file descriptor
that is open for the device. All read and write operations to the disk use a
byte-offset relative to the start of the first block on the disk.
A rawFs file system is created by default if inserted media does not contain a
recognizable file system.
8.6.1 Configuring VxWorks for rawFs
To use the rawFs file system, configure VxWorks with the INCLUDE_RAWFS and
INCLUDE_XBD components.
If you are using a device driver that is not designed for use with the XBD facility,
you must use the INCLUDE_XBD_BLK_DEV wrapper component in addition to
INCLUDE_XBD. See XBD Block Device Wrapper, p.404 for more information.
505
VxWorks
Kernel Programmer's Guide, 6.6
Set the NUM_RAWFS_FILES parameter of the INCLUDE_RAWFS component to the
desired maximum open file descriptor count. For information about using
multiple file descriptors with what is essentially a single large file, see 8.6.4 rawFs
File I/O, p.509.
8.6.2 Creating a rawFs File System
The rawFs file system is the default file system. It is created automatically when
VxWorks cannot instantiate a known file system such as dosFs, HRFS, or cdromFs.
Unlike dosFs and HRFS, rawFs does not have a formatter. There are no particular
data structures on the media that signify the disk as being raw. To create a rawFs
file system manually, the current file system must be un-instantiated and replaced
with rawFs. Having two or more file systems on the same media can produce
instabilities in the VxWorks system. Hence, when instantiating a new file system
the previous one must be removed.
See Example 8-10Creating a rawFs File System, p.508 for code that illustrates how this
can be done. (See 8.2 File System Monitor, p.455 for information about default
creation of rawFs.)
The rawFs library rawFsLib is initialized automatically at boot time. The
rawFsInit( ) routine is called by the usrRoot( ) task after starting the VxWorks
system. The rawFsInit( ) routine takes a single parameter, the maximum number
of rawFs file descriptors that can be open at one time. This count is used to allocate
a set of descriptors; a descriptor is used each time a rawFs device is opened. The
parameter can be set with the NUM_RAWFS_FILES configuration parameter of the
INCLUDE_RAWFS component
The rawFsInit( ) routine also makes an entry for the rawFs file system in the I/O
system driver table (with iosDrvInstall( )). This entry specifies the entry points for
rawFs file operations, for all devices that use the rawFs file system. The driver
number assigned to the rawFs file system is placed in a global variable,
rawFsDrvNum.
After the rawFs file system is initialized, one or more devices must be created.
Devices are created with the device driver’s device creation routine
(xxDevCreate( )). The driver routine returns a pointer to a block device descriptor
structure (BLK_DEV). The BLK_DEV structure describes the physical aspects of the
device and specifies the routines in the device driver that a file system can call.
Immediately after its creation, the block device has neither a name nor a file system
associated with it. To initialize a block device for use with rawFs, the
already-created block device must be associated with rawFs and a name must be
506
8 Local File Systems
8.6 Raw File System: rawFs
assigned to it. This is done with the rawFsDevInit( ) routine. Its parameters are the
name to be used to identify the device and a pointer to the block device descriptor
structure (BLK_DEV):
RAW_VOL_DESC *pVolDesc;
BLK_DEV
*pBlkDev;
pVolDesc = rawFsDevInit ("DEV1:", pBlkDev);
The rawFsDevInit( ) call assigns the specified name to the device and enters the
device in the I/O system device table (with iosDevAdd( )). It also allocates and
initializes the file system’s volume descriptor for the device. It returns a pointer to
the volume descriptor to the caller; this pointer is used to identify the volume
during certain file system calls.
Note that initializing the device for use with rawFs does not format the disk. That
is done using an ioctl( ) call with the FIODISKFORMAT function.
NOTE: No disk initialization (FIODISKINIT) is required, because there are no file
system structures on the disk. Note, however, that rawFs accepts that ioctl( )
function code for compatibility with other file systems; in such cases, it performs
no action and always returns OK.
507
8
VxWorks
Kernel Programmer's Guide, 6.6
Example 8-10
Creating a rawFs File System
This example illustrates creating a rawFs file system.
int fd;
device_t xbd;
/* Map some XBD names. Use :0 and :1 since the disk may or may not have
partitions */
fsmNameMap ("/ata:0", "/rawfs");
fsmNameMap ("/ata:1", "/rawfs");
xbd = ataXbdDevCreate (0,0,0,0,"/ata");
/* Get an file descriptor to the current file system */
fd = open ("/rawfs", 0, 0);
/* Register on the path instantiator event */
/* The ejection of the current file system is asynchronous and is handled by
another task.
Depending on relative priorities this may not happen
immediately so the path wait even
facility is used. Each file system will
trip this event when they instatiate to let
waiting task that it is ready.
*/
fsPathAddedEventSetup (&waitData, "/rawfs");
fd = open ("/rawfs", 0, 0);
/* Eject the current file system and put rawfs in its place */
ioctl (fd, XBD_SOFT_EJECT, (int)XBD_TOP);
/* Our FD is now invalid */
/* Wait for the path to instantiate */
fsWaitForPath(&waitData);
Once the call to fsWaitForPath( ) returns the rawfs file system is ready.
8.6.3 Mounting rawFs Volumes
A disk volume is mounted automatically, generally during the first open( ) or
creat( ) operation. (Certain ioctl( ) functions also cause the disk to be mounted.)
The volume is again mounted automatically on the first disk access following a
ready-change operation.
!
CAUTION: Because device names are recognized by the I/O system using simple
substring matching, file systems should not use a slash (/) alone as a name or
unexpected results may occur.
508
8 Local File Systems
8.6 Raw File System: rawFs
8.6.4 rawFs File I/O
To begin I/O operations upon a rawFs device, first open the device using the
standard open( ) routine (or the creat( ) routine). Data on the rawFs device is
written and read using the standard I/O routines write( ) and read( ). For more
information, see 7.4 Basic I/O, p.365.
The character pointer associated with a file descriptor (that is, the byte offset where
the read and write operations take place) can be set by using ioctl( ) with the
FIOSEEK function.
Multiple file descriptors can be open simultaneously for a single device. These
must be carefully managed to avoid modifying data that is also being used by
another file descriptor. In most cases, such multiple open descriptors use FIOSEEK
to set their character pointers to separate disk areas.
8.6.5 I/O Control Functions Supported by rawFsLib
The rawFs file system supports the ioctl( ) functions shown in Table 8-5. The
functions listed are defined in the header file ioLib.h. For more information, see
the API references for rawFsLib and for ioctl( ) in ioLib.
Table 8-5
I/O Control Functions Supported by rawFsLib
Function
Decimal
Value
Description
FIODISKCHANGE
13
Announces a media change.
FIODISKFORMAT
5
Formats the disk (device driver function).
FIOFLUSH
2
Same as FIOSYNC.
FIOGETNAME
18
Gets the device name of the fd.
FIONREAD
1
Gets the number of unread bytes on the device.
FIOSEEK
7
Sets the current byte offset on the device.
FIOSYNC
21
Writes out all modified file descriptor buffers.
FIOUNMOUNT
39
Un-mounts a disk volume.
FIOWHERE
8
Returns the current byte position on the device.
509
8
VxWorks
Kernel Programmer's Guide, 6.6
8.7 CD-ROM File System: cdromFs
The VxWorks CD-ROM file system, cdromFs allows applications to read data from
CDs formatted according to the ISO 9660 standard file system with or without the
Joliet extensions. This section describes how cdromFs is organized, configured,
and used.
The cdromFs library, cdromFsLib, lets applications read any CD-ROMs, CD-Rs, or
CD-RWs (collectively called CDs) that are formatted in accordance with the ISO
9660 file system standard, with or without the Joliet extensions. ISO 9660
interchange level 3, implementation level 2, is supported. Note that multi-extent
files, interleaved files, and files with extended attribute records are supported.
The following CD features and ISO 9660 features are not supported:
■
■
■
■
Multi-volume sets
Record format files
CDs with a sector size that is not a power of two2
Multi-session CD-R or CD-RW3
After initializing cdromFs and mounting it on a CD-ROM block device, you can
access data on that device using the standard POSIX I/O calls: open( ), close( ),
read( ), ioctl( ), readdir( ), and stat( ). The write( ) call always returns an error.
The cdromFs utility supports multiple drives, multiple open files, and concurrent
file access. When you specify a pathname, cdromFS accepts both forward slashes
(/) and back slashes (\) as path delimiters. However, the backslash is not
recommended because it might not be supported in future releases.
The initialization sequence for the cdromFs file system is similar to installing a
dosFs file system on a SCSI or ATA device.
After you have created the CD file system device (8.7.2 Creating and Using cdromFs,
p.512), use ioctl( ) to set file system options. The files system options are described
below:
CDROMFS_DIR_MODE_SET/GET
These options set and get the directory mode. The directory mode controls
whether a file is opened with the Joliet extensions, or without them. The
directory mode can be set to any of the following:
2. Therefore, mode 2/form 2 sectors are not supported, as they have 2324 bytes of user data
per sector. Both mode 1/form 1 and mode 2/form 1 sectors are supported, as they have 2048
bytes of user data per sector.
3. The first session (that is, the earliest session) is always read. The most commonly desired
behavior is to read the last session (that is, the latest session).
510
8 Local File Systems
8.7 CD-ROM File System: cdromFs
MODE_ISO9660
Do not use the Joliet extensions.
MODE_JOLIET
Use the Joliet extensions.
MODE_AUTO
Try opening the directory first without Joliet, and then with Joliet.
!
CAUTION: Changing the directory mode un-mounts the file system. Therefore,
any open file descriptors are marked as obsolete.
CDROMFS_STRIP_SEMICOLON
8
This option sets the readdir( ) strip semicolon setting to FALSE if arg is 0, and
to TRUE otherwise. If TRUE, readdir( ) removes the semicolon and following
version number from the directory entries retrieved.
CDROMFS_GET_VOL_DESC
This option returns, in arg, the primary or supplementary volume descriptor
by which the volume is mounted. arg must be of type T_ISO_PVD_SVD_ID, as
defined in cdromFsLib.h. The result is the volume descriptor, adjusted for the
endianness of the processor (not the raw volume descriptor from the CD). This
result can be used directly by the processor. The result also includes some
information not in the volume descriptor, such as which volume descriptor is
in use.
For information on using cdromFs( ), see the API reference for cdromFsLib.
8.7.1 Configuring VxWorks for cdromFs
To configure VxWorks with cdromFs, add the INCLUDE_CDROMFS and
INCLUDE_XBD components to the kernel. Add other required components (such
as SCSI or ATA) depending on the type of device).
If you are using a device driver that is not designed for use with the XBD facility,
you must use the INCLUDE_XBD_BLK_DEV wrapper component in addition to
INCLUDE_XBD. See XBD Block Device Wrapper, p.404 for more information.
If you are using an ATAPI device, make appropriate modifications to the ataDrv,
ataResources[ ] structure array (if needed). This must be configured appropriately
for your hardware platform.
511
VxWorks
Kernel Programmer's Guide, 6.6
8.7.2 Creating and Using cdromFs
This section describes the steps for creating a block device for the CD-ROM,
creating a cdromFsLib device, mounting the file system, and accessing the media.
The steps are performed from the shell, and shell show routines are used to display
information.
Step 1:
Create a Block Device
Create a block device. The following is an example for an ATAPI master device
upon the secondary ATA controller:
-> xbd = ataXbdDevCreate(1,0,0,0,"/cdrom")
New symbol "xbd" added to kernel symbol table.
xbd = 0xca4fe0: value = 262145 = 0x4000
CDROMFS file system is created automatically if a CD is present in the drive.
Step 2:
Verify cdromFs is instantiated
If a CD is present in the drive, the device name appears in the devs output.
-> devs
drv name
0 /null
1 /tyCo/0
1 /tyCo/1
4 /fd0
5 /ata0a
9 yow-grand:
10 /vio
3 /cdrom:0
Step 3:
Open the Root Directory
This step is optional. It is only required if you plan to perform Step 4 or Step 5, both
of which use the file descriptor obtained in this step.
-> fd = open ("/cdrom:0", 0, 0)
New symbol "fd" added to kernel symbol table.
fd = 0x18cef98: value = 4 = 0x4
In the command-line sequence above, the first 0 is the value of O_RDONLY in
fcntl.h.
Remember to close the root directory in Step 6.
Step 4:
Set readdir( ) To Omit Version Numbers from Its Output
This step is optional.
512
8 Local File Systems
8.7 CD-ROM File System: cdromFs
The strip semicolon mode controls whether readdir( ) returns version numbers.
After mounting, the strip semicolon mode defaults to FALSE, meaning that version
numbers will be returned. If you do not want version numbers, type the following:
-> ioctl (fd, 0x740002, 1)
In the command-line sequence above, 0x740002 is the value of
CDROMFS_STRIP_SEMICOLON in cdromFsLib.h.
Step 5:
Specify Which Volume Descriptor To Use
This step is optional.
The directory mode controls which volume descriptor is used to open a file or
directory. After mounting, the directory mode defaults to MODE_AUTO, meaning
that all volume descriptors will be tried. The directory mode can be changed to
MODE_ISO9660 to use only the ISO 9660 volume descriptors, or to MODE_JOLIET
to use only the Joliet volume descriptors. If either of these modes is selected, and
if the CD does not contain the selected volume descriptor, an
S_cdromFsLib_UNKNOWN_FILE_SYSTEM error is recorded.
-> ioctl (fd, 0x740000, 0)
In the command-line sequence above, 0x740000 is the value of
CDROMFS_DIR_MODE_SET, and 0 is the value of MODE_ISO9660. Both are located
in cdromFsLib.h.
Step 6:
Close the Root Directory
This step is required only if you opened the root directory in Step 3.
-> close (fd)
513
8
VxWorks
Kernel Programmer's Guide, 6.6
Step 7:
Check the Configuration
You can check the CD-ROM configuration using cdromFsVolConfigShow( ):
-> cdromFsVolConfigShow "/cdrom:0"
device config structure ptr
device name
bytes per physical sector
0x18d4dd8
/cdrom:0
2048
Primary directory hierarchy:
514
volume descriptor number
descriptor logical sector
descriptor offset in sector
standard ID
volume descriptor version
UCS unicode level (0=ISO9660)
system ID
volume ID
volume size
number of logical blocks
volume set size
volume sequence number
logical block size
path table memory size (bytes)
path table size on CD (bytes)
path table entries
volume set ID
:1
:16
:0
:CD001
:1
:0
:
:DELL_P1110
:37773312 = 36 MB
:18444 = 0x480c
:1
:1
:2048
:364
:364
:21
:
volume publisher ID
:
volume data preparer ID
:
volume application ID
:NERO - BURNING ROM
copyright file name
abstract file name
bibliographic file name
creation date
modification date
expiration date
effective date
value = 0 = 0x0
:none
:none
:none
:13.07.2000
:13.07.2000
:00.00.0000
:00.00.0000
12:30:00:00
12:30:00:00
00:00:00:00
00:00:00:00
8 Local File Systems
8.7 CD-ROM File System: cdromFs
8.7.3 I/O Control Functions Supported by cdromFsLib
The cdromFs file system supports the ioctl( ) functions. These functions, and their
associated constants, are defined in the header files ioLib.h and cdromFsLib.h.
Table 8-6 describes the ioctl( ) functions that cdromFsLib supports. For more
information, see the API references for cdromFsLib and for ioctl( ) in ioLib.
Table 8-6
ioctl( ) Functions Supported by cdromFsLib
Function Constant
Decimal
Description
CDROMFS_DIR_MODE_GET
7602176
Gets the volume descriptor(s) used to open files.
CDROMFS_DIR_MODE_SET
7602177
Sets the volume descriptor(s) used to open files.
CDROMFS_GET_VOL_DESC
7602179
Gets the volume descriptor that is currently in use.
CDROMFS_STRIP_SEMICOLON
7602178
Sets the readdir( ) strip version number setting.
FIOFSTATGET
38
Gets file status information (directory entry data).
FIOGETNAME
18
Gets the filename of the file descriptor.
FIOLABELGET
33
Gets the volume label.
FIONREAD
1
Gets the number of unread bytes in a file.
FIONREAD64
52
Gets the number of unread bytes in a file (64-bit
version).
FIOREADDIR
37
Reads the next directory entry.
FIOSEEK
7
Sets the current byte offset in a file.
FIOSEEK64
53
Sets the current byte offset in a file (64-bit version).
FIOUNMOUNT
39
Un-mounts a disk volume.
FIOWHERE
8
Returns the current byte position in a file.
FIOWHERE64
54
Returns the current byte position in a file (64-bit
version).
515
8
VxWorks
Kernel Programmer's Guide, 6.6
8.7.4 Version Numbers
cdromFsLib has a 4-byte version number. The version number is composed of four
parts, from most significant byte to least significant byte:
■
■
■
■
major number
minor number
patch level
build
The version number is returned by cdromFsVersionNumGet( ) and displayed by
cdromFsVersionNumDisplay( ).
8.8 Read-Only Memory File System: ROMFS
ROMFS is a simple, read-only file system that represents and stores files and
directories in a linear way (similar to the tar utility). It is installed in RAM with the
VxWorks system image at boot time. The name ROMFS stands for Read-Only
Memory File System; it does not imply any particular relationship to ROM media.
ROMFS provides the ability to bundle VxWorks applications—or any other files
for that matter—with the operating system. No local disk or network connection
to a remote disk is required for executables or other files. When VxWorks is
configured with the ROMFS component, files of any type can be included in the
operating system image simply by adding them to a ROMFS directory on the host
system, and then rebuilding VxWorks. The build produces a single system image
that includes both the VxWorks and the files in the ROMFS directory.
When VxWorks is booted with this image, the ROMFS file system and the files it
holds are loaded with the kernel itself. ROMFS allows you to deploy files and
operating system as a unit. In addition, process-based applications can be coupled
with an automated startup facility so that they run automatically at boot time.
ROMFS thereby provides the ability to create fully autonomous, multi-process
systems.
ROMFS can also be used to store applications that are run interactively for
diagnostic purposes, or for applications that are started by other applications
under specific conditions (errors, and so on).
516
8 Local File Systems
8.8 Read-Only Memory File System: ROMFS
8.8.1 Configuring VxWorks with ROMFS
VxWorks must be configured with the INCLUDE_ROMFS component to provide
ROMFS facilities.
8.8.2 Building a System With ROMFS and Files
Configuring VxWorks with ROMFS and applications involves the following steps:
1.
Create a ROMFS directory in the project directory on the host system, using
the name /romfs.
2.
Copy the application files into the /romfs directory.
3.
Rebuild VxWorks.
For example, adding a process-based application called myVxApp.vxe from the
command line would look like this:
cd c:\myInstallDir\vxworks-6.1\target\proj\wrSbc8260_diab
mkdir romfs
copy c:\allMyVxApps\myVxApp.vxe romfs
make TOOL=diab
The contents of the romfs directory are automatically built into a ROMFS file
system and combined with the VxWorks image.
The ROMFS directory does not need to be created in the VxWorks project
directory. It can also be created in any location on (or accessible from) the host
system, and the make utility’s ROMFS_DIR macro used to identify where it is in
the build command. For example:
make TOOL=diab ROMFS_DIR="c:\allMyVxApps"
Note that any files located in the romfs directory are included in the system image,
regardless of whether or not they are application executables.
8.8.3 Accessing Files in ROMFS
At run-time, the ROMFS file system is accessed as /romfs. The content of the
ROMFS directory can be browsed using the ls and cd shell commands, and
accessed programmatically with standard file system routines, such as open( ) and
read( ).
For example, if the directory
installDir/vxworks-6.x/target/proj/wrSbc8260_diab/romfs has been created on the
517
8
VxWorks
Kernel Programmer's Guide, 6.6
host, the file foo copied to it, and the system rebuilt and booted; then using cd and
ls from the shell (with the command interpreter) looks like this:
[vxWorks *]# cd /romfs
[vxWorks *]# ls
.
..
foo
[vxWorks *]#
And foo can also be accessed at run-time as /romfs/foo by any applications
running on the target.
8.8.4 Using ROMFS to Start Applications Automatically
ROMFS can be used with various startup mechanisms to start process-based
applications automatically when VxWorks boots.
See the VxWorks Application Programmer’s Guide: Applications and Processes for more
information.
8.9 Target Server File System: TSFS
The Target Server File System (TSFS) is designed for development and diagnostic
purposes. It is a full-featured VxWorks file system, but the files are actually located
on the host system.
NOTE: TSFS is not designed for use with large files (whether application
executables or other files), and performance may suffer when they are greater than
50 KB. For large files, use FTP or NFS instead of TSFS
TSFS provides all of the I/O features of the network driver for remote file access
(see 7.8.5 Non-NFS Network Devices, p.400), without requiring any target
resources—except those required for communication between the target system
and the target server on the host. The TSFS uses a WDB target agent driver to
transfer requests from the VxWorks I/O system to the target server. The target
server reads the request and executes it using the host file system. When you open
a file with TSFS, the file being opened is actually on the host. Subsequent read( )
and write( ) calls on the file descriptor obtained from the open( ) call read from and
write to the opened host file.
518
8 Local File Systems
8.9 Target Server File System: TSFS
The TSFS VIO driver is oriented toward file I/O rather than toward console
operations. TSFS provides all the I/O features that netDrv provides, without
requiring any target resource beyond what is already configured to support
communication between target and target server. It is possible to access host files
randomly without copying the entire file to the target, to load an object module
from a virtual file source, and to supply the filename to routines such as ld( ) and
copy( ).
Each I/O request, including open( ), is synchronous; the calling target task is
blocked until the operation is complete. This provides flow control not available in
the console VIO implementation. In addition, there is no need for WTX protocol
requests to be issued to associate the VIO channel with a particular host file; the
information is contained in the name of the file.
Consider a read( ) call. The driver transmits the ID of the file (previously
established by an open( ) call), the address of the buffer to receive the file data, and
the desired length of the read to the target server. The target server responds by
issuing the equivalent read( ) call on the host and transfers the data read to the
target program. The return value of read( ) and any errno that might arise are also
relayed to the target, so that the file appears to be local in every way.
For detailed information, see the API reference for wdbTsfsDrv.
Socket Support
TSFS sockets are operated on in a similar way to other TSFS files, using open( ),
close( ), read( ), write( ), and ioctl( ). To open a TSFS socket, use one of the
following forms of filename:
"TCP:hostIP:port"
"TCP:hostname:port"
The flags and permissions arguments are ignored. The following examples show
how to use these filenames:
fd = open("/tgtsvr/TCP:phobos:6164",0,0);
/* open socket and connect */
/* to server phobos
*/
fd = open("/tgtsvr/TCP:150.50.50.50:6164",0,0);
/* open socket and
/* connect to server
/* 150.50.50.50
*/
*/
*/
The result of this open( ) call is to open a TCP socket on the host and connect it to
the target server socket at hostname or hostIP awaiting connections on port. The
resultant socket is non-blocking. Use read( ) and write( ) to read and write to the
TSFS socket. Because the socket is non-blocking, the read( ) call returns
519
8
VxWorks
Kernel Programmer's Guide, 6.6
immediately with an error and the appropriate errno if there is no data available
to read from the socket. The ioctl( ) usage specific to TSFS sockets is discussed in
the API reference for wdbTsfsDrv. This socket configuration allows VxWorks to
use the socket facility without requiring sockLib and the networking modules on
the target.
Error Handling
Errors can arise at various points within TSFS and are reported back to the original
caller on the target, along with an appropriate error code. The error code returned
is the VxWorks errno which most closely matches the error experienced on the
host. If a WDB error is encountered, a WDB error message is returned rather than
a VxWorks errno.
Configuring VxWorks for TSFS Use
To use TSFS, configure VxWorks with the INCLUDE_WDB_TSFS component. This
creates the /tgtsvr file system on the target.
The target server on the host system must also be configured for TSFS. This
involves assigning a root directory on your host to TSFS (see the discussion of the
target server -R option in Security Considerations, p.520). For example, on a PC host
you could set the TSFS root to c:\myTarget\logs.
Having done so, opening the file /tgtsvr/logFoo on the target causes
c:\myTarget\logs\logFoo to be opened on the host by the target server. A new file
descriptor representing that file is returned to the caller on the target.
Security Considerations
While TSFS has much in common with netDrv, the security considerations are
different (also see 7.8.5 Non-NFS Network Devices, p.400). With TSFS, the host file
operations are done on behalf of the user that launched the target server. The user
name given to the target as a boot parameter has no effect. In fact, none of the boot
parameters have any effect on the access privileges of TSFS.
In this environment, it is less clear to the user what the privilege restrictions to
TSFS actually are, since the user ID and host machine that start the target server
may vary from invocation to invocation. By default, any host tool that connects to
a target server which is supporting TSFS has access to any file with the same
520
8 Local File Systems
8.9 Target Server File System: TSFS
authorizations as the user that started that target server. However, the target
server can be locked (with the -L option) to restrict access to the TSFS.
The options which have been added to the target server startup routine to control
target access to host files using TSFS include:
-R Set the root of TSFS.
For example, specifying -R /tftpboot prepends this string to all TSFS filenames
received by the target server, so that /tgtsvr/etc/passwd maps to
/tftpboot/etc/passwd. If -R is not specified, TSFS is not activated and no TSFS
requests from the target will succeed. Restarting the target server without
specifying -R disables TSFS.
8
-RW Make TSFS read-write.
The target server interprets this option to mean that modifying operations
(including file create and delete or write) are authorized. If -RW is not
specified, the default is read only and no file modifications are allowed.
NOTE: For more information about the target server and the TSFS, see the tgtsvr
command reference. For information about specifying target server options from
Workbench, see the Wind River Workbench User’s Guide: Setting Up Your Hardware
and the Wind River Workbench User’s Guide: New Target Server Connections.
Using the TSFS to Boot a Target
For information about using the TSFS to boot a targets, see 3.11 Booting From the
Host File System Using TSFS, p.155.
521
VxWorks
Kernel Programmer's Guide, 6.6
522
9
Network File System: NFS
9.1 Introduction 523
9.2 Configuring VxWorks for an NFS Client 524
9.3 Creating an NFS Client 529
9.4 Configuring VxWorks for an NFS Server 535
9.5 Creating an NFS Server 540
9.1 Introduction
VxWorks provides an implementation of the Network File System (NFS)
application protocol, versions 2 and 3.
The first part of this chapter describes how to configure and use an NFS client,
which enables a VxWorks target to mount remote file systems and access the
contents of those file systems as if they were local. The second part of the chapter
describes how to configure and use an NFS server, which enables a VxWorks
target to export local systems to remote network systems.
NOTE: VxWorks does not normally provide authentication services for NFS
requests. If you need the NFS server to authenticate incoming requests, see the
nfsdInit( ) and mountdInit( ) reference entries for information on authorization
hooks.
523
VxWorks
Kernel Programmer's Guide, 6.6
9.2 Configuring VxWorks for an NFS Client
The VxWorks supports the following NFS client configuration components, which
you can use to include the modules needed to create an NFS client:
■
■
■
■
■
Core NFS Client
NFS Client All
NFS v2 Client
NFS v3 Client
NFS Mount All
These components are described in detail below.
Core NFS Client
The INCLUDE_CORE_NFS_CLIENT component provides nfsCommon, a library
that provides core functionality for NFS clients. This library also provides
parameters that you can use to specify the NFS user ID, group identifier, and the
maximum length of the pathname for a file.
This component contains the following configuration parameters:
NFS_USER_ID
Synopsis: User identifier for NFS access.
Default: 2001
NFS_GROUP_ID
Synopsis: Group identifier for NFS access.
Default: 100
NFS_MAXPATH
Synopsis: Maximum file path length.
Default: 255
NFS Client All
Including the INCLUDE_NFS_CLIENT_ALL component is equivalent to selecting
both INCLUDE_NFS2_CLIENT and INCLUDE_NFS3_CLIENT, the components that
pull in client support for NFS version 2 and NFS version 3 respectively.
524
9 Network File System: NFS
9.2 Configuring VxWorks for an NFS Client
This component requires the INCLUDE_NFS2_CLIENT and
INCLUDE_NFS3_CLIENT components.
NFS v2 Client
The INCLUDE_NFS2_CLIENT component provides nfs2Lib and nfs2Drv, which
provide support of an NFS version two client. Using this client, you can mount
exported NFS (version two) directories to your VxWorks target.
This component requires the INCLUDE_CORE_NFS_CLIENT component.
Initialization
This component also configures the VxWorks image to initialize the NFS v2 client,
which includes a call to nfsAuthUnixSet( ):
nfsAuthUnixSet ( sysBootParams.hostName, NFS_USER_ID, NFS_GROUP_ID, 0,
(int *) 0);
Values for the NFS_USER_ID and NFS_GROUP_ID parameters are taken from the
required INCLUDE_CORE_NFS_CLIENT component.
Parameters
This component contains the following configuration parameters:
NFS2_CLIENT_CACHE_DEFAULT_NUM_LINES
Synopsis: Default number of cache lines.
Default: 16
The NFS client implementation uses a persistent cache, which is structured as
a multi-line buffer cache. This parameter configures the default number of
lines in the NFS v2 client cache. The value can be decreased to as low as 1, or
increased.
You can modify the number of cache lines, either at build time or at run-time.
To configure the cache at run time, call the routine usrNfs2CacheInit( ):
usrNfs2CacheInit (UINT32 numLines, UINT32 lineSize, UINT32 options);
NFS2_CLIENT_CACHE_DEFAULT_LINE_SIZE
Synopsis: Default number of bytes in cache line.
Default: 16384
525
9
VxWorks
Kernel Programmer's Guide, 6.6
The NFS client implementation uses a persistent cache, which is structured as
a multi-line buffer cache. The cache line size must be a power of two.
This parameter configures the default size of the NFS v2 client cache line. The
default value for this is 16384 (16 kB). When combined with the default
number of cache lines, this yields a default cache size of 256 kB.
You can modify the number of cache line size, either at build time or at
run-time. To configure the cache at run time, call the routine
usrNfs2CacheInit( ):
usrNfs2CacheInit (UINT32 numLines, UINT32 lineSize, UINT32 options);
Changing the size of the cache will not affect any existing cache. It will only
impact future caches.
NFS2_CLIENT_CACHE_DEFAULT_OPTIONS
Synopsis: Default cache options.
Default: 0
This parameter configures the default options for the NFS v2 client cache. The
two valid settings for this parameter are:
0
The default value of zero (0) means that the cache will collect the written
data and only send it to the server when the cache line is full, or it needs
to be flushed (no options), which is the default value
1, NFS_CACHE_WRITE_THROUGH
A value of one means that the cache will be write-through.
You can modify the cache options, either at build time or at run-time. To
configure the cache at run time, call the routine usrNfs2CacheInit( ):
usrNfs2CacheInit (UINT32 numLines, UINT32 lineSize, UINT32 options);
NFS v3 Client
The INCLUDE_NFS3_CLIENT component provides the nfs3Lib and nfs3Drv
libraries, which provide support of an NFS version three client. Using this client,
you can mount exported NFS (version three) directories to your VxWorks target.
This component requires the INCLUDE_CORE_NFS_CLIENT component.
526
9 Network File System: NFS
9.2 Configuring VxWorks for an NFS Client
Initialization
This component also configures the VxWorks image to initialize the NFS v3 client,
which includes a call to nfsAuthUnixSet( ):
nfsAuthUnixSet ( sysBootParams.hostName, NFS_USER_ID, NFS_GROUP_ID, 0,
(int *) 0);
Values for the NFS_USER_ID and NFS_GROUP_ID parameters to this routine are
taken from the required INCLUDE_CORE_NFS_CLIENT component.
Parameters
This component contains the following configuration parameters:
NFS3_CLIENT_CACHE_DEFAULT_NUM_LINES
9
Synopsis: Default number of cache lines.
Default: 16
The NFS client implementation uses a persistent cache, which is structured as
a multi-line buffer cache. This parameter configures the default number of
lines in the NFS v3 client cache. The value can be decreased to as low as 1, or
increased.
You can modify the number of cache lines, either at build time or at run-time.
To configure the cache at run time, call the routine usrNfs3CacheInit( ):
usrNfs3CacheInit (UINT32 numLines, UINT32 lineSize, UINT32 options);
NFS3_CLIENT_CACHE_DEFAULT_LINE_SIZE
Synopsis: Default number of bytes in cache line.
Default: 16384
The NFS client implementation uses a persistent cache, which is structured as
a multi-line buffer cache. The cache line size must be a power of two.
This parameter configures the default size of the NFS v3 client cache line. The
default value for this is 16384 (16 kB). When combined with the default
number of cache lines, this yields a default cache size of 256 kB.
You can modify the number of cache line size, either at build time or at
run-time. To configure the cache at run time, call the routine
usrNfs3CacheInit( ):
usrNfs3CacheInit (UINT32 numLines, UINT32 lineSize, UINT32 options);
527
VxWorks
Kernel Programmer's Guide, 6.6
Changing the size of the cache will not affect any existing cache. It will only
impact future caches.
NFS3_CLIENT_CACHE_DEFAULT_OPTIONS
Synopsis: Default cache options.
Default: 0
This parameter configures the default options for the NFS v3 client cache. The
two valid settings for this parameter are:
0
The default value of zero (0) means that the cache will collect the written
data and only send it to the server when the cache line is full, or it needs
to be flushed (no options), which is the default value
1, NFS_CACHE_WRITE_THROUGH
A value of one means that the cache will be write-through.
You can modify the cache options, either at build time or at run-time. To
configure the cache at run time, call the routine usrNfs3CacheInit( ):
usrNfs3CacheInit (UINT32 numLines, UINT32 lineSize, UINT32 options);
Configuring NFS3 Client Writes
The NFS v3 client has one additional configurable parameter that is not available
on NFS v2. According to the RFC, the NFS v3 client can dictate to an NFS v3 server
how it should perform the write operations. At runtime, the NFS v3 client can be
set to inform the server that it should perform writes one of the following styles:
■
■
■
UNSTABLE
FILE_SYNC
DATA_SYNC
The default setting is UNSTABLE.
You can use two routines to configure these options at run-time:
■
■
nfs3StableWriteSet(stable_how mode) lets you set the mode
nfs3StableWriteGet( ) routine gets the current mode
NOTE: This option does not exist in NFS v2.
528
9 Network File System: NFS
9.3 Creating an NFS Client
NFS Mount All
The INCLUDE_NFS_MOUNT_ALL component configures a VxWorks target to
make a boot-time call to nfsMountAll( ). This routine automatically mounts all file
systems exported by the boot host system.
This component requires the INCLUDE_NFS_CLIENT_ALL component.
This component contains the following configuration parameter:
GROUP_EXPORTS
Synopsis: Remove check for GroupName during MountAll.
Default: FALSE
An NFS server can specify that an exported file system is intended for particular
group of clients. Before mounting an exported file system, a client typically checks
that it is a group member. If the client is not a group member, it does not mount
the file system. To configure the client to skip this check, set this parameter to
TRUE. The default of FALSE ensures that the check occurs.
9.3 Creating an NFS Client
To create an NFS client and use it to access a remote networked file system:
1.
Call nfs2Drv( ) or nfs3Drv( ) to initialize support for NFS v2 or v3 or both.
Normally, you should not need to call either of these routines explicitly.
VxWorks calls these routines automatically at boot time if the image includes
INCLUDE_NFS_CLIENT_ALL. For NFS v2 only, use INCLUDE_NFS2_CLIENT.
To initialize NFS v3 only, use INCLUDE_NFS3_CLIENT.
2.
Call nfsAuthUnixSet( ) to configure the user name and user ID the device
should use when communicating with a particular host.
VxWorks calls nfsAuthUnixSet( ) automatically at boot time if the image
contains INCLUDE_NFS_CLIENT_ALL, INCLUDE_NFS2_CLIENT, or
INCLUDE_NFS3_CLIENT. The values for user name and user ID are supplied
by the parameters to INCLUDE_CORE_NFS_CLIENT.
3.
Call hostAdd( ) to configure the host table to know of the host system from
which you want to mount an exported file system.
529
9
VxWorks
Kernel Programmer's Guide, 6.6
VxWorks calls hostAdd( ) for the host system automatically at boot time. If
you want to mount file systems from other remote systems, you need to make
and explicit hostAdd( ) call for those systems.
4.
Call nfsMount( ) or nfsMountAll( ) to actually mount a remote file system.
The nfsMountAll( ) routine queries the specified remote system for a list of
exported file names and then creates NFS client device instances for each
exported file system. To unmount a file system, use nfsUnmount( ). Use
nfsDevShow( ) to display a list of the mounted NFS devices.
As a convenience, the INCLUDE_NFS_MOUNT_ALL component configures an
image to make a boot time call to nfsMountAll( ) to mount all file systems
exported by the boot host.
The following sections supplement and expand on the procedure outlined above.
Exporting File Systems from the Remote NFS Server
For a UNIX NFS server, the /etc/exports file specifies which of the server’s file
systems are exported for mounting by remote NFS clients. If a file system on a
UNIX NFS server is not listed in /etc/exports, the file system is not exported, which
means other machines cannot use NFS to mount it. For example, consider an
/etc/exports file that contains the line:
/usr
The server exports /usr without restriction. If you want to limit access to this
directory, you can include additional parameters on the line. For example:
1.
On the UNIX box, login as root (super user).
2.
Edit: /etc/exports
3.
Specify the path and permission for the file system that you would export.
For example: /usr * (rw)
For more information on these parameters, consult your UNIX system
documentation.
4.
Export the file system, run: exportfs -ra
5.
On the UNIX target, run the NFS daemon: rpc.nfsd
6.
Run: rpc.rquotad
7.
To run mount the daemon, run: rpc.mountd
To check whether NFS is running, use: rpcinfo -p.
530
9 Network File System: NFS
9.3 Creating an NFS Client
Windows systems also support NFS. Thus, it is possible to configure a directory on
a Windows system so that it is exported over NFS. The exact procedures for doing
so depend upon the particular network package you purchased. For more
information, consult the documentation included with your Windows networking
package.
Setting Your NFS Client Name, User ID, and Group ID
Internally, NFS depends upon RPC to handle the remote execution of the
commands (open, read, write, and others) that access the data in the remote file
system. Associated with the RPC protocol is an authentication system known as
AUTH_UNIX. This authentication system requires RPC peers to provide a user
name, a user ID, and a group name. The recipient of an RPC message uses this
information to decide whether to honor or ignore the RPC request.
On a VxWorks host, you can set the NFS user name, user ID, and group name
using the NFS_GROUP_ID and NFS_USER_ID parameters included in the
INCLUDE_CORE_NFS_CLIENT component. You can also set these values by calling
nfsAuthUnixSet( ) or nfsAuthUnixPrompt( ). For example, to use
nfsAuthUnixSet( ) to set the NFS user ID to 1000 and the NFS group ID to 200 for
the machine mars, you would call nfsAuthUnixSet( ) as follows:
-> nfsAuthUnixSet "mars", 1000, 200, 0
The nfsAuthUnixPrompt( ) routine provides a more interactive way of setting the
NFS authentication parameters from the shell.
On UNIX systems, a user ID is specified in the file /etc/passwd. A list of groups that
a user belongs to is specified in the file /etc/group. To configure a default user ID
and group ID, set NFS_USER_ID and NFS_GROUP_ID. The NFS authentication
parameters will take on these values at system startup. If NFS file access is
unsuccessful, make sure that the configuration is correct.
Mounting a Remote File System
After setting your NFS client name, user ID, and group ID, you are ready to call
nfsMount( ) to mount any file system exported by a known host. To add a system
to the list of hosts known to a VxWorks system, call hostAdd( ):
hostAdd ("host", "IPaddress" )
This routine associates a host name with an IP address. Thus, if you wanted to
mount a file system exported by a system called mars, you would need to have
531
9
VxWorks
Kernel Programmer's Guide, 6.6
already called hostAdd( ) for mars. For example, if mars were at 192.168.10.1 you
would need to call hostAdd( ) as follows:
hostAdd ("mars", "192.168.10.1" )
If mars exports a file system called /usr, you can now use a call to nfsMount( ) to
create a local mount of that remotely exported file system. The syntax of an
nfsMount( ) call is as follows:
nfsMount ("hostName", "hostFileSys", "localName")
hostName
The host name of the NFS server that exports the file system you want to
mount.
hostFileSys
The name of the host file system or subdirectory as it is known on the
exporting NFS server system.
localName
The local name to assign to the file system.
For example, if you wanted to mount a remote file system, /d0/, on your target,
wrs, as a device called /myDevice0/, you would make the following call to
nfsMount( ):
nfsMount ("wrs", "/d0/", "/myDevice0/");
The VxWorks target now has access to the contents of /d0/, although using the
device name, /myDevice0/. For example, if the remote device stores the file,
/d0/bigdog, you can access this file from the wrs target using the pathname,
/myDevice0/bigdog. If you want the local device to use the same device name
as is used on the exporting system, use a NULL as the third parameter of the
nfsMount( ) call. For example:
nfsMount ("wrs", "/d0/", NULL);
Or, from the kernel shell:
-> nfsMount "wrs", "/d0/"
On the VxWorks target, the nfsMount( ) call creates the local device, /d0/.
Thus, on the target, the pathname to bigdog is the same as on the exporting
system; that is: /d0/bigdog.
If you do not need to mount the remote file system under a new name, you should
consider using nfsMountAll( ) instead of nfsMount( ). A call to nfsMountAll( )
mounts all file systems that are exported from the remote system and that are
accessible to the specified client.
532
9 Network File System: NFS
9.3 Creating an NFS Client
The syntax of nfsMountAll( ) is as follows:
nfsMountAll( "hostName", "clientName", quietFlag )
hostName
The name of the host from which you want to mount all exported file
systems.
clientName
The name of a client specified in an access list, if any. A NULL clientName
mounts only those file systems that are accessible to any client.
quietFlag
A boolean value that tells nfsMountAll( ) whether to execute in verbose
or silent mode. FALSE indicates verbose mode, and TRUE indicates quiet
mode.
533
9
VxWorks
Kernel Programmer's Guide, 6.6
Using ioctl( ) With Open Files from a Mounted Directory
After opening a file in a mounted directory, you can work with the file using the
ioctl( ) control functions listed in Table 9-1.
Table 9-1
Supported I/O Control Functions for Files Accessed through NFS
IOCTL
Description
FIOGETNAME
Gets the file name of fd and copies it to the buffer referenced
by nameBuf:
status = ioctl (fd, FIOGETNAME, &nameBuf);
FIONREAD
Copies to nBytesUnread the number of bytes remaining in
the file specified by fd:
status = ioctl (fd, FIONREAD, &nBytesUnread);
FIOSEEK
Sets the current byte offset in the file to the position specified
by newOffset. If the seek goes beyond the end-of-file, the file
grows. The end-of-file pointer gets moved to the new
position, and the new space is filled with zeros:
status = ioctl (fd, FIOSEEK, newOffset);
FIOSYNC
Flush data to the remote NFS file. It takes no additional
argument:
status = ioctl (fd, FIOSYNC, 0);
FIOWHERE
Returns the current byte position in the file. This is the byte
offset of the next byte to be read or written. It takes no
additional argument:
position = ioctl (fd, FIOWHERE, 0);
FIOREADDIR
Reads the next directory entry. Use the third argument in the
ioctl( ) call to supply a pointer to a directory descriptor of
type DIR.
DIR dirStruct;
fd = open ("directory", O_RDONLY);
status = ioctl (fd, FIOREADDIR, &dirStruct);
Normally, you do not use the FIOREADDIR functionality
directly. Instead, you would call readdir( ). See the reference
entry for dirLib.
534
9 Network File System: NFS
9.4 Configuring VxWorks for an NFS Server
Table 9-1
Supported I/O Control Functions for Files Accessed through NFS (cont’d)
IOCTL
Description
FIOFSTATGET
Gets file status information (directory entry data). Use the
third argument in the ioctl( ) call to supply a pointer to a stat
structure that is filled with data describing the specified file.
For example:
struct stat statStruct;
fd = open ("file", O_RDONLY);
status = ioctl (fd, FIOFSTATGET, &statStruct);
Normally, you do not use the FIOFSTATGET functionality
directly. Instead, you would stat( ) or fstat( ) routines get file
information. See the manual entry for dirLib.
FIOFSTATFSGET
Gets the file system parameters for and open file descriptor.
Use the third argument in the ioctl( ) call to supply a pointer
to a statfs structure that is filled with data describing the
underlying file system.
statfs statfsStruct;
fd = open ("directory", O_RDONLY);
status = ioctl (fd, FIOFSTATFSGET, &statfsStruct);
Normally, you do not use the FIOFSTATFSGET functionality
directly. Instead, you would stat( ) or fstat( ) routines get file
information. See the manual entry for dirLib.
9.4 Configuring VxWorks for an NFS Server
VxWorks supports the following NFS server configuration components, which
you can use to include the modules needed to create an NFS server:
■
■
■
■
NFS Server
NFS server All
NFS server V2
NFS server V3
These components are discussed in detail below.
535
9
VxWorks
Kernel Programmer's Guide, 6.6
NFS Server
The INCLUDE_CORE_NFS_SERVER component provides nfsd, the library that
supplies the NFS server daemon initialization routines.
For more information on these routines, see the nfsd reference entry. This
component also supplies configuration parameters that set basic configuration
values for the server.
NOTE: The NFS server maintains a reply cache of up to 64 recently transmitted
reply messages. The purpose of this cache is to deal with RPC retransmissions. If
the XID of an incoming NFS request matches that of a previously replied message,
the cached reply is sent back instead of actually executing the NFS request.
Each entry in the server reply cache requires 8800 bytes for the entry itself and 48
bytes for each cache node. Thus, if there were three entries in the cache, the
memory usage would be 26,544 bytes, which is (3 * 8800) + (3 * 48) bytes. If there
were the full 64 entries in the cache, the memory usage would be 566,272 bytes.
If you think you have noticed a memory leak in NFS, please consider whether this
memory use accounts for the discrepancy.
This component contains the following configuration parameters:
NFS_MAXPATH
Synopsis: Maximum length of a file path string (excludes file name).
Default: 255
NFS_USER_ID
Synopsis: User identifier for NFS access.
Default: 2001
NFS_MAXFILENAME
Synopsis: Maximum file name length. Valid values range from 1 to 99.
Default: 40
This parameter specifies the maximum length of a filename. If a filename is longer
than this value then it is truncated before returning it to the client. This parameter
should be set carefully since it affects the amount of memory allocated by the
server.
536
9 Network File System: NFS
9.4 Configuring VxWorks for an NFS Server
NFS_GROUP_ID
Synopsis: Group identifier for NFS access.
Default: 100
NFS server All
Including the INCLUDE_NFS_SERVER_ALL component is equivalent to selecting
both INCLUDE_NFS3_SERVER and INCLUDE_NFS2_SERVER.
This component requires the INCLUDE_NFS3_SERVER, INCLUDE_NFS2_SERVER,
and INCLUDE_NFS_SERVER_INSTALL components.
9
NFS server V2
The INCLUDE_NFS2_SERVER component provides nfs2dLib and mount1Lib. The
nfs2dLib library implements the NFS version 2 procedures as specified in the
Network File System Protocol Specification (RFC 1094).
The mount1Lib library implements a mount version one server to support
mounting VxWorks file systems remotely. The mount server is an implementation
of version 1 of the mount protocol as defined in RFC 1094.
The INCLUDE_NFS2_SERVER component requires the
INCLUDE_CORE_NFS_SERVER and INCLUDE_NFS_SERVER_INSTALL
components.
NOTE: VxWorks does not normally provide authentication services for NFS
requests. If you need to authenticate incoming requests, see the documentation for
nfsdInit( ) and mountdInit( ) for information about authorization hooks.
To actually export a file system, you must call nfsExport( ).
Table 9-2 lists the requests that RFC 1813 accepted from clients. For details of their
use, see RFC 1094, NFS: Network File System Protocol Specification.
NFS server V3
The INCLUDE_NFS3_SERVER component provides nfs3dLib and mount3Lib. The
nfs3dLib library provides an implementation of the NFS version 3 procedures as
specified in the RFC 1813 (Network File System Protocol Specification). The
537
VxWorks
Kernel Programmer's Guide, 6.6
mount3Lib library implements a mount server. The mount server is an
implementation of version 3 mount protocol as defined in RFC 1813.
NFS version 3 requires mount versions 1 and 3. Also required are
INCLUDE_CORE_NFS_SERVER and INCLUDE_NFS_SERVER_INSTALL.
To actually export a file system, you must call nfsExport( ).
The INCLUDE_NFS3_SERVER component requires the
INCLUDE_CORE_NFS_SERVER and INCLUDE_NFS_SERVER_INSTALL
components.
NOTE: VxWorks does not normally provide authentication services for NFS
requests. If you need to authenticate incoming requests, see the documentation for
nfsdInit( ) and mountdInit( ) for information about authorization hooks.
The following requests are accepted from clients. For details of their use, see RFC
1813: NFS: Network File System Protocol Specification.
Table 9-2
RFC 1813 Supported Client Requests
Procedure Name
Procedure Number
NFSPROC_NULL
0
NFSPROC_GETATTR
1
NFSPROC_SETATTR
2
NFSPROC_LOOKUP
3
NFSPROC_ACCESS
4
NFSPROC_READLINK
5 – not supported, limitation in DOSFS
NFSPROC_READ
6
NFSPROC_WRITE
7
NFSPROC_CREATE
8
NFSPROC_MKDIR
9
NFSPROC_SYMLINK
10 – not supported
NFSPROC_MKNOD
11 – not supported
NFSPROC_REMOVE
12
538
9 Network File System: NFS
9.4 Configuring VxWorks for an NFS Server
Table 9-2
RFC 1813 Supported Client Requests (cont’d)
Procedure Name
Procedure Number
NFSPROC_RMDIR
13
NFSPROC_RENAME
14
NFSPROC_LINK
15 – not supported
NFSPROC_READDIR
16
NFSPROC_READDIRPLUS
17
NFSPROC_FSSTAT
18
NFSPROC_FSINFO
19
NFSPROC_PATHCONF
20
NFSPROC_COMMIT
21
9
539
VxWorks
Kernel Programmer's Guide, 6.6
9.5 Creating an NFS Server
To set up an NFS file server on a VxWorks target, you need to configure the
VxWorks target to include the appropriate NFS server configuration components,
described in 9.4 Configuring VxWorks for an NFS Server, p.535. You also need a file
system for export; for information about file systems, see 8. Local File Systems.
After you have a file system to export, call nfsExport( ) to export the specified file
system.
Initializing an NFS File System for Export
The following code fragment creates a RAM drive, initializes it for dosFs, and
exports the file system for NFS clients on the network:
unsigned myBlockSize; /* block size in bytes */
unsigned myTotalSize; /* disk size in bytes */
myBlockSize = 512;
myTotalSize = 16777216; /* 16Mb */
xbdRamDiskDevCreate (myBlockSize, myTotalSize, FALSE, "/ramDrv");
dosFsVolFormat ("/ramDrv", 2, 0);
nfsExport ("/ramDrv", 0, FALSE, 0);
!
CAUTION: For NFS-exportable file systems, the device name absolutely must not
end in a slash.
Exporting a File System through NFS
After you have an exportable file system, call nfsExport( ) to make it available to
NFS clients on your network. Then mount the file system from the remote NFS
client using the facilities of that system. The following example shows how to
export the new file system from a VxWorks target called vxTarget, and how to
mount it from a typical UNIX system.
1.
After the file system (/export in this example) is initialized, the following
routine call specifies it as a file system to be exported with NFS:
nfsExport ("/export", 0, FALSE, 0);
The first three arguments specify the name of the file system to export; the
VxWorks NFS export ID (0 means to assign one automatically); and whether
540
9 Network File System: NFS
9.5 Creating an NFS Server
to export the file system as read-only. The last argument is a placeholder for
future extensions.
2.
!
To mount the file system from another machine, see the system documentation
for that machine. Specify the name of the VxWorks system that exports the file
system, and the name of the desired file system. You can also specify a
different name for the file system as seen on the NFS client.
CAUTION: On UNIX systems, you normally need root access to mount file systems.
For example, on a typical UNIX system, the following command (executed
with root privilege) mounts the /export file system from the VxWorks system
vxTarget, using the name /mnt for it on UNIX:
9
-> /etc/mount vxTarget:/export /mnt
541
VxWorks
Kernel Programmer's Guide, 6.6
542
10
Flash File System Support:
TrueFFS
10.1 Introduction 543
10.2 Overview of Implementation Steps 545
10.3 Creating a System with TrueFFS 546
10.4 Using TrueFFS Shell Commands 558
10.5 Using TrueFFS With HRFS 560
10.1 Introduction
TrueFFS is a flash management facility that provides access to flash memory by
emulating disk access.
It provides VxWorks with block device functionality, which allows either the
dosFs file system (with or without TRFS support) or the HRFS file system to be
used to access flash memory in the same manner as a disk. For information about
the file system facilities, see 8.5 MS-DOS-Compatible File System: dosFs, p.478),
7.8.9 Transaction-Based Reliable File System Facility: TRFS, p.405, and 8.4 Highly
Reliable File System: HRFS, p.459.
In addition, TrueFFS provides full flash media management capabilities.
TrueFFS is a VxWorks-compatible implementation of M-Systems FLite, version
2.0. This system is reentrant, thread-safe, and supported on all CPU architectures
that host VxWorks. TrueFFS consists of the following four layers:
543
VxWorks
Kernel Programmer's Guide, 6.6
■
The core layer, which connects the other layers and handles global facilities,
such as back-grounding, garbage collection, timers, and other system
resources. This layer provides the block device interface for a file system.
■
The flash translation layer, which maintains the block allocation map that
associates the file system’s view of the storage medium with erase blocks in
flash.
■
The Memory Technology Device (MTD) layer, which implements the low-level
programming of the flash medium (map, read, write, and erase functions).
■
The socket layer, which provides an interface between TrueFFS and the board
hardware with board-specific hardware access routines.
Figure 10-1 illustrates the relationship between the file system, TrueFFS layers,
and the flash medium itself.
Figure 10-1
File System, TrueFFS Layers, and Flash
File System
Core
Layer
TrueFFS
Translation Layer
MTDs
Socket Layer
Flash Memory
This chapter provides instructions for using TrueFFS with the MTDs and drivers
that are included in this release. It provides quick-start material for configuring
TrueFFS and formatting TrueFFS drives, and thus presents the basic steps required
to use the default TrueFFS facilities with your application.
It also provides information about creating a boot image region that excludes
TrueFFS, and about writing the boot image to that region.
If you must customize or create new socket drivers or MTDs, or would simply like
more detailed information about TrueFFS technology, see the VxWorks Device
Driver Developer’s Guide: Flash File System Support with TrueFFS.
544
10 Flash File System Support: TrueFFS
10.2 Overview of Implementation Steps
NOTE: This version of the TrueFFS product is a block device driver to VxWorks
that, although intended to be file system neutral, is provided for use with the dosFs
file system or the HRFS file system. The configuration steps for using TrueFFS with
dosFs and HRFS are slightly different.
10.2 Overview of Implementation Steps
This section provides an overview of how to use TrueFFS with VxWorks BSPs that
provide support for the TrueFFS component. To determine if your BSP provides
TrueFFS support, see the online BSP reference documentation (or the file
installDir/vxworks-6.x/target/config/bspName/target.ref).
You may need to write certain sub-components for your application. This is most
often required for the MTD layer. See the VxWorks Device Driver Developer’s Guide:
Flash File System Support with TrueFFS for information in this regard.
To determine if this release provides an MTD suitable for your flash hardware, see
10.3.1 Selecting an MTD, p.546 and Including the MTD Component, p.550.
Step 1:
Select an MTD Component
Choose an MTD, appropriate for your hardware, from those provided with the
TrueFFS product. See 10.3.1 Selecting an MTD, p.546.
Step 2:
Identify the Socket Driver
Ensure that you have a working socket driver. The socket driver is a source code
component, implemented in the file sysTffs.c. For BSPs that already support
TrueFFS, the socket driver is fully defined and located in the BSP directory. See
10.3.2 Identifying the Socket Driver, p.547.
Step 3:
Configure the System
Configure your system for TrueFFS by adding the appropriate VxWorks
components. Minimum support requires components for a file system and the four
TrueFFS layers. See 10.3.3 Configuring VxWorks with TrueFFS, p.548.
545
10
VxWorks
Kernel Programmer's Guide, 6.6
Step 4:
Build the System
Build the system from Workbench or from the command line with the vxprj
facility. See 10.3.4 Building the System, p.551.
Step 5:
Boot the Target and Format the Drives
Boot the target and then format the drives. See 10.3.5 Formatting the Flash, p.552.
Step 6:
Create a Flash Region for Boot Code
Optionally, create a boot image region that excludes TrueFFS, and write the boot
image to that region (this region can alternatively be used for a system startup log
or configuration data). See 10.3.6 Reserving a Region in Flash for a Boot Image, p.554.
Step 7:
Mount the Drive
Mount the file system on a TrueFFS flash drive. See 10.3.7 Mounting the Drive,
p.556.
Step 8:
Test the Drive
Test your drive(s). See 10.3.8 Testing the Drive, p.557.
10.3 Creating a System with TrueFFS
This section provides detailed instructions on configuring VxWorks with the
required TrueFFS and file system components, building the system, formatting the
flash, mounting the drive, and testing the drive. It also provides information about
creating a region for a boot image.
10.3.1 Selecting an MTD
Determine whether any of the MTDs provided with this release support the device
that you intend to use for TrueFFS. Devices are usually identified by their JEDEC
IDs. If you find an MTD appropriate to your flash device, you can use that MTD.
These drivers are also provided in binary form; so you do not need to compile the
MTD source code unless you have modified it.
546
10 Flash File System Support: TrueFFS
10.3 Creating a System with TrueFFS
The directory installDir/vxworks-6.x/target/src/drv/tffs contains the source code
for the following types of MTD components:
■
MTDs that work with several of the devices provided by Intel, AMD, Fujitsu,
and Sharp.
■
Two generic MTDs that can be used for devices complying with CFI.
To better support the out-of-box experience, these MTDs attempt to cover the
widest possible range of devices (in their class) and of bus architectures.
Consequently, the drivers may not provide the performance you may need for the
run-time environment that you want to target. Also note that your hardware
configuration may require some changes to the generic drivers in order to work.
Consider such requirements as word or byte access limitations, or unusual array
configurations.
If the performance and size of the drivers provided do not match your
requirements, you can modify them to better suit your needs. For more
information, see the VxWorks Device Driver Developer’s Guide: Flash File System
Support with TrueFFS.
NOTE: For the list of the MTD components and details about adding the MTD
component to your system, see Including the MTD Component, p.550.
10.3.2 Identifying the Socket Driver
The socket driver that you include in your system must be appropriate for your
BSP. Some BSPs include socket drivers, others do not. The socket driver file is
sysTffs.c and, if provided, it is located in your BSP directory.
If your BSP does not provide this file, follow the procedure described in the
VxWorks Device Driver Developer’s Guide: Flash File System Support with TrueFFS,
which explains how to port a stub version to your hardware.
In either case, the build process requires that a working socket driver (sysTffs.c)
be located in the BSP directory. For more information, see Adding the Socket Driver,
p.551.
547
10
VxWorks
Kernel Programmer's Guide, 6.6
10.3.3 Configuring VxWorks with TrueFFS
To configure a VxWorks systems with TrueFFS, you must include:
■
Components to fully support either the dosFs or HRFS file system.
■
The INCLUDE_XBD_BLKDEV component (see XBD Block Device Wrapper,
p.404).
■
The core layer TrueFFS component, INCLUDE_TFFS.
Some BSPs—like the PowerQuiccII— have a custom MTD in sysTffs.c and do
not need any additional selections. For this BSP you must identify the flash in
which TrueFFS will reside (INCLUDE_SIMM_TFFS).
■
At least one software module from each of the other three TrueFFS layers
(translation, MTD, and socket).
You can configure and build your system either from the command line or with
Workbench. For general information on configuration procedures, see the Wind
River Workbench User’s Guide and the VxWorks Command-Line Tools User’s Guide.
For either configuration and build method, special consideration must be given to
cases in which either the socket driver or the MTD, or both, are not provided. The
drivers must be registered and MTDs need appropriate component descriptions.
For more information, see the VxWorks Device Driver Developer’s Guide: Flash File
System Support with TrueFFS. For supported BSPs this is usually taken care of
automatically by including components. It is more of a concern when adding
TrueFFS support to a BSP.
NOTE: Included with TrueFFS are sources for several MTDs and socket drivers.
The MTDs are in installDir/vxworks-6.x/target/src/drv/tffs. The socket drivers are
defined in the sysTffs.c files provided in the
installDir/vxworks-6.x/target/config/bspname directory for each BSP that supports
TrueFFS.
Including File System Components
File system support—for either dosFs or HRFS—must be included in your system
for TrueFFS to be useful. For information about these file systems and components
upon which they depend, see 8.5 MS-DOS-Compatible File System: dosFs, p.478 and
8.4 Highly Reliable File System: HRFS, p.459.
548
10 Flash File System Support: TrueFFS
10.3 Creating a System with TrueFFS
There are other file system components that are not required, but which may be
useful. These components add support for the basic functionality needed to use a
file system, such as the commands ls, cd, copy, and so forth (which are provided
by the INCLUDE_DISK_UTIL component).
Including the XBD Wrapper Component
VxWorks must be configured with the INCLUDE_XBD_BLKDEV component for the
TrueFFS block device driver (which is not XBD-compatible). For more
information, see XBD Block Device Wrapper, p.404).
Including the Core TrueFFS Component
10
VxWorks must be configured with the TrueFFS core component, INCLUDE_TFFS.
Including this component triggers the correct sequence of events, at boot time, for
initializing this product. It also ensures that the socket driver is included in your
system (see Adding the Socket Driver, p.551).
Including Utility Components
TrueFFS provides optional utility components for automatic drive detection, show
routines, and writing a boot image to flash.
INCLUDE_TFFS_MOUNT
Including this component adds automatic detection (on booting) of existing
formatted TrueFFS drives.
INCLUDE_TFFS_SHOW
Including this component adds two TrueFFS configuration display utilities,
tffsShow( ) and tffsShowAll( ) for use from the shell.
The tffsShow( ) routine prints device information for a specified socket
interface. It is particularly useful when trying to determine the number of
erase units required to write a boot image (10.3.6 Reserving a Region in Flash for
a Boot Image, p.554). The tffsShowAll( ) routine provides the same information
for all socket interfaces registered with VxWorks. The tffsShowAll( ) routine
can be used from the shell to list the drives in the system. The drives are listed
in the order in which they were registered. This component is not included by
default.
549
VxWorks
Kernel Programmer's Guide, 6.6
INCLUDE_TFFS_BOOT_IMAGE
Including this component provides the tffsBootImagePut( ) routine (in
sysTffs.o). The routine is used to write a boot image to flash memory (see
Writing the Boot Image to Flash, p.556)
NOTE: INCLUDE_TFFS_BOOT_IMAGE is included by default for some BSPs.
Including the MTD Component
Add the MTD component appropriate to your flash device (10.3.1 Selecting an
MTD, p.546) to your system. The MTD components for flash devices from Intel,
AMD, Fujitsu, and Sharp, are described in Table 10-1. For more information about
support for these devices, see the VxWorks Device Driver Developer’s Guide: Flash File
System Support with TrueFFS.
Table 10-1
Components for TrueFFS MTDs
Component
Device
INCLUDE_MTD_CFISCS
CFI/SCS device.
INCLUDE_MTD_CFIAMD
CFI-compliant AMD and Fujitsu devices.
INCLUDE_MTD_I28F016
Intel 28f016 device.
INCLUDE_MTD_I28F008
Intel 28f008 device.
INCLUDE_MTD_AMD
AMD, Fujitsu: 29F0{40,80,16} 8-bit devices.
If you have written your own MTD, you must be sure that it is correctly defined
for inclusion in the system, and that it explicitly requires the transition layer. See
the VxWorks Device Driver Developer’s Guide: Flash File System Support with TrueFFS
for information.
Including the Translation Layer Component
Choose the translation layer appropriate to the technology used by your flash
medium. The main variants of flash devices are NOR and NAND. TrueFFS
provides support for:
■
550
NOR devices.
10 Flash File System Support: TrueFFS
10.3 Creating a System with TrueFFS
■
NAND devices that conform to the SSFDC specification.
The translation layer is provided in binary form only. The translation layer
components are listed in Table 10-2.
Table 10-2
Components for TrueFFS Translation Layers
Component
Description
INCLUDE_TL_FTL
The translation layer for NOR flash devices. If you can
execute code in flash, your device uses NOR logic.
INCLUDE_TL_SSFDC
The translation layer for devices that conform to Toshiba
Solid State Floppy Disk Controller Specifications.
TrueFFS supports only those NAND devices that comply
with the SSFDC specification.
The component descriptor files specify the dependency between the translation
layers and the MTDs; therefore, when configuring through Workbench or the
command-line vxprj facility, you do not need to explicitly select a translation layer.
The build process handles it for you.
For more information about the translation layer, see the VxWorks Device Driver
Developer’s Guide: Flash File System Support with TrueFFS.
Adding the Socket Driver
Inclusion of the socket driver is automatic for BSPs that provide the driver. By
including the core TrueFFS component, INCLUDE_TFFS, in VxWorks, the build
process checks for a socket driver, sysTffs.c, in the BSP directory and includes that
file in the system.
If your BSP does not provide a socket driver, follow the procedure described in the
VxWorks Device Driver Developer’s Guide: Flash File System Support with TrueFFS for
writing a socket driver. To include the socket driver in your system, a working
version of the socket driver (sysTffs.c) must be located in you BSP directory.
10.3.4 Building the System
Build the system from Workbench or from the command line with vxprj.
551
10
VxWorks
Kernel Programmer's Guide, 6.6
10.3.5 Formatting the Flash
!
WARNING: If the flash array for your system is used for a boot image (a boot loader
or self-booting VxWorks system) as well as file system media, space must be
reserved for the boot image before you format the flash for TrueFFS. For more
information, see 10.3.6 Reserving a Region in Flash for a Boot Image, p.554.
First, boot your system. After the system boots and registers the socket driver(s),
start the shell. From the shell, run tffsDevFormat( ) or sysTffsFormat( ) to format
the flash memory for use with TrueFFS—use the latter if the BSP provides it
(sysTffsFormat( ) performs some setup operations and then calls
tffsDevFormat( )).
For example, the tffsDevFormat( ) routine takes two arguments, a drive number
and a format argument:
tffsDevFormat (int tffsDriveNo, int formatArg);
NOTE: You can format the flash medium even if though block device driver has
not yet been associated with the flash.
NOTE: The size of the flash media available for use is reduced by one sector for
TrueFFS internal use.
Specifying the Drive Number
The first argument for tffsDevFormat( ), tffsDriveNo, is the drive number (socket
driver number), which identifies the flash medium to be formatted. Most systems
have a single flash drive, but TrueFFS supports up to five.
The socket registration process determines the drive number. Drive numbers are
assigned to the flash devices on the basis of the order in which the socket drivers
are registered in sysTffsInit( ) at boot time. The first registered is drive 0, the
second is drive 1, and so on up to 4. Details of this process are described in see the
VxWorks Device Driver Developer’s Guide: Flash File System Support with TrueFFS.
Specifying Format Options
The second argument for tffsDevFormat( ), formatArg, is a pointer to a
tffsDevFormatParams structure (cast to an int). This structure describes how the
552
10 Flash File System Support: TrueFFS
10.3 Creating a System with TrueFFS
volume should be formatted. The tffsDevFormatParams structure is defined in
installDir/vxworks-6.x/target/h/tffs/tffsDrv.h:
typedef struct
{
tffsFormatParams formatParams;
unsigned
formatFlags;
}tffsDevFormatParams;
TFFS_STD_FORMAT_PARAMS
To facilitate calling tffsDevFormat( ) from the shell, you can simply pass zero (or
a NULL pointer) for the second argument, formatArg. Doing makes use of the
TFFS_STD_FORMAT_PARAMS macro, which defines default values for the
tffsDevFormatParams structure. The macro defines the default values used in
formatting a flash disk device.
Do not use this macro if the flash device is shared with a boot loader. If the BSP
provides sysTffsFormat( ) use that routine instead.
TFFS_STD_FORMAT_PARAMS is defined in tffsDrv.h as:
#define TFFS_STD_FORMAT_PARAMS {{0, 99, 1, 0x10000l, NULL, {0,0,0,0},
NULL, 2, 0, NULL}, FTL_FORMAT_IF_NEEDED}
If the second argument, formatArg, is zero, tffsDevFormat( ) uses the default
values from this macro.
The macro passes values for both the first and second members of the
tffsDevFormatParams structure. These are:
formatParams = {0, 99, 1, 0x10000l, NULL, {0,0,0,0}, NULL, 2, 0, NULL}
formatFlags = FTL_FORMAT_IF_NEEDED
The meaning of these default values, and other possible arguments for the
members of this structure, are described below.
formatParams
The formatParams member is of the type tffsFormatParams. Both this structure,
and the default values used by the TFFS_STD_FORMAT_PARAMS macro, are
defined in installDir/vxworks-6.x/target/h/tffs/tffsDrv.h.
If you use the TFFS_STD_FORMAT_PARAMS macro, the default values will format
the entire flash medium for use with TrueFFS.
If you want to store a boot image in flash, you must change the value of the
bootImageLen member of the tffsFormatParams structure to reserve a region
flash for the image, which is separate from that used by TrueFFS (for more
information see 10.3.6 Reserving a Region in Flash for a Boot Image, p.554).
553
10
VxWorks
Kernel Programmer's Guide, 6.6
formatFlags
The second member of the tffsDevFormatParams structure, formatFlags,
determines the option used to format the drive. There are several possible values
for formatFlags, which are listed in Table 10-3.
Table 10-3
Options for formatFlags
Macro
Value
Meaning
FTL_FORMAT
1
FAT and FTL formatting
FTL_FORMAT_IF_NEEDED
2
FAT formatting, FTL formatting if needed
NO_FTL_FORMAT
0
FAT formatting only
The default macro TFFS_STD_FORMAT_PARAMS passes
FTL_FORMAT_IF_NEEDED as the value for this argument.
10.3.6 Reserving a Region in Flash for a Boot Image
In order to use flash media for a boot image as well as a file system, a portion of
the flash memory (a fallow region) must be reserved so that it is excluded from the
area subject to formatting and the run-time operations of TrueFFS. Note that the
fallow region can be used for purposes other than boot loader code (for example,
a system startup log or configuration data).
If TrueFFS is used with the vxWorks_romResident VxWorks image type, TrueFFS
must be read-only. For information about boot image types, see 3.3 Boot Loader
Image Types, p.133 and 2.4.1 VxWorks Image Types, p.15.
Reserving a Fallow Region
To create a boot image region, TrueFSS must be configured so that it is excluded
from an area of flash that is large enough to accommodate the boot image. If your
BSP provides sysTffsFormat( ), you can use the routine to do so. Otherwise, you
must modify the tffsFormatParams structure.
Using sysTffsFormat( )
Some BSPs provide an optional, BSP-specific, helper routine sysTffsFormat( ),
which can be used to preserve a boot image region.
554
10 Flash File System Support: TrueFFS
10.3 Creating a System with TrueFFS
This routine first sets up a pointer to a tffsFormatParams structure that has been
initialized with a value for bootImageLen (which defines the boot image region);
then it calls tffsDevFormat( ).
Several BSPs, among them the ads860 BSP, include a sysTffsFormat( ) routine that
reserves 0.5 MB for the boot image. For example:
STATUS sysTffsFormat (void)
{
STATUS status;
tffsDevFormatParams params =
{
#define HALF_FORMAT
/* lower 0.5MB for bootimage, upper 1.5MB for TFFS */
#ifdef HALF_FORMAT
{0x80000l, 99, 1, 0x10000l, NULL, {0,0,0,0}, NULL, 2, 0, NULL},
#else
{0x000000l, 99, 1, 0x10000l, NULL, {0,0,0,0}, NULL, 2, 0, NULL},
#endif /* HALF_FORMAT */
FTL_FORMAT_IF_NEEDED
};
/* assume that the drive number 0 is SIMM */
status = tffsDevFormat (0, (int)¶ms);
return (status);
}
For more examples of sysTffsFormat( ) usage, see the socket drivers in
installDir/vxworks-6.x/target/src/drv/tffs/sockets. If your BSP does not provide a
sysTffsFormat( ) routine, then create a similar routine, or pass the appropriate
argument to tffsDevFormat( ).
Modifying the tffsFormatParams Structure
If your BSP does not provide sysTffsFormat( ), you must modify the
tffsFormatParams structure to reserve a fallow region before you call
tffsDevFormat( ).
Change the bootImageLen member of the tffsFormatParams structure to a value
that is at least as large as the boot image. The area defined by bootImageLen is
excluded from TrueFFS activity (formatting and wear-leveling).
For more information about bootImageLen and other members of the structure,
see Specifying Format Options, p.552.
555
10
VxWorks
Kernel Programmer's Guide, 6.6
Writing the Boot Image to Flash
Once you have created a boot image region, you can write the boot image to the
flash device using tffsBootImagePut( ). This routine bypasses TrueFFS (and its
translation layer) and writes directly into any location in flash memory. However,
because tffsBootImagePut( ) relies on a call to tffsRawio( ), you cannot use this
routine once the TrueFFS volume is mounted.
The arguments to tffsBootImagePut( ) are the following:
driveNo
The same drive number as the one used as input to the format routine.
offset
Offset from the start of flash at which the image is written (most often specified
as zero).
filename
Pointer to the boot image.
!
WARNING: Because tffsBootImagePut( ) lets you write directly to any area of flash,
it is possible to accidentally overwrite and corrupt the TrueFFS-managed area of
flash if you do not specify the parameters correctly. For more information about
how to use this utility, see the reference entry for tffsBootImagePut( ) in the
VxWorks API reference.
10.3.7 Mounting the Drive
Use the usrTffsConfig( ) routine to mount the file system on a TrueFFS flash drive.
Its arguments are the following:
drive
Specifies the drive number of the TFFS flash drive; valid values are 0 through
the number of socket interfaces in BSP.
removable
Specifies whether the media is removable. Use 0 for non-removable, 1 for
removable.
fileName
Specifies the mount point, for example, '/tffs0/'.
556
10 Flash File System Support: TrueFFS
10.3 Creating a System with TrueFFS
The following example runs usrTffsConfig( ) to attach a drive to the file system,
and then runs devs to list all drivers:
-> usrTffsConfig 0,0,"/flashDrive0/"
-> devs
drv
0
1
1
5
6
2
name
/null
/tyCo/0
/tyCo/1
host:
/vio
/flashDrive0/
Internally, usrTffsConfig( ) calls other routines, passing the parameters it is given.
Among these routines is tffsDevCreate( ), which creates a TrueFFS block device on
top of the socket driver. This routine takes, as input, a number (0 through 4) that
identifies the socket driver on top of which to construct the TrueFFS block device.
The tffsDevCreate( ) call uses this number as an index into the array of FLSocket
structures. This number is visible later to the file system as the driver number. The
xbdBlkDevCreateSync( ) routine is then called, which sends a message the file
system handler. If the TFFS device has a valid file system installed (dosFS or HRFS)
the file system handler mounts the underlying file system. If there is no valid file
system on the device, rawfs is instantiated on the device. The user can then format
the device with the file system of choice.
After mounting the drive, create the file system (dosFs or HRFS).
10.3.8 Testing the Drive
One way to test your drive is by copying a text file from the host (or from another
type of storage medium) to the flash file system on the target. Then, copy the file
to the console or to a temporary file for comparison, and verify the content. The
following example (using dosFs on TrueFFS) is run from the shell:
->@copy "host:/home/myHost/.cshrc" "/flashDrive0/myCshrc"
Copy Ok: 4266 bytes copied
Value = 0 = 0x0
->@copy "/flashDrive0/myCshrc"
...
...
...
Copy Ok: 4266 bytes copied
Value = 0 = 0x0
557
10
VxWorks
Kernel Programmer's Guide, 6.6
NOTE: The copy command requires the appropriate configuration of dosFs
support components. For more information on dosFs, see 8.5.1 Configuring
VxWorks for dosFs, p.478.
10.4 Using TrueFFS Shell Commands
This section illustrates using TrueFFS shell commands (with the C interpreter) to
access the flash file system. These examples assume that the flash has been
configured into the system in the default configuration for the BSPs shown. For
detailed information on creating a system with TrueFFS, 10.2 Overview of
Implementation Steps, p.545 and 10.3 Creating a System with TrueFFS, p.546.
Target with a Board-Resident Flash Array and a Boot Image Region
This example uses sysTffsFormat( ) to format board-resident flash, preserving the
boot image region. It does not update the boot image, so no call is made to
tffsBootImagePut( ). Then, it mounts the non-removable RFA medium as drive
number 0.
At the shell prompt, type the following commands:
-> sysTffsFormat
-> usrTffsConfig 0,0,"/RFA/"
Target with a Board-Resident Flash Array and a PCMCIA Slot
This example formats RFA and PCMCIA flash for two drives.
The first lines of this example format the board-resident flash by calling the helper
routine, sysTffsFormat( ), which preserves the boot image region. This example
does not update the boot image. It then mounts the drive, numbering it as 0 and
passing 0 as the second argument to usrTffsConfig( ). Zero is used because RFA
is non-removable.
The last lines of the example format PCMCIA flash, passing default format values
to tffsDevFormat( ) for formatting the entire drive. Then, it mounts that drive.
Because PCMCIA is removable flash, it passes 1 as the second argument to
usrTffsConfig( ). (See 10.3.7 Mounting the Drive, p.556 for details on the arguments
to usrTffsConfig( ).)
558
10 Flash File System Support: TrueFFS
10.4 Using TrueFFS Shell Commands
Insert a flash card in the PCMCIA socket. At the shell prompt, type the following
commands:
->
->
->
->
sysTffsFormat
usrTffsConfig 0,0,"/RFA/"
tffsDevFormat 1,0
usrTffsConfig 1,1,"/PCMCIA1/"
Target with a Board-Resident Flash Array and No Boot Image Region Created
This example formats board-resident flash using the default parameters to
tffsDevFormat( ), as described in 10.3.5 Formatting the Flash, p.552. Then, it mounts
the drive, passing 0 as the drive number and indicating that the flash is
non-removable.
At the shell prompt, type the following commands:
10
-> tffsDevFormat 0,0
-> usrTffsConfig 0,0,"/RFA/"
Target with Two PCMCIA Slots
This example formats PCMCIA flash for two drives. Neither format call preserves
a boot image region. Then, it mounts the drives, the first is numbered 0, and the
second is numbered 1. PCMCIA is a removable medium.
Insert a flash card in each PCMCIA socket. At the shell prompt, type the following
commands:
->
->
->
->
tffsDevFormat
usrTffsConfig
tffsDevFormat
usrTffsConfig
0,0
0,1,"/PCMCIA1/"
1,0
1,1,"/PCMCIA2/"
559
VxWorks
Kernel Programmer's Guide, 6.6
10.5 Using TrueFFS With HRFS
This section provides information about implementing TrueFFS with HRFS
programmatically, as well as interactively from the kernel shell.
10.5.1 TrueFFS With HRFS Code Example
The following code fragment illustrates the procedure for creating a TrueFFS block
device, creating an XBD block device wrapper, and formatting the TrueFFS media
for an HRFS file system.
/* create block device for the entire disk, */
if ((pBlkDev = tffsDevCreate (0, 0)) == NULL)
{
printErr ("tffsDevCreate failed.");
return (ERROR);
}
/* wrap this block device into an XBD */
xbd = xbdBlkDevCreateSync(pBlkDev, "/tffs0");
printf("Wrap XBD=0x%x
/*
* If the HRFS file system exists already, it will be
* automatically instantiated and we are done.
*/
/*
* But if a file system does not exist or it isn't HRFS
* format it.
*/
hrfsFormat ("/tffs0", 0, 0, 0);
For a description of the full set of steps needed to implement HRFS with TrueFFS,
see 10.5.2 TrueFFS With HRFS Shell Command Example, p.560.
10.5.2 TrueFFS With HRFS Shell Command Example
The following steps illustrate the procedure for implementing TrueFFS with HRFS
using the kernel shell.
560
10 Flash File System Support: TrueFFS
10.5 Using TrueFFS With HRFS
1.
Perform a low-level format of the flash device. This step must be performed
before you use the flash device for the first time.
> sysTffsFormat
value = 0 = 0x0
2.
Create the TrueFFS block device.
> dev = tffsDevCreate (0,0)
New symbol "dev" added to kernel symbol table.
dev = 0x1ffb840: value = 33536128 = 0x1ffb880 = dev + 0x40
3.
Create an XBD device wrapper.
> xbdBlkDevCreateSync dev, "/tffs"
Instantiating /tffs:0 as rawFs, device = 0x10001 value = 1 = 0x1
4.
Format the TrueFFS device for HRFS.
> hrfsDiskFormat "/tffs:0"
Formatting /tffs:0 for HRFS v1.2
Instantiating /tffs:0 as rawFs, device = 0x10001 Formatting...OK.
value = 0 = 0x0
Note that the hrfsDiskFormat( ) routine is designed for convenient use from
the shell; the hrfsFormat( ) routine is used in code.
5.
The TFFS device is now ready for use. List the contents.
> ll "/tffs:0"
Listing Directory /tffs:0:
drwxrwxrwx 1 0 0 8192 Jan 1 00:05 ./
drwxrwxrwx 1 0 0 8192 Jan 1 00:05 ../
value = 0 = 0x0
After the media and file system and been prepared, for subsequent reboots the
procedure is slightly simpler as the flash media does not have to be formatted for
TrueFFS and HRFS.
1.
Create the TrueFFS block device.
> dev = tffsDevCreate (0,0)
New symbol "dev" added to kernel symbol table.
dev = 0x461eb8: value = 33535048 = 0x1ffb448
2.
Create an XBD device wrapper.
> xbdBlkDevCreateSync dev, "/tffs"
value = 1 = 0x1
3.
The TFFS device is now ready for use. List the contents.
> ll "/tffs:0"
Listing Directory /tffs:0:
drwxrwxrwx 1 0 0 8192 Jan 1 00:05 ./
drwxrwxrwx 1 0 0 8192 Jan 1 00:05 ../
value = 0 = 0x0
561
10
VxWorks
Kernel Programmer's Guide, 6.6
562
11
Error Detection and Reporting
11.1 Introduction 563
11.2 Configuring Error Detection and Reporting Facilities 564
11.3 Error Records 566
11.4 Displaying and Clearing Error Records 568
11.5 Fatal Error Handling Options 569
11.6 Using Error Reporting APIs in Application Code 572
11.7 Sample Error Record 573
11.1 Introduction
VxWorks provides an error detection and reporting facility to help debugging
software faults. It does so by recording software exceptions in a specially
designated area of memory that is not cleared between warm reboots. The facility
also allows for selecting system responses to fatal errors, with alternate strategies
for development and deployed systems.
563
VxWorks
Kernel Programmer's Guide, 6.6
The key features of the error detection and reporting facility are:
■
A persistent memory region in RAM used to retain error records across warm
reboots.
■
Mechanisms for recording various types of error records.
■
Error records that provide detailed information about run-time errors and the
conditions under which they occur.
■
The ability to display error records and clear the error log from the shell.
■
Alternative error-handing options for the system’s response to fatal errors.
■
Macros for implementing error reporting in user code.
The hook routines described in the edrLib API reference can be used as the basis
for implementing custom functionality for non-RAM storage for error records.
For more information about error detection and reporting routines in addition to
that provided in this chapter, see the API reference entries for edrLib, edrShow,
edrErrLogLib, and edrSysDbgLib.
For information about related facilities, see 6.8 Memory Error Detection, p.331.
NOTE: This chapter provides information about facilities available in the VxWorks
kernel. For information about facilities available to real-time processes, see the
corresponding chapter in the VxWorks Application Programmer’s Guide.
11.2 Configuring Error Detection and Reporting Facilities
To use the error detection and reporting facilities:
■
VxWorks must be configured with the appropriate components.
■
A persistent RAM memory region must be configured, and it must be
sufficiently large to hold the error records.
■
Optionally, users can change the system’s default response to fatal errors.
564
11 Error Detection and Reporting
11.2 Configuring Error Detection and Reporting Facilities
11.2.1 Configuring VxWorks
To use the error detection and reporting facility, the kernel must be configured
with the following components:
■
■
■
■
INCLUDE_EDR_PM
INCLUDE_EDR_ERRLOG
INCLUDE_EDR_SHOW
INCLUDE_EDR_SYSDBG_FLAG
As a convenience, the BUNDLE_EDR component bundle may be used to include all
of the above components.
11.2.2 Configuring the Persistent Memory Region
The persistent-memory region is an area of RAM at the top of system memory
specifically reserved for error records. It is protected by the MMU and the
VxWorks vmLib facilities. The memory is not cleared by warm reboots, provided
a VxWorks 6.x boot loader is used.
NOTE: The persistent memory region is not supported for the symmetric multipro-
cessing (SMP) configuration of VxWorks. For general information about VxWorks
SMP and for information about migration, see 15. VxWorks SMP and
15.15 Migrating Code to VxWorks SMP, p.702.
A cold reboot always clears the persistent memory region. The pmInvalidate( )
routine can also be used to explicitly destroy the region (making it unusable) so
that it is recreated during the next warm reboot.
The persistent-memory area is write-protected when the target system includes an
MMU and VxWorks has been configured with MMU support.
The size of the persistent memory region is defined by the PM_RESERVED_MEM
configuration parameter. By default the size is set to six pages of memory.
By default, the error detection and reporting facility uses one-half of whatever
persistent memory is available. If no other applications require persistent memory,
the component may be configured to use almost all of it. This can be accomplished
by defining EDR_ERRLOG_SIZE to be the size of PM_RESERVED_MEM less the size
of one page of memory.
If you increase the size of the persistent memory region beyond the default, you
must create a new boot loader with the same PM_RESERVED_MEM value. The
memory area between RAM_HIGH_ADRS and sysMemTop( ) must be big enough
565
11
VxWorks
Kernel Programmer's Guide, 6.6
to copy the VxWorks boot loader. If it exceeds the sysMemTop( ) limit, the boot
loader may corrupt the area of persistent memory reserved for core dump storage
when it loads VxWorks. The boot loader, must therefore be rebuilt with a lower
RAM_HIGH_ADRS value.
!
WARNING: If the boot loader is not properly configured (as described above), this
could lead into corruption of the persistent memory region when the system boots.
The EDR_RECORD_SIZE parameter can be used to change the default size of error
records. Note that for performance reasons, all records are necessarily the same
size.
The pmShow( ) shell command (for the C interpreter) can be used to display the
amount of allocated and free persistent memory.
For more information about persistent memory, see 6.6 Reserved Memory, p.328
and the pmLib API reference.
!
WARNING: A VxWorks 6.x boot loader must be used to ensure that the persistent
memory region is not cleared between warm reboots. Prior versions of the boot
loader may clear this area.
11.2.3 Configuring Responses to Fatal Errors
The error detection and reporting facilities provide for two sets of responses to
fatal errors. See 11.5 Fatal Error Handling Options, p.569 for information about these
responses, and various ways to select one for a run-time system.
11.3 Error Records
Error records are generated automatically when the system experiences specific
kinds of faults. The records are stored in the persistent memory region of RAM in
a circular buffer. Newer records overwrite older records when the persistent
memory buffer is full.
566
11 Error Detection and Reporting
11.3 Error Records
The records are classified according to two basic criteria:
■
■
event type
severity level
The event type identifies the context in which the error occurred (during system
initialization, or in a process, and so on).
The severity level indicates the seriousness of the error. In the case of fatal errors,
the severity level is also associated with alternative system’s responses to the error
(see 11.5 Fatal Error Handling Options, p.569).
The event types are defined in Table 11-1, and the severity levels in Table 11-2.
Table 11-1
Table 11-2
Event Types
Type
Description
INIT
System initialization events.
BOOT
System boot events.
REBOOT
System reboot (warm boot) events.
KERNEL
VxWorks kernel events.
INTERRUPT
Interrupt handler events.
RTP
Process environment events.
USER
Custom events (user defined).
11
Severity Levels
Severity Level
Description
FATAL
Fatal event.
NONFATAL
Non-fatal event.
WARNING
Warning event.
INFO
Informational event.
The information collected depends on the type of events that occurs. In general, a
complete fault record is recorded. For some events, however, portions of the
567
VxWorks
Kernel Programmer's Guide, 6.6
record are excluded for clarity. For example, the record for boot and reboot events
exclude the register portion of the record.
Error records hold detailed information about the system at the time of the event.
Each record includes the following generic information:
■
■
■
■
■
■
■
■
■
date and time the record was generated
type and severity
operating system version
task ID
process ID, if the failing task in a process
task name
process name, if the failing task is in a process
source file and line number where the record was created
a free form text message
It also optionally includes the following architecture-specific information:
■
■
■
■
■
memory map
exception information
processor registers
disassembly listing (surrounding the faulting address)
stack trace
11.4 Displaying and Clearing Error Records
The edrShow library provides a set of commands for the shell’s C interpreter that
are used for displaying the error records created since the persistent memory
region was last cleared. See Table 11-3.
Table 11-3
Shell Commands for Displaying Error Records
Command
Action
edrShow( )
Show all records.
edrFatalShow( )
Show only FATAL severity level records.
edrInfoShow( )
Show only INFO severity level records.
edrKernelShow( )
Show only KERNEL event type records.
568
11 Error Detection and Reporting
11.5 Fatal Error Handling Options
Table 11-3
Shell Commands for Displaying Error Records (cont’d)
Command
Action
edrRtpShow( )
Show only RTP (process) event type records.
edrUserShow( )
Show only USER event type records.
edrIntShow( )
Show only INTERRUPT event type records.
edrInitShow( )
Show only INIT event type records.
edrBootShow( )
Show only BOOT event type records.
edrRebootShow( )
Show only REBOOT event type records.
The shell’s command interpreter provides comparable commands. See the API
references for the shell, or use the help edr command.
In addition to displaying error records, each of the show commands also displays
the following general information about the error log:
■
■
■
■
■
■
■
total size of the log
size of each record
maximum number of records in the log
the CPU type
a count of records missed due to no free records
the number of active records in the log
the number of reboots since the log was created
See the edrShow API reference for more information.
11.5 Fatal Error Handling Options
In addition to generating error records, the error detection and reporting facility
provides for two modes of system response to fatal errors for each event type:
■
■
debug mode, for lab systems (development)
deployed mode, for production systems (field)
The difference between these modes is in their response to fatal errors in processes
(RTP events). In debug mode, a fatal error in a process results in the process being
569
11
VxWorks
Kernel Programmer's Guide, 6.6
stopped. In deployed mode, as fatal error in a process results in the process being
terminated.
The operative error handling mode is determined by the system debug flag (see
11.5.2 Setting the System Debug Flag, p.571). The default is deployed mode.
Table 11-4 describes the responses in each mode for each of the event types. It also
lists the routines that are called when fatal records are created.
The error handling routines are called response to certain fatal errors. Only fatal
errors—and no other event types—have handlers associated with them. These
handlers are defined in installDir/vxworks-6.x/target/config/comps/src/edrStub.c.
Developers can modify the routines in this file to implement different system
responses to fatal errors. The names of the routines, however, cannot be changed.
Table 11-4
FATAL Error-Handling Options
Event Type
Debug Mode
Deployed
Mode
(default)
INIT
Reboot
Reboot
edrInitFatalPolicyHandler( )
KERNEL
Stop failed
task
Stop failed
task
edrKernelFatalPolicyHandler( )
INTERRUPT
Reboot
Reboot
edrInterruptFatalPolicyHandler( )
RTP
Stop process
Delete
process
edrRtpFatalPolicyHandler( )
Error Handling Routine
Note that when the debugger is attached to the target, it gains control of the system
before the error-handling option is invoked, thus allowing the system to be
debugged even if the error-handling option calls for a reboot.
11.5.1 Configuring VxWorks with Error Handling Options
In order to provide the option of debug mode error handling for fatal errors,
VxWorks must be configured with the INCLUDE_EDR_SYSDBG_FLAG
component, which it is by default. The component allows a system debug flag to
be used to select debug mode, as well as reset to deployed mode (see 11.5.2 Setting
the System Debug Flag, p.571). If INCLUDE_EDR_SYSDBG_FLAG is removed from
VxWorks, the system defaults to deployed mode (see Table 11-4).
570
11 Error Detection and Reporting
11.5 Fatal Error Handling Options
11.5.2 Setting the System Debug Flag
How the error detection and reporting facility responds to fatal errors, beyond
merely recording the error, depends on the setting of the system debug flag. When
the system is configured with the INCLUDE_EDR_SYSDBG_FLAG component, the
flag can be used to set the handling of fatal errors to either debug mode or
deployed mode (the default).
For systems undergoing development, it is obviously desirable to leave the system
in a state that can be more easily debugged; while in deployed systems, the aim is
to have them recover as best as possible from fatal errors and continue operation.
The debug flag can be set in any of the following ways:
■
Statically, with boot loader configuration.
■
Interactively, at boot time.
■
Programmatically, using APIs in application code.
11
When a system boots, the banner displayed on the console displays information
about the mode defined by the system debug flag. For example:
ED&R Policy Mode: Deployed
The modes are identified as Debug, Deployed, or Permanently Deployed. The
latter indicates that the INCLUDE_EDR_SYSDBG_FLAG component is not included
in the system, which means that the mode is deployed and that it cannot be
changed to debug.
For more information, see the following sections and the API reference entry for
edrSysDbgLib (in particular with regard to the edrSystemDebugModeSet( )
routine).
Setting the Debug Flag Statically
The system can be set to either debug mode or deployed mode with the f boot
loader parameter when a boot loader is configured and built. The value of 0x000 is
used to select deployed mode. The value of 0x400 is used to select debug mode. By
default, it is set to deployed mode.
For information about configuring and building boot loaders, see 3.7 Customizing
and Building Boot Loaders, p.146.
571
VxWorks
Kernel Programmer's Guide, 6.6
Setting the Debug Flag Interactively
To change the system debug flag interactively, stop the system when it boots. Then
use the c command at the boot-loader command prompt. Change the value of the
the f parameter: use 0x000 for deployed mode (the default) or to 0x400 for debug
mode.
Setting the Debug Flag Programmatically
The state of the system debug flag can also be changed in user code with the
edrSysDbgLib API.
11.6 Using Error Reporting APIs in Application Code
The edrLib.h file provides a set of convenient macros that developers can use in
their source code to generate error messages (and responses by the system to fatal
errors) under conditions of the developers choosing.
The macros have no effect if VxWorks has not been configured with error detection
and reporting facilities. Code, therefore, must not be conditionally compiled to
make use of these facilities.
The edrLib.h file is in installDir/vxworks-6.x/target/h.
The following macros are provided:
EDR_USER_INFO_INJECT (trace, msg)
Creates a record in the error log with an event type of USER and a severity of
INFO.
EDR_USER_WARNING_INJECT (trace, msg)
Creates a record in the error log with event type of USER and a severity of
WARNING.
EDR_USER_FATAL_INJECT (trace, msg)
Creates a record in the error log with event type of USER and a severity of
FATAL.
572
11 Error Detection and Reporting
11.7 Sample Error Record
All the macros use the same parameters. The trace parameter is a boolean value
indicating whether or not a traceback should be generated for the record. The msg
parameter is a string that is added to the record.
11.7 Sample Error Record
The following is an example of a record generated by a failed kernel task:
==[1/1]==============================================================
Severity/Facility:
FATAL/KERNEL
Boot Cycle:
1
OS Version:
6.0.0
Time:
THU JAN 01 05:15:07 1970 (ticks = 1134446)
Task:
"kernelTask" (0x0068c6c8)
Injection Point:
excArchLib.c:2523
fatal kernel task-level exception!
<<<<>>>>
0x00100000 -> 0x002a48dc: kernel
<<<<>>>>
data access
Exception current instruction address: 0x002110cc
Machine Status Register: 0x0000b032
Data Access Register: 0x50000000
Condition Register: 0x20000080
Data storage interrupt Register: 0x40000000
<<<<>>>>
r0
r3
r6
r9
r12
r15
r18
r21
r24
r27
r30
lr
cr
scSrTblPtr
=
=
=
=
=
=
=
=
=
=
=
=
=
=
0x00210ff8
0x00213a10
0x0068c6c8
0x00000000
0x0000007f
0x00000000
0x00000000
0x00000000
0x00000000
0x00000000
0x00000000
0x00210ff8
0x20000080
0x0047fe4c
sp
r4
r7
r10
r13
r16
r19
r22
r25
r28
r31
ctr
xer
srTblPtr
=
=
=
=
=
=
=
=
=
=
=
=
=
=
0x006e0f50
0x00003032
0x0000003a
0x00000002
0x00000000
0x00000000
0x00000000
0x00000000
0x00000000
0x00000000
0x50000000
0x0024046c
0x20000000
0x0047fe4c
r2
r5
r8
r11
r14
r17
r20
r23
r26
r29
msr
pc
pgTblPtr
573
=
=
=
=
=
=
=
=
=
=
=
=
=
0x00000000
0x00000001
0x00000000
0x00000002
0x00000000
0x00000000
0x00000000
0x00000000
0x00000000
0x006e0f74
0x0000b032
0x002110cc
0x00481000
11
VxWorks
Kernel Programmer's Guide, 6.6
<<<<>>>>
0x2110ac
0x2110b0
0x2110b4
0x2110b8
0x2110bc
0x2110c0
0x2110c4
0x2110c8
*0x2110cc
0x2110d0
0x2110d4
0x2110d8
0x2110dc
0x2110e0
0x2110e4
0x2110e8
2c0b0004
41820024
2c0b0008
41820030
4800004c
3c600021
83e1001c
38633a10
a09f0000
48000048
83e1001c
3c600021
38633a15
809f0000
48000034
83e1001c
cmpi
bc
cmpi
bc
b
lis
lwz
addi
lhz
b
lwz
lis
addi
lwz
b
lwz
crf0,0,r11,0x4 # 4
0xc,2, 0x2110d4 # 0x002110d4
crf0,0,r11,0x8 # 8
0xc,2, 0x2110e8 # 0x002110e8
0x211108 # 0x00211108
r3,0x21 # 33
r31,28(r1)
r3,r3,0x3a10 # 14864
r4,0(r31)
0x211118 # 0x00211118
r31,28(r1)
r3,0x21 # 33
r3,r3,0x3a15 # 14869
r4,0(r31)
0x211118 # 0x00211118
r31,28(r1)
<<<<>>>>
0x0011047c vxTaskEntry
0x00211258 d
574
+0x54 : 0x00211244 ()
+0x18 : memoryDump ()
12
Target Tools
12.1 Introduction 576
12.2 Kernel Shell 577
12.3 Kernel Object-Module Loader 603
12.4 Kernel Symbol Tables 617
12.5 Show Routines 624
12.6 WDB Target Agent 626
12.7 Common Problems 643
575
VxWorks
Kernel Programmer's Guide, 6.6
12.1 Introduction
The Wind River host development environment provides tools that reside and
execute on the host machine. This approach conserves target memory and
resources. However, there are many situations in which it is desirable to make use
of target-resident facilities: a target-resident shell, kernel object-module loader,
debugging facilities, and system symbol table. The uses for these target-resident
tools include the following:
■
Debugging a deployed system over a serial connection.
■
Developing and debugging network protocols, where it is useful to see the
target's view of a network.
■
Loading kernel modules from a target disk, from ROMFS, or over the network,
and running them interactively (or programmatically).
The target based tools are partially independent of each other. For example, the
kernel shell may be used without the kernel object-module loader, and vice versa.
However, for any of the other individual tools to be completely functional, the
system symbol table is required.
In some situations, it may be useful to use both the host-resident development
tools and the target-resident tools at the same time. In this case, additional facilities
are required so that both environments maintain consistent views of the system.
For more information, see 12.4.5 Synchronizing Host and Kernel Modules List and
Symbol Table, p.623.
For the most part, the target-resident facilities work the same as their host
development environment counterparts. For more information, see the
appropriate chapters of the Wind River Workbench User’s Guide and the VxWorks
Command-Line Tools User’s Guide.
This chapter describes the target-resident kernel shell, kernel object-module
loader, debugging facilities, and system symbol table. It also provides an overview
of the most commonly used VxWorks show routines, which are executed from the
shell. In addition, it describes the WDB target agent. WDB is a target-resident,
run-time facility required for connecting host tools with a VxWorks target system.
576
12 Target Tools
12.2 Kernel Shell
12.2 Kernel Shell
For the most part, the target-resident kernel shell works the same as the host shell
(also known as WindSh—for Wind Shell).1 The kernel shell, however, supports
only the C interpreter and command interpreter (see 12.2.2 Kernel and Host Shell
Differences, p.578 for information about other differences).
For detailed information about the host shell and the shell interpreters, see the
Wind River Workbench Host Shell User’s Guide, and the online Wind River Host Shell
API Reference.
Multiple kernel shell sessions may be run simultaneously, which allows for
simultaneous access to the target from the host console and remote connections
made with telnet or rlogin.
NOTE: The kernel shell operates only with int, long, short, char, double, or float
data types.
12
12.2.1 C Interpreter and Command Interpreter
The kernel shell includes both a C interpreter and a command interpreter. Their
basic differences are as follows:
■
The command interpreter is designed primarily for starting, monitoring, and
debugging real-time process (RTP) applications. It can also be used in
conjunction with the kernel object module loader to load and unload kernel
object modules. It provides a UNIX-like shell environment.
■
The C interpreter is designed primarily for monitoring and debugging
kernel-based code. It can be used for loading, running, and unloading object
modules in conjunction with the kernel object-module loader. In addition, it
provides some APIs for starting and monitoring RTP applications. The C
interpreter operates on C routines.
For detailed information about the interpreters, see the Wind River Workbench Host
Shell User’s Guide.
For information about the commands supported by each interpreter, see Interpreter
Commands and References, p.578.
1. In versions of VxWorks prior to 6.0, the kernel shell was called the target shell. The new
name reflects the fact that the target-resident shell runs in the kernel and not in a process.
577
VxWorks
Kernel Programmer's Guide, 6.6
For information about adding new commands to the command interpreter, and
creating interpreters for the kernel shell, see 12.2.19 Adding Custom Commands to the
Command Interpreter, p.594 and 12.2.20 Creating a Custom Interpreter, p.599.
Switching Between Interpreters
To switch between the shell’s C and command interpreters, use the cmd command
when the C interpreter is active to invoke the command interpreter, and the C
command when the command interpreter is active to invoke the C interpreter. The
following example illustrates switching from the C interpreter to the command
interpreter and back again (note the difference in the shell prompt for each
interpreter):
-> cmd
[vxWorks *]# C
->
You can also execute a command from the interpreter that is not active.
Interpreter Commands and References
For information about individual C interpreter routines, see the usrLib, dbgLib,
and usrShellHistLib sections in the VxWorks Kernel API Reference, as well as
entries for the various show routine libraries.
The dbgLib routines are particularly useful (for example, semaphores can be
created and manipulated from the shell). Note that the kernel shell can also call any
C routine that returns a data type supported by the shell (int, long, short, char,
double, or float).
For information about the command interpreter commands, see the VxWorks
Kernel Shell Command Reference.
For information about help available from the kernel shell itself, see 12.2.6 Using
Kernel Shell Help, p.585.
12.2.2 Kernel and Host Shell Differences
The major differences between the target and host shells are:
■
578
The host and kernel shells do not provide exactly the same set of commands.
The kernel shell, for example, has commands related to network, shared data,
12 Target Tools
12.2 Kernel Shell
environment variables, and some other facilities that are not provided by the
host shell. However, the host and kernel shells provide a very similar set of
commands for their command and C interpreters.
■
Each shell has its own distinct configuration parameters, as well as those that
are common to both.
■
Both shells include a command and a C interpreter. The host shell also
provides a Tcl interpreter and a gdb interpreter. The gdb interpreter has about
40 commands and is intended for debugging processes (RTPs); and it
references host file system paths.
■
For the host shell to work, VxWorks must be configured with the WDB target
agent component. For the kernel shell to work, VxWorks be configured with
the kernel shell component, as well as the target-resident symbol tables
component.
■
The host shell can perform many control and information functions entirely on
the host, without consuming target resources.
■
The kernel shell does not require any Wind River host tool support.
■
The host shell uses host system resources for most functions, so that it remains
segregated from the target. This means that the host shell can operate on the
target from the outside, whereas the kernel shell is part of the VxWorks kernel.
For example, because the kernel shell task is created with the taskSpawn( )
VX_UNBREAKABLE option, it is not possible to set breakpoints on a function
executed within the kernel shell task context. Therefore, the user must create a
new task, with sp( ), to make breakable calls. For example, from the kernel
shell you must do this:
-> b printf
-> sp printf, "Test\n"
Whereas from the host shell you can do this:
-> b printf
-> printf ("Test\n")
Conflicts in task priority may also occur while using the kernel shell.
!
WARNING: Shell commands must be used in conformance with the routine
prototype, or they may cause the system to hang.
■
The kernel shell has its own set of terminal-control characters, unlike the host
shell, which inherits its setting from the host window from which it was
invoked. (See 12.2.7 Using Kernel Shell Control Characters, p.586.)
579
12
VxWorks
Kernel Programmer's Guide, 6.6
■
The kernel shell correctly interprets the tilde operator in pathnames for UNIX
and Linux host systems (or remote file systems on a UNIX or Linux host
accessed with ftp, rsh, NFS, and so on), whereas the host shell cannot. For
example, the following command executed from the kernel shell (with the C
interpreter) by user panloki would correctly locate the kernel module
/home/panloki/foo.o on the host system and load it into the kernel:
-> ld < ~/foo.o
■
When the kernel shell encounters a string literal (“...”) in an expression, it
allocates space for the string, including the null-byte string terminator, plus
some additional overhead.2 The value of the literal is the address of the string
in the newly allocated storage. For example, the following expression allocates
12-plus bytes from the target memory pool, enters the string in that memory
(including the null terminator), and assigns the address of the string to x:
-> x = "hello there"
The following expression can be used to return the memory to the target
memory pool (see the memLib reference entry for information on memory
management):
-> free (x)
Furthermore, even when a string literal is not assigned to a symbol, memory
is still permanently allocated for it. For example, the following expression uses
memory that is never freed:
-> printf ("hello there")
This is because if strings were only temporarily allocated, and a string literal
was passed to a routine being spawned as a task, by the time the task executed
and attempted to access the string, the kernel shell would have already
released (and possibly even reused) the temporary storage where the string
was held.
If the accumulation of memory used for strings has an adverse effect on
performance after extended development sessions with the kernel shell, you
can use the strFree() routine (with the C interpreter) or the equivalent
string free command (with the command interpreter).
The host shell also allocates memory on the target if the string is to be used
there. However, it does not allocate memory on the target for commands that
can be performed at the host level (such as lkup( ), ld( ), and so on).
2. The amount of memory allocated is rounded up to the minimum allocation unit for the
architecture in question, plus the amount for the header for that block of memory.
580
12 Target Tools
12.2 Kernel Shell
12.2.3 Configuring VxWorks With the Kernel Shell
The functionality of the kernel shell is provided by a suite of components, some of
which are required, and others of which are optional.
Required Components
To use the kernel shell, you must configure VxWorks with the INCLUDE_SHELL
component. The configuration parameters for this component are described in
Table 12-1.
You must also configure VxWorks with components for symbol table support,
using either the INCLUDE_STANDALONE_SYM_TBL or INCLUDE_NET_SYM_TBL
component. For information about configuring VxWorks with symbol tables, see
12.4.1 Configuring VxWorks with Symbol Tables, p.618.
Table 12-1
INCLUDE_SHELL Configuration Parameters
12
Configuration Parameter
Description
SHELL_SECURE
Access the kernel shell attached to the console
through a login access.
SHELL_STACK_SIZE
Default stack size of kernel shell task.
SHELL_TASK_NAME_BASE Default basename for the kernel shell tasks.
SHELL_TASK_PRIORITY
Priority of the kernel shell tasks.
SHELL_TASK_OPTIONS
Spawning options for the kernel shell tasks.
SHELL_START_AT_BOOT
The kernel shell is launched automatically at boot
time on the console.
SHELL_COMPATIBLE
The kernel shell is configured to be compatible with
the vxWorks 5.5 shell: one shell session, global I/O
redirected, shell task options without the
VX_PRIVATE_ENV bit.
SHELL_DEFAULT_CONFIG
The default configuration parameters for the kernel
shell can be set using this string.
SHELL_FIRST_CONFIG
The configuration parameters for the initial kernel
shell session can be set using this string.
581
VxWorks
Kernel Programmer's Guide, 6.6
Table 12-1
INCLUDE_SHELL Configuration Parameters
Configuration Parameter
Description
SHELL_REMOTE_CONFIG
The configuration parameters for the kernel shell
sessions started for a remote connection can be set
using this string.
Optional Components
Table 12-2 describes components that provide additional shell functionality.
Table 12-2
Optional Shell Components
Component
Description
INCLUDE_DEBUG
Debugging facilities, such as disassembly,
task stack trace, setting a breakpoint,
stepping, and so on.
INCLUDE_SHELL_BANNER
Display the shell banner on startup.
INCLUDE_SHELL_VI_MODE
Editing mode similar to the vi editing mode.
INCLUDE_SHELL_EMACS_MODE
Editing mode similar to the emacs editing
mode.
INCLUDE_SHELL_INTERP_C
C interpreter. See 12.2.1 C Interpreter and
Command Interpreter, p.577.
INCLUDE_SHELL_INTERP_CMD
Command interpreter. See 12.2.1 C
Interpreter and Command Interpreter, p.577.
INCLUDE_STARTUP_SCRIPT
Kernel shell startup script facility.
INCLUDE_SHELL_HISTORY_FILE
Shell history commands.
Table 12-3 describes components that provide additional command interpreter
functionality. They must be used with the INCLUDE_SHELL_INTERP_CMD
component (described above in Table 12-2).
582
12 Target Tools
12.2 Kernel Shell
Table 12-3
Command Interpreter Components
Component
Description
INCLUDE_DISK_UTIL_SHELL_CMD
File system shell commands.
INCLUDE_EDR_SHELL_CMD
Error detection and reporting
shell commands.
INCLUDE_TASK_SHELL_CMD
Task shell commands.
INCLUDE_DEBUG_SHELL_CMD
Debug shell commands.
INCLUDE_SYM_SHELL_CMD
Symbol shell commands.
INCLUDE_VM_SHOW_SHELL_CMD
Virtual memory show shell
commands.
INCLUDE_ADR_SPACE_SHELL_CMD
Address space shell
commands.
INCLUDE_SHARED_DATA_SHOW_SHELL_CMD Shared data show shell
commands.
INCLUDE_MEM_EDR_SHELL_CMD
Memory detection and
reporting shell commands
INCLUDE_MEM_EDR_RTP_SHELL_CMD
Memory detection and
reporting shell commands for
processes (RTPs).
INCLUDE_MODULE_SHELL_CMD
Kernel loader shell command.
INCLUDE_UNLOADER_SHELL_CMD
Kernel unloader shell
command.
INCLUDE_SHL_SHELL_CMD
Shared library commands for
processes.
INCLUDE_RTP_SHELL_CMD
Process shell commands.
INCLUDE_RTP_SHOW_SHELL_CMD
Process show shell
commands.
INCLUDE_HISTORY_FILE_SHELL_CMD
Shell history commands.
Additional components that are useful are the following:
583
12
VxWorks
Kernel Programmer's Guide, 6.6
INCLUDE_DISK_UTIL
Provides file utilities, such as ls and cd (it is required by
INCLUDE_DISK_UTIL_SHELL_CMD).
INCLUDE_SYM_TBL_SHOW
Provides symbol table show routines, such as lkup.
It can also be useful to include components for the kernel object-module loader
and unloader (see 12.3.1 Configuring VxWorks with the Kernel Object-Module Loader,
p.604). These components are required for the usrLib commands that load
modules into, and unload modules from, the kernel (see 12.2.10 Loading and
Unloading Kernel Object Modules, p.587).
Note that the BUNDLE_STANDALONE_SHELL and BUNDLE_NET_SHELL
component bundles are also available to provide for a standalone kernel shell or a
networked kernel shell.
12.2.4 Configuring the Kernel Shell
The kernel shell can be configured statically with various VxWorks component
parameter options (as part of the configuration and build of the operating stem),
as well as configured dynamically from the shell terminal for a shell session.
The default configuration is defined for all shell sessions of the system with the
component parameter SHELL_DEFAULT_CONFIG. However, the configuration for
the initial shell session launched at boot time can be set differently with the
SHELL_FIRST_CONFIG parameter, and the configuration for remote sessions
(telnet or rlogin) can be set with SHELL_REMOTE_CONFIG.
Each of these component parameters provide various sets of shell configuration
variables that can be set from the command line. These include INTERPRETER,
LINE_EDIT_MODE, VXE_PATH, AUTOLOGOUT, and so on.
Some of the configuration variables are dependent on the inclusion of other
VxWorks components in the operating system. For example, RTP_CREATE_STOP
is only available if VxWorks is configured with process support and the command
interpreter component (INCLUDE_RTP and INCLUDE_SHELL_INTERP_CMD).
With the C interpreter, shConfig( ) can be used to reconfigure the shell
interactively. Similarly, using the command interpreter, the shell configuration can
be displayed and changed with the set config command.
584
12 Target Tools
12.2 Kernel Shell
Some useful configuration variables are:
INTERPRETER
Identify the interpreter, either C or Cmd. The default is the first interpreter
registered (the C interpreter).
LINE_EDIT_MODE
Set the line edit mode, either emacs or vi. The default is the first line edit mode
style registered (vi mode).
LINE_LENGTH
Set the shell line length (it cannot be changed dynamically). The default is 256
characters.
12.2.5 Starting the Kernel Shell
The kernel shell starts automatically after VxWorks boots, by default. If a console
window is open over a serial connection, the shell prompt appears after the shell
banner.
For information about booting VxWorks, and starting a console window, see the
Wind River Workbench User’s Guide: Setting up Your Hardware.
The shell component parameter SHELL_START_AT_BOOT controls if an initial
shell session has to be started (TRUE) or not (FALSE). Default is TRUE. If set to
FALSE, the shell session does not start. It is up to the user to start it either
programmatically (from an application), from the host shell, from a telnet or rlogin
shell session or from the wtxConsole (a host tool). Use shellInit( ) or
shellGenericInit( ) to start a shell session.
Note that when a user calls a routine from the kernel shell, the routine is executed
in the context of the shell task. So that if the routine hangs, the shell session will
hang as well.
12.2.6 Using Kernel Shell Help
For either the C or the command interpreter, the help command displays the basic
set of interpreter commands.
For information about references with detailed information on interpreter
commands, see Interpreter Commands and References, p.578. Also see the Wind River
Workbench Host Shell User’s Guide for information about interpreter use.
585
12
VxWorks
Kernel Programmer's Guide, 6.6
12.2.7 Using Kernel Shell Control Characters
The kernel shell has its own set of terminal-control characters, unlike the host shell,
which inherits its setting from the host window from which it was invoked.
Table 12-4 lists the kernel shell’s terminal-control characters. The first four of these
are defaults that can be mapped to different keys using routines in tyLib (see also
tty Special Characters, p.394).
Table 12-4
kernel shell Terminal Control Characters
Command
Description
CTRL+C
Aborts and restarts the shell.
However, if a process is launched with the command interpreter
(using rtp exec), the function of CTRL+C changes. It is used to
interrupt the process.
CTRL+D
Logs out when the terminal cursor is at the beginning of a line.
CTRL+H
Deletes a character (backspace).
CTRL+Q
Resumes output.
CTRL+S
Temporarily suspends output.
CTRL+U
Deletes an entire line.
CTRL+W
If a process is launched with the command interpreter (using
rtp exec), this key sequence suspends the process running in the
foreground.
CTRL+X
Reboots (trap to the ROM monitor).
ESC
Toggles between input mode and edit mode (vi mode only).
The shell line-editing commands are the same as they are for the host shell. See the
ledLib API references.
12.2.8 Kernel Shell History
The history of kernel shell activity can be recorded with the histSave( ) and
histLoad( ) commands for the C interpreter, and the history save and history load
commands for the command interpreter. The commands allow you to save the
586
12 Target Tools
12.2 Kernel Shell
shell history to, and load it from, a file. The commands are provided by the
INCLUDE_SHELL_HISTORY_FILE component and by the
INCLUDE_HISTORY_FILE_SHELL_CMD component.
For more information about these commands, see the usrShellHistLib entry in the
VxWorks Kernel API Reference (for the C interpreter), and the VxWorks Kernel Shell
Command Reference (for the command interpreter).
12.2.9 Defining Kernel Shell Command Aliases
Aliases can be created for shell commands, as with a UNIX shell. They can be
defined programatically using the shellCmdAliasAdd( ) and
shellCmdAliasArrayAdd( ) routines (see Sample Custom Commands, p.599 for
examples).
For information about creating command aliases interactively, see the VxWorks
Command-Line Tools User’s Guide.
12
12.2.10 Loading and Unloading Kernel Object Modules
Kernel object modules can be dynamically loaded into a running VxWorks kernel
with the target-resident loader. For information about configuring VxWorks with
the loader, and about its use, see 12.3 Kernel Object-Module Loader, p.603.
NOTE: For information about working with real-time processes from the shell, see
the Wind River Workbench Host Shell User’s Guide, the VxWorks Application
Programmer’s Guide: Applications and Processes, and the online Wind River Host Shell
API Reference.
The following is a typical load command from the shell, in which the user
downloads appl.o using the C interpreter:
-> ld < /home/panloki/appl.o
The ld( ) command loads an object module from a file, or from standard input into
the kernel. External references in the module are resolved during loading.
Once an application module is loaded into target memory, subroutines in the
module can be invoked directly from the shell, spawned as tasks, connected to an
interrupt, and so on. What can be done with a routine depends on the flags used
to download the object module (visibility of global symbols or visibility of all
symbols).
587
VxWorks
Kernel Programmer's Guide, 6.6
Modules can be reloaded with reld( ), which unloads the previously loaded
module of the same name before loading the new version. Modules can be
unloaded with unld( ).
For more information about ld, see the VxWorks API reference for usrLib. For
more information about reld( ) and unld( ), see the VxWorks API reference for
unldLib. Note that these routines are meant for use from the shell only; they
cannot be used programmatically.
Undefined symbols can be avoided by loading modules in the appropriate order.
Linking independent files before download can be used to avoid unresolved
references if there are circular references between them, or if the number of
modules is unwieldy. The static linker ldarch can be used to link interdependent
files, so that they can only be loaded and unloaded as a unit. (See Statically Linking
Kernel Application Modules, p.63)
Unloading a code module releases all of the resources used when loading the
module, as far as that is possible. This includes removing symbols from the target’s
symbol table, removing the module from the list of modules loaded in the kernel,
removing the text, data, and bss segments from the kernel memory they were
stored in, and freeing that memory. It does not include freeing memory or other
resources (such as semaphores) that were allocated or created by the module itself
while it was loaded.
12.2.11 Debugging with the Kernel Shell
The kernel shell includes task-level debugging utilities for kernel space if VxWorks
has been configured with the INCLUDE_DEBUG component. For information
about the debugging commands available, see the dgbLib entry in the VxWorks
Kernel API Reference for the C interpreter; and see the VxWorks Kernel Shell
Command Reference for the command interpreter.
Debugging SMP Systems with the Kernel Shell
The kernel shell can be used for task mode debugging of SMP systems, but it
cannot be used for system mode debugging. Software breakpoints are always
persistent—that is, retained in target memory (for UP systems they are not). Kernel
shell debug commands are, however, not affected by persistent software
breakpoints. For example, disassembling an address on which a software
breakpoint is installed displays the real instruction.
588
12 Target Tools
12.2 Kernel Shell
Also note the following with regard to using the kernel shell to debug SMP
systems:
■
Breakpoint exceptions that occur while holding an ISR-callable spinlock are
ignored.
■
Breakpoint exceptions that occur while holding a task-only spinlock are
ignored.
■
Breakpoint exceptions that occur while interrupts are locked are ignored.
■
The output of different tasks can be intermingled. For example, creating a task
that performs a printf( ) operation, as follows:
-> sp printf,"Hello world."
Task spawned: id = 0x61707478, name = Hello world.t1
value = 1634759800 = 0x61707478 = 'x'
->
The internal shell mechanism prevents intermingling of lines, but the
intermingling of output cannot be prevented if other tasks or exception
handlers print characters in the midst of a shell output.
For information about the SMP configuration of VxWorks, see 15. VxWorks SMP.
12.2.12 Aborting Routines Executing from the Kernel Shell
Occasionally it is desirable to abort the shell’s evaluation of a statement. For
example, an invoked routine can loop excessively, suspend, or wait on a
semaphore. This can happen because of errors in the arguments specified in the
invocation, errors in the implementation of the routine, or oversight regarding the
consequences of calling the routine. In such cases it is usually possible to abort and
restart the kernel shell task. This is done by pressing the special target-shell abort
character on the keyboard, CTRL+C by default. This causes the kernel shell task to
restart execution at its original entry point. Note that the abort key can be changed
to a character other than CTRL+C by calling tyAbortSet( ).
When restarted, the kernel shell automatically reassigns its system standard input
and output streams to the original assignments they had when the kernel shell was
first spawned. Thus any kernel shell redirections are canceled, and any executing
shell scripts are aborted.
589
12
VxWorks
Kernel Programmer's Guide, 6.6
The abort facility works only if the following are true:
■
excTask( ) is running.
■
The driver for the particular keyboard device supports it (all
VxWorks-supplied drivers do).
Also, you may occasionally enter an expression that causes the kernel shell to incur
a fatal error such as a bus/address error or a privilege violation. Such errors
normally result in the suspension of the offending task, which allows further
debugging.
However, when such an error is incurred by the kernel shell task, VxWorks
automatically restarts the kernel shell, because further debugging is impossible
without it. Note that for this reason, as well as to allow the use of breakpoints and
single-stepping, it is often useful when debugging to spawn a routine as a task
instead of just calling it directly from the kernel shell.
When the kernel shell is aborted for any reason, either because of a fatal error or
because it is aborted from the terminal, a task trace is displayed automatically. This
trace shows where the kernel shell was executing when it died.
Note that an offending routine can leave portions of the system in a state that may
not be cleared when the kernel shell is aborted. For instance, the kernel shell might
have taken a semaphore, which cannot be given automatically as part of the abort.
12.2.13 Console Login Security
Console login security can be provided for the kernel shell by adding the
INCLUDE_SECURITY component to the VxWorks configuration. In addition, the
shell’s SHELL_SECURE component parameter must be set to TRUE (it is set to
FALSE by default).
With this configuration, the shell task is not launched at startup. Instead, a login
task runs on the console, waiting for the user to enter a valid login ID and
password. After validation of the login, the shell task is launched for the console.
When the user logs out from the console, the shell session is terminated, and a new
login task is launched.
Also see Remote Login Security, p.592.
590
12 Target Tools
12.2 Kernel Shell
12.2.14 Using a Remote Login to the Kernel Shell
Users can log into a VxWorks system with telnet and rlogin and use the kernel
shell, provided that VxWorks has been configured with the appropriate
components. VxWorks can also be configured with a remote-login security feature
that imposes user ID and password constraints on access to the system.
Note that VxWorks does not support rlogin access from the VxWorks system to
the host.
Remote Login With telnet and rlogin
When VxWorks is first booted, the shell’s terminal is normally the system console.
You can, however, use telnet to access the kernel shell from a host over the
network if VxWorks is built with the INCLUDE_TELNET_CLIENT component
(which can be configured with the TELNETD_MAX_CLIENTS parameter). This
component creates the tTelnetd task when the system boots. It is possible to start
several shells for different network connections. (Remote login is also available
with the wtxConsole tool.)
To access the kernel shell over the network, use the telnet command with the name
of the target VxWorks system. For example:
% telnet myVxBox
UNIX host systems can also use rlogin to access to the kernel shell from the host.
VxWorks must be configured with the INCLUDE_RLOGIN component to create the
tRlogind task.
To end an rlogin connection to the shell, you can do any of the following:
■
Use the CTRL+D key combination.
■
Use the logout( ) command with the shell’s C interpreter, or the logout
command with the command interpreter.
■
Type the tilde and period characters at the shell prompt:
-> ~.
591
12
VxWorks
Kernel Programmer's Guide, 6.6
Remote Login Security
VxWorks can be configured with a remote-login security feature that imposes user
ID and password constraints on access to the system. The INCLUDE_SECURITY
component provides this facility.
Note that loginEncryptInstall( ) allows for use of other encryption routines (such
as SHA512).
A user is then prompted for a login user name and password when accessing the
VxWorks target remotely. The default login user name and password provided
with the supplied system image is target and password.
The default user name and password can be changed with the loginUserAdd( )
routine, as follows:
-> loginUserAdd "fred", "encrypted_password"
The default user name and password can be changed with loginUserAdd( ), which
requires an encrypted password. To create an encrypted password, use the
vxencrypt tool on the host system. The tool prompts you to enter a password, and
then displays the encrypted version. The user name and password can then be
changed with the loginUserAdd( ) command with the shell’s C interpreter. For
example, mysecret is encrypted as bee9QdRzs, and can be used with the user
name fred as follows to change the default settings:
-> loginUserAdd "fred", "
bee9QdRzs"
To define a group of login names, include a list of loginUserAdd( ) calls in a
startup script and run the script after the system has been booted. Or include the
loginUserAdd( ) calls in usrAppInit( ); for information in this regard, see
2.6.10 Configuring VxWorks to Run Applications Automatically, p.66.
NOTE: The values for the user name and password apply only to remote login into
the VxWorks system. They do not affect network access from VxWorks to a remote
system; See Wind River Network Stack for VxWorks 6 Programmer’s Guide.
The remote-login security feature can be disabled at boot time by specifying the
flag bit 0x20 (SYSFLAG_NO_SECURITY) in the flags boot parameter.
Also see 12.2.13 Console Login Security, p.590.
592
12 Target Tools
12.2 Kernel Shell
12.2.15 Launching a Shell Script Programmatically
A simple way to have a script executed programmatically by an interpreter (the
command interpreter for example) is as follows:
fdScript = open ("myScript", O_RDONLY);
shellGenericInit ("INTERPRETER=Cmd", 0, NULL, &shellTaskName, FALSE, FALSE,
fdScript, STD_OUT, STD_ERR); do
taskDelay (sysClkRateGet ());
while (taskNameToId (shellTaskName) != ERROR); close (fdScript);
The do/while loop is necessary for waiting for the shell script to terminate.
12.2.16 Executing Shell Commands Programmatically
There is no system( ) API as there is for a UNIX operating system. In order to be
able to execute shell commands from an application, the same technique as
described above can be used. It is not a file handle that is passed to the
shellGenericInit( ) API, but a pseudo-device slave file descriptor (see the API
reference for the ptyDrv library for information about pseudo-devices).
The application writes the commands it wants to be executed into the master side
of the pseudo-device. A pseudo-code representation of this might be as follows:
fdSlave = open ("system.S", O_RDWR);
fdMaster = open ("system.M", O_RDWR);
shellGenericInit ("INTERPRETER=Cmd", 0, NULL, &shellTaskName, FALSE, FALSE,
fdSlave, STD_OUT, STD_ERR);
taskDelay (sysClkRateGet ());
write (fdMaster, "pwd\n", 4);
close (fdMaster);
close (fdSlave);
12.2.17 Accessing Kernel Shell Data Programmatically
Shell data is available with the shellDataLib library. This allows the user to
associate data values with a shell session (uniquely per shell session), and to access
them at any time. This is useful to maintain default values, such as the memory
dump width, the disassemble length, and so on. These data values are not
accessible interactively from the shell, only programatically.
593
12
VxWorks
Kernel Programmer's Guide, 6.6
12.2.18 Using Kernel Shell Configuration Variables
Shell configuration variables are available using the shellConfigLib library. This
allows the user to define default configurations for commands or for the shell
itself. Such variables already exist for the shell (see the configuration variables
RTP_CREATE_STOP or LINE_EDIT_MODE). They behave similarly to environment
strings in a UNIX shell. These variables can be common to all shell sessions, or local
to a shell session. They can be modified and displayed interactively by the shell
user with the command set config or the shell routine shConfig( ).
12.2.19 Adding Custom Commands to the Command Interpreter
The kernel shell’s command interpreter consists of a line parser and of a set of
commands. It can be extended with the addition of custom commands written in
C. (The host shell’s command interpreter can likewise be extended, but with
commands written in Tcl.)
The syntax of a command statement is standard shell command-line syntax,
similar to that used with the UNIX sh shell or the Windows cmd shell. The syntax
is:
command [options] [arguments]
Blank characters (such as a space or tab) are valid word separators within a
statement. A blank character can be used within an argument string if it is escaped
(that is, prefixed with the back-slash character) or if the argument is quoted with
double quote characters. The semicolon character is used as a command separator,
used for entering multiple commands in a single input line. To be used as part of
an argument string, a semicolon must escaped or quoted.
The command parser splits the statement string into a command name string and
a string that consists of the options and the arguments. These options and
arguments are then passed to the command routine.
The command name may be a simple name (one word, such as reboot) or a
composite name (several words, such as task info). Composite names are useful
for creating classes of commands (commands for tasks, commands for processes,
and so on).
Options do not need to follow any strict format, but the standard UNIX option
format is recommended because it is handled automatically by the command
parser. If the options do not follow the standard UNIX format, the command
routine must parse the command strings to extract options and arguments. See
Defining a Command, p.596 for more information.
594
12 Target Tools
12.2 Kernel Shell
The standard UNIX option format string is:
-character [extra option argument]
The special double-dash option (--) is used in the same way as in UNIX. That is, all
elements that follow it are treated as an argument string, and not options. For
example, if the command test that accepts option -a, -b and -f (the latter with an
extra option argument), then the following command sets the three options, and
passes the arg string as an argument:
test -a -b -f arg
However, the next command only sets the -a option. Because they follow the
double-dash, the -b, -f and arg elements of the command are passed to the C
routine of the test command as strings:
test -a -- -b -f file arg
The command interpreter only handles strings. As a consequence, the arguments
of a command routine are strings as well. It is up to the command routine to
transform the strings into numerical values if necessary.
For information about symbol access syntax, see the material on the command
interpreter in the VxWorks Command-Line Tools User’s Guide.
Creating A New Command
The command interpreter is designed to allow customers to add their own
commands to the kernel shell.
Commands are stored in an internal database of the command interpreter. The
information describing a command are defined by a C structure that contains:
■
■
■
■
■
■
The command name.
A pointer to the C routine for the command.
The command options string (if needed).
A short description of the command.
A full description of the command.
A usage synopsis.
The command descriptions and the synopsis are used by the help command.
A command is registered with the command interpreter database along with a
topic name. The topic is used by the help command to display related commands.
For example, to display all commands related to the memory, you would use the
command help memory.
595
12
VxWorks
Kernel Programmer's Guide, 6.6
This section describes the conventions used for creating commands and provides
information about examples of commands that can serve as models. Also see the
shellInterpCmdLib API reference.
It may also be useful to review the other shell documentation in order to facilitate
the task of writing new commands. See the shellLib, shellDataLib, and
shellConfigLib API references, as well as the material on the command interpreter
in the VxWorks Command-Line Tools User’s Guide.
Defining a Command
A command is defined for the interpreter by a C structure composed of various
strings and a function pointer, with the following elements:
nameStruct = {"cmdFullname",
func,
"opt",
"shortDesc",
"fullDesc",
"%s [synopsis]"};
The string cmdFullname is the name of the command. It may be a composite name,
such as foo bar. In this case, foo is the top-level command name, and bar is a
sub-command of foo. The command name must be unique.
The func element is the name of the routine to call for that command name.
The string opt can be used in several different ways to define how option input is
passed to the command routine:
■
If opt is not NULL, it describes the possible options that the command accepts.
Each option is represented as a single character (note that the parser is case
sensitive). If an option takes an argument, a colon character (:) must be added
after the option character. For example, the following means that the
command accepts the -a, -v, and -f arg options:
avf:
■
If opt not NULL, and consists only of a semicolon, the option input is passed to
the command routine as a single string. It is up to the routine to extract options
and arguments.
■
If opt is NULL, the parser splits the input line into tokens and passes them as
traditional argc/argv parameters to the command routine.
Note that the command routine must be coded in a manner appropriate to how the
opt string is used in the command-definition structure (see Writing the Command
Routine, p.597).
596
12 Target Tools
12.2 Kernel Shell
The string shortDesc is a short description of the command. A sequence of string
conversion characters (%s) within that string is replaced by the command name
when the description is displayed. The string should not be ended by a new-line
character (\n).
The string fullDesc is the full description of the command. A sequence of string
conversion characters (%s) within that string is replaced by the command name
when the description is displayed. This description should contain the explanation
of the command options. The string should not be ended by newline (\n)
character.
The string synopsis is the synopsis of the command. A sequence of string
conversion characters (%s) within that string is replaced by the command name
when the synopsis is displayed. The string should not be ended by a newline (\n)
character.
The description and synopsis strings are used by the command interpreter’s help
command.
The rules for the C language command routines associated with the command
structures are described in Writing the Command Routine, p.597. See Sample Custom
Commands, p.599 for examples.
Writing the Command Routine
This section describes how to write the C routine for a command, including how
the routine should handle command options.
The command definition structure and the command routine must be coordinated,
most obviously with regard to the command routine name, but also with regard to
the opt element of the structure defining the command:
■
If the opt element is not equal to NULL, the declaration of the routine must
include the options array:
int func
(
SHELL_OPTION options[]
...
)
/* options array */
In this declaration, options is a pointer on the argument array of the
command.
■
However, if the opt element is NULL, the declaration of the routine must
include argc/argv elements, as follow:
597
12
VxWorks
Kernel Programmer's Guide, 6.6
int func
(
int
char **
...
)
argc,
arcv
/* number of argument */
/* pointer on the array of arguments */
In this declaration, argc is the number of arguments of the command, and argv
is an array that contains the argument strings.
In the first case the parser populates the options[] array.
In the second case it splits and passes the arguments as strings using argc/argv to
the routine.
When the opt element is used to define options, the order in which they are listed
is significant, because that is the order in which they populate the options[] array.
For example, if opt is avf:
■
Option a is described by the first cell of the options array: options[0].
■
Option v is described by the second cell of the options array: options[1].
■
Option f is described by the third cell of the options array: options[2].
■
The argument of option f is options[2].string.
Each cell of the options array passed by the parser to func is composed of a boolean
value (TRUE if the option is set, FALSE if not) and a pointer to a string (pointer to
an argument), if so defined. For example, if -a has been defined, the value of
options[0].isSet value is TRUE. Otherwise it is FALSE.
A boolean value indicates if it is the last cell of the array. If the option string opt is
only a colon, the argument string of the command is passed to func without any
processing, into the string field of the first element of the options array.
The return value of the command routine is an integer. By convention, a return
value of zero means that the command has run without error. Any other value
indicates an error value.
See Defining a Command, p.596 for information about the command-definition
structure. See Sample Custom Commands, p.599 for examples of command
structures and routines.
Registering a New Command
The shell commands have to be registered against the shell interpreter. This can be
done at anytime; the shell component does not need to be initialized before
commands can be registered.
598
12 Target Tools
12.2 Kernel Shell
A command is registered in a topic section. The topic name and the topic
description must also be registered in the command interpreter database. The
routine used to do so is shellCmdTopicAdd( ). This routine accepts two
parameters: a unique topic name and a topic description. The topic name and
description are displayed by the help command.
Two routines are used to register commands: shellCmdAdd( ) adds a single
command, and shellCmdArrayAdd( ) adds an array of commands.
See Sample Custom Commands, p.599 for information about code that illustrates
command registration.
Sample Custom Commands
For an example of custom command code, see
installDir/vxworks-6.x/target/src/demo/shell/tutorialShellCmd.c. In addition to
illustrating how to implement custom commands, it shows how to create
command aliases (also see 12.2.9 Defining Kernel Shell Command Aliases, p.587).
The code can be used with Wind River Workbench as a downloadable kernel
module project, or included in a kernel project. For information about using
Workbench, see the Wind River Workbench User’s Guide.
It can also be built from the command line with the command make CPU=cpuType.
For example:
make CPU=PENTIUM2
This resulting module can then be loaded into the kernel, using Workbench
(including the host shell) or kernel shell.
The tutorialShellCmdInit( ) routine must be called to register the commands
before they can be executed, regardless of how the code is implemented.
12.2.20 Creating a Custom Interpreter
The kernel shell is designed to allow customers to add their own interpreter. Two
interpreters are provided by Wind River: the C interpreter and the command
interpreter. An interpreter receives the user input line from the shell, validates the
input against its syntax and grammar, and performs the action specified by the
input line (such as redirection, calling a VxWorks function, reading or writing
memory, or any other function that an interpreter might perform).
599
12
VxWorks
Kernel Programmer's Guide, 6.6
Within the shell, an interpreter is defined by a set of static information and an
interpreter context.
The static information about the interpreter is:
■
an interpreter context initialization routine
■
a parsing routine
■
an evaluation routine
■
a completion routine
■
a restart routine
■
an interpreter context finalization routine (to release any resources)
■
an interpreter name (which must be unique)
■
a default interpreter prompt
This information is set when the interpreter is registered with the shell using the
shellInterpRegister( ) routine.
The interpreter context is created the first time the interpreter is used. It is unique
to each interpreter and to each shell session. When it has been created, the shell
calls the interpreter context initialization routine so that the interpreter can add
any private data that it requires to that context. The pInterpParam field of the
interpreter’s context structure SHELL_INTERP_CTX can be used for that purpose.
Each time a line is entered from terminal or read from a script, the shell calls the
parsing routine of the current interpreter so that it can evaluate the line.
The arguments to the parsing routine are the interpreter context, a pointer to the
input line string, and a boolean value indicating whether or not the shell session is
interactive.
The job of the parsing routine is to split the line into meaningful words, according
to the interpreter’s syntax, and to perform the appropriate actions.
The evaluation routine is called by the shell whenever an evaluation for that
interpreter is required, using the shellInterpEvaluate( ) routine. Usually, the
evaluation routine and the parsing routine can share most of their code. The
evaluation routine returns an evaluation value.
The completion routine is called by the shell whenever a completion key is hit
(completion keys are defined by the line editing mode of the shell; see the ledLib
API reference). It is up to the interpreter to perform word completion according to
its syntax, and to the position of the cursor in the line.
600
12 Target Tools
12.2 Kernel Shell
It is up to the interpreter to release its resource whenever a shell session terminates
or is restarted. The finalize and restart routines specified at registration time are
called by the shell for this purpose.
The restart routine is called by the shell whenever the shell task is restarted, either
because it has taken an exception or it was restarted by the user. This routine is
useful for releasing any resources reserved by the interpreter (memory blocks,
semaphores, and so on).
The context-finalize routine is called when the shell session is terminated. It is used
to free any resources allocated by the interpreter.
Note that the stream redirection characters must be handled by the interpreter
itself (for example, the command interpreter’s <, > and >> redirection characters).
The interpreter name is a string that uniquely identifies the interpreter among any
others registered with the shell.
The default interpreter prompt is the string that identifies the interpreter visually.
For example, the C interpreter prompt is the string:
->
The interpreter prompt may contain format conversion characters that are
dynamically replaced when printed by another string. For example %/ is replaced
by the current shell session path, %n is replaced by the user name, and so on (see
the shellPromptLib API reference for more information). Moreover, it is possible
to add new format strings with shellPromptFmtStrAdd( ). For example the
command-related process adds the format string %c to display the name of the
current working memory context.
The current interpreter is defined by the shell configuration variable named
INTERPRETER (the C macro SHELL_CFG_INTERP is defined for it in shellLib.h).
A shell user can switch to is own interpreter by setting the value of the
configuration variable INTERPRETER to its interpreter name. This can be done
either interactively or programmatically. For illustrative purposes, the following
commands change the interpreter from C to command and back again at the shell
command line:
-> shConfig "INTERPRETER=Cmd"
[vxWorks]# set config INTERPRETER=C
->
It is also possible to created custom commands to allow for switching to and from
a custom interpreter, similar to those used to switch between the command and C
interpreters (C is used with the command interpreter, and cmd with the C
interpreter to switch between the two).
601
12
VxWorks
Kernel Programmer's Guide, 6.6
The following code fragment sets the command interpreter for the current session:
shellConfigValueSet (CURRENT_SHELL_SESSION, SHELL_CFG_INTERP, "Cmd");
For more information, see the set config command and the
shellConfigValueSet( ) routine in the shellConfigLib API reference.
Sample Custom Interpreter
For an example of a interpreter code, see:
installDir/vxworks-6.x/target/src/demo/shell/shellInterpDemo.c. The sample
interpreter is called DEMO, and it has only six commands:
■
■
■
■
■
■
task to create a task
list to list the tasks
kill to destroy a task
dump to dump the memory contents
sym to access symbol
C to switch back to the C interpreter
The DEMO code illustrates how to create a small interpreter with a few
commands, and how to make use of some of the routines exported by the shell
(from the shellDataLib and shellConfigLib libraries).
To use this interpreter in the shell, you can build and download it as a
downloadable kernel module. To register the commands, the initialization
function shellInterpDemoInit( ) has to be called first. You can load it, register it,
and set it for the current session as follows:
-> ld < shellInterpDemo.o
-> shellInterpRegister (shellInterpDemoInit)
-> shConfig "INTERPRETER=DEMO"
DEMO #
If you choose to link the module with VxWorks instead of downloading it, you
have to make the shellInterpDemoInit( ) call in the user startup code (see
2.6.8 Linking Kernel Application Object Modules with VxWorks, p.64 and
2.6.10 Configuring VxWorks to Run Applications Automatically, p.66).
The code can be used with Wind River Workbench as a downloadable kernel
module project, or included in a kernel project. For information about using
Workbench, see the Wind River Workbench User’s Guide.
602
12 Target Tools
12.3 Kernel Object-Module Loader
12.3 Kernel Object-Module Loader
The target-resident, VxWorks kernel object-module loader lets you add object
modules to the kernel at run-time. This operation, called loading, or downloading,
allows you to install kernel-space applications or to extend the operating system
itself. (For the sake of brevity, the kernel object-module loader is also referred to
simply as the kernel loader, or loader, in this section).
The downloaded code can be a set of routines, meant to be used by some other
code (the equivalent of a library in other operating systems), or it can be an
application, meant to be executed by a task or a set of tasks. The units of code that
can be downloaded are referred to as object modules.
The ability to load individual object modules brings significant flexibility to the
development process, in several different ways. The primary use of this facility
during development is to unload, recompile, and reload object modules under
development. The alternative is to link the developed code into the VxWorks
image, to rebuild this image, and to reboot the target, every time the development
code must be recompiled.
The kernel loader also enables you to dynamically extend the operating system,
since once code is loaded, there is no distinction between that code and the code
that was compiled into the image that booted.
Finally, you can configure the kernel loader to optionally handle memory
allocation, on a per-load basis, for modules that are downloaded. This allows
flexible use of the target's memory. The loader can either dynamically allocate
memory for downloaded code, and free that memory when the module is
unloaded; or, the caller can specify the addresses of memory that has already been
allocated. This allows the user more control over the layout of code in memory. For
more information, see 12.3.5 Specifying Memory Locations for Loading Objects, p.610.
The functionality of the kernel loader is provided by two components: the loader
proper, which installs the contents of object modules in the target system's
memory; and the unloader, which uninstalls object modules. In addition, the
loader relies on information provided by the system symbol table.
!
CAUTION: Do not unload an object module while its tasks are running. Doing so
may result in unpredictable behavior.
603
12
VxWorks
Kernel Programmer's Guide, 6.6
NOTE: The target-resident kernel object-module loader is sometimes confused
with the boot loader, which is used to load the kernel image into memory.
Although these two tools perform similar functions, and share some support code,
they are separate entities. The boot loader loads only complete system images, and
does not perform relocations. See 3. Boot Loader.
12.3.1 Configuring VxWorks with the Kernel Object-Module Loader
By default, the kernel object-module loader is not included in VxWorks. To use the
loader, you must configure VxWorks with the INCLUDE_LOADER component.
Adding the INCLUDE_LOADER component automatically includes several other
components that together provide complete loader functionality. These
components are:
INCLUDE_UNLOADER
Provides facilities for unloading object modules.
INCLUDE_MODULE_MANAGER
Provides facilities for managing loaded modules and obtaining information
about them. For more information, see the VxWorks API reference for
moduleLib.
INCLUDE_SYM_TBL
Provides facilities for storing and retrieving symbols. For more information,
see 12.4 Kernel Symbol Tables, p.617 and the VxWorks API reference for
symLib.
INCLUDE_SYM_TBL_INIT
Specifies a method for initializing the system symbol table.
!
CAUTION: If you want to use the target-resident symbol tables and kernel
object-module loader in addition to the host tools, you must configure VxWorks
with the INCLUDE_WDB_MDL_SYM_SYNC component to provide host-target
symbol table and module synchronization. This component is included by default
when both the kernel loader and WDB agent are included in VxWorks. For more
information, see 12.4.4 Using the VxWorks System Symbol Table, p.622.
The kernel loader and unloader are discussed further in subsequent sections, and
in the VxWorks API references for loadLib and unldLib.
604
12 Target Tools
12.3 Kernel Object-Module Loader
12.3.2 Kernel Object-Module Loader API
The API routines, shell C interpreter commands, and shell command interpreter
commands available for loading and unloading kernel modules are described in
Table 12-5 and Table 12-6.
Note that the kernel loader routines can be called directly from the C interpreter or
from code. The shell commands, however, should only be called from the shell and
not from within programs.3 In general, shell commands handle auxiliary
operations, such as opening and closing a file; they also print their results and any
error messages to the console.
Table 12-5
Routines for Loading and Unloading Object Modules
Routine
Description
loadModule( )
Loads an object module.
loadModuleAt( )
Loads an object module into a specific memory location.
unldByModuleId( )
Unloads an object module by specifying a module ID.
unldByNameAndPath( ) Unloads an object module by specifying name and path.
unldByGroup( )
Table 12-6
Unloads an object module by specifying its group.
Shell C Interpreter Commands for Object Modules
Command
Description
ld( )
Loads an object module into kernel memory.
reld( )
Unloads and reloads an object module (specified by filename or
module ID).
unld( )
Unloads an object module (specified by filename or module ID) from
kernel memory.
The use of some of these routines and commands is discussed in the following
sections.
3. In future releases, calling shell commands programmatically may not be supported.
605
12
VxWorks
Kernel Programmer's Guide, 6.6
For detailed information, see the loadLib, unldLib, and usrLib API references, the
shell command reference, as well as 12.3.3 Summary List of Kernel Object-Module
Loader Options, p.606.
12.3.3 Summary List of Kernel Object-Module Loader Options
The kernel loader's behavior can be controlled using load flags passed to loadLib
and unldLib routines. Many of these flags can be combined (using a logical OR
operation); some are mutually exclusive. The tables in this section group these
options by category.
Table 12-7
Kernel Loader and Unloader Options for C++
Hex
Value
Option
Description
LOAD_CPLUS_XTOR_AUTO
0x1000 Call C++ constructors on loading.
LOAD_CPLUS_XTOR_MANUAL
0x2000 Do not call C++ constructors on loading.
UNLD_CPLUS_XTOR_AUTO
0x20 Call C++ destructors on unloading.
UNLD_CPLUS_XTOR_MANUAL
0x40 Do not call C++ destructors on unloading. If this
option is used, be sure that the destructor routines
are not used to release resources back to the system
(such as memory or semaphores); or the caller may
first run any static destructors by calling
cplusDtors( ).
Table 12-8
Kernel Loader Options for Symbol Registration
Hex
Value
Option
LOAD_NO_SYMBOLS
606
0x2
Description
No symbols from the module are registered in the system's
symbol table. Consequently, linkage against the code module
is not possible. This option is useful for deployed systems,
when the module is not supposed to be used in subsequent
link operations.
12 Target Tools
12.3 Kernel Object-Module Loader
Table 12-8
Kernel Loader Options for Symbol Registration (cont’d)
Hex
Value
Option
Description
LOAD_LOCAL_SYMBOLS
0x4
Only local (private) symbols from the module are registered in
the system's symbol table. No linkage is possible against this
code module’s public symbols. This option is not very useful
by it self, but is one of the base options for
LOAD_ALL_SYMBOLS.
LOAD_GLOBAL_SYMBOLS
0x8
Only global (public) symbols from the module are registered
in the system's symbol table. No linkage is possible against
this code module's private symbols. This is the kernel loader's
default when the loadFlags parameter is left as NULL.
LOAD_ALL_SYMBOLS
0xC
Local and global symbols from the module are registered in
the system's symbol table. This option is useful for debugging.
12
Table 12-9
Kernel Loader Options for Code Module Visibility
Hex
Value
Option
HIDDEN_MODULE
Table 12-10
Description
0x10 The code module is not visible from the moduleShow( )
routine or the host tools. This is useful on deployed systems
when an automatically loaded module should not be
detectable by the user. It only affects user visibility, and does
not affect linking with other modules.
Kernel Unloader Options for Breakpoints and Hooks
Option
UNLD_KEEP_BREAKPOINTS
Hex
Value
0x1
Description
The breakpoints are left in place when the code
module is unloaded. This is useful for debugging, as
all breakpoints are otherwise removed from the
system when a module is unloaded.
607
VxWorks
Kernel Programmer's Guide, 6.6
Table 12-10
Kernel Unloader Options for Breakpoints and Hooks (cont’d)
Hex
Value
Option
0x2
UNLD_FORCE
Table 12-11
Description
By default, the kernel unloader does not remove the
text sections when they are used by hooks in the
system. This option forces the unloader to remove
the sections anyway, at the risk of unpredictable
results.
Kernel Loader Options for Resolving Weak Symbols
Option
Hex Value
Description
LOAD_WEAK_MATCH_ALL
0x20000 A weak symbol is ignored if there is an existing
global symbol with the same name. If none is found,
the weak symbol is registered as a global. This is the
default behavior.
LOAD_WEAK_MATCH_NONE
0x10000 The loader always registers the weak symbol as a
global, regardless of any existing definition. This
option provides the behavior of the VxWorks 5.x
loader.
608
12 Target Tools
12.3 Kernel Object-Module Loader
Table 12-12
Kernel Loader Options for Resolving Common Symbols
Option
Hex Value
Description
LOAD_COMMON_MATCH_NONE
0x100
This option prevents any matching with
already-existing symbols. Common symbols are
added to the symbol table unless
LOAD_NO_SYMBOLS is set. This is the default
option. (Note that this option is only in effect for the
specific load operation in which it is used; that is, it
has no effect on subsequent load operations that do
not use it.)
LOAD_COMMON_MATCH_USER
0x200 Seeks a matching symbol in the system symbol
table, but considers only symbols in user modules,
not symbols that were in the original booted image.
If no matching symbol exists, this option behaves
like LOAD_COMMON_MATCH_NONE.
LOAD_COMMON_MATCH_ALL
0x400 Seeks a matching symbol in the system symbol
table, considering all symbols. If no matching
symbol exists, this option behaves like
LOAD_COMMON_MATCH_NONE.
If several matching symbols exist for the options LOAD_COMMON_MATCH_USER
and LOAD_COMMON_MATCH_ALL, the symbol most recently added to the
symbol table is used.
Table 12-13
Fully-Linked Module Load Support
Hex
Value
Option
LOAD_FULLY_LINKED
Description
0x20 Provides for loading fully linked modules (that is, modules
without any unresolved symbols or relocations).
Note that symbol tables are not required when VxWorks is configured with
support for loading fully-linked object modules (the option is listed in
Table 12-13). For more information, see loadModuleAt( ) in the VxWorks Kernel
API Reference.
609
12
VxWorks
Kernel Programmer's Guide, 6.6
12.3.4 Loading C++ Modules into the Kernel
For information about loading C++ modules from the shell, see 13.4 Using C++ in
Signal Handlers and ISRs, p.649. Also see 12.3.3 Summary List of Kernel Object-Module
Loader Options, p.606 for C++ kernel loader and unloader options.
12.3.5 Specifying Memory Locations for Loading Objects
By default, the kernel object-module loader allocates the memory necessary to
hold a code module. It is also possible to specify where in memory any or all of the
text, data, and bss segments of an object module should be installed using the
loadModuleAt( ) command. If an address is specified for a segment, then the caller
must allocate sufficient space for the segment at that address before calling the
load routine. If no addresses are specified, the kernel loader allocates one
contiguous area of memory for all three of the segments.
For any segment that does not have an address specified, the loader allocates the
memory (using memPartAlloc( ) or, for aligned memory, using memalign( )). The
base address can also be set to the value LOAD_NO_ADDRESS, in which case the
loader replaces the LOAD_NO_ADDRESS value with the actual base address of the
segment once the segment is installed in memory.
The basic unit of information in a relocatable ELF object file is a section. In order to
minimize memory fragmentation, the loader gathers sections so that they form the
logical equivalent of an ELF segment. For simplicity, these groups of sections are
also referred to as segments. For more information, see ELF Object Module Format,
p.612).
The kernel loader creates three segments: text, data, and bss. When gathering
sections together to form segments, the sections are placed into the segments in the
same order in which they occur in the ELF file. It is sometimes necessary to add
extra space between sections to satisfy the alignment requirements of all of the
sections. When allocating space for one or more segments, care must be taken to
ensure that there is enough space to permit all of the sections to be aligned
properly. (The alignment requirement of a section is given as part of the section
description in the ELF format. The binary utilities readelfarch and objdumparch
can be used to obtain the alignment information.)
In addition, the amount of padding required between sections depends on the
alignment of the base address. To ensure that there will be enough space without
knowing the base address in advance, allocate the block of memory so that it is
aligned to the maximum alignment requirement of any section in the segment. So,
610
12 Target Tools
12.3 Kernel Object-Module Loader
for instance, if the data segment contains sections requiring 128 and 264 byte
alignment, in that order, allocate memory aligned on 264 bytes.
The kernel unloader can remove the segments from wherever they were installed,
so no special instructions are required to unload modules that were initially loaded
at specific addresses. However, if the base address was specified in the call to the
loader, then, as part of the unload, unloader does not free the memory area used
to hold the segment. This allocation was performed by the caller, and the
de-allocation must be as well.
12.3.6 Guidelines and Caveats for Kernel Object-Module Loader Use
The following sections describe the criteria used to load modules and issues with
loading that may need to be taken into account.
Relocatable Object Files
12
Relocatable object files are used for modules that can be dynamically loaded into
the VxWorks kernel and run. In contrast to an executable file, which is fully linked
and ready to run at a specified address, a relocatable file is an object file for which
text and data sections are in a transitory form, meaning that some addresses are
not yet known. Relocatable object modules are generated by the compiler with .o
extension (similar to the ones produced as an intermediate step between the
application source files—.c, .s, .cpp— and the corresponding executable files that
run in VxWorks processes).
Relocatable files are used for downloadable modules because the layout of the
VxWorks image and downloaded code in memory are not available to a compiler
running on a host machine. Therefore, the code handled by the target-resident
kernel loader must be in relocatable form, rather than an executable. The loader
itself performs some of the same tasks as a traditional linker in that it prepares the
code and data of an object module for the execution environment. This includes the
linkage of the module's code and data to other code and data.
Once installed in the system's memory, the entity composed of the object module's
code, data, and symbols is called a code module. For information about installed
code modules, see the VxWorks API reference for moduleLib.
611
VxWorks
Kernel Programmer's Guide, 6.6
ELF Object Module Format
An relocatable ELF object file is essentially composed of two categories of
elements: the headers and the sections. The headers describe the sections, and the
sections contain the actual text and data to be installed.
An executable ELF file is a collection of segments, which are aggregations of
sections. The kernel object-module loader performs an aggregation step on the
relocatable object files that is similar to the process carried out by toolchains when
producing an executable ELF file. The resulting image consists of one text segment,
one data segment, and one bss segment. (A general ELF executable file may have
more than one segment of each type, but the loader uses the simpler model of at
most one segment of each type.) The loader installs the following categories of
sections in the system's memory:
■
text sections that hold the application's instructions
■
data sections that hold the application's initialized data
■
bss sections that hold the application's un-initialized data
■
read-only data sections that hold the application's constant data
Read-only data sections are placed in the text segment by the loader.
Linking and Reference Resolution
The kernel object-module loader performs some of the same tasks as a traditional
linker in that it prepares the code and data of an object module for the execution
environment. This includes the linkage of the module's code and data to other code
and data.
The loader is unlike a traditional linker in that it does this work directly in the
target system's memory, and not in producing an output file.
In addition, the loader uses routines and variables that already exist in the
VxWorks system, rather than library files, to relocate the object module that it
loads. The system symbol table (see 12.4.4 Using the VxWorks System Symbol Table,
p.622) is used to store the names and addresses of functions and variables already
installed in the system.This has the side effect that once symbols are installed in the
system symbol table, they are available for future linking by any module that is
loaded. Moreover, when attempting to resolve undefined symbols in a module,
the loader uses all global symbols compiled into the target image, as well as all
global symbols of previously loaded modules. As part of the normal load process,
all of the global symbols provided by a module are registered in the system symbol
612
12 Target Tools
12.3 Kernel Object-Module Loader
table. You can override this behavior by using the LOAD_NO_SYMBOLS load flag
(see Table 12-8).
The system symbol table allows name clashes to occur. For example, suppose a
symbol named func exists in the system. A second symbol named func is added to
the system symbol table as part of a load. From this point on, all links to func are
to the most recently loaded symbol. See also, 12.4.1 Configuring VxWorks with
Symbol Tables, p.618.
Load Sequence Requirements and Caveats
The kernel object-module loader loads code modules in a sequential manner. That
is, a separate load is required for each separate code module. The user must,
therefore, consider dependencies between modules and the order in which they
must be loaded to link properly.
Suppose a user has two code modules named A_module and B_module, and
A_module references symbols that are contained in B_module. The user may
either use the host-resident linker to combine A_module and B_module into a
single module, or they should load B_module first, and then load A_module.
When code modules are loaded, they are irreversibly linked to the existing
environment; meaning that, once a link from a module to an external symbol is
created, that link cannot be changed without unloading and reloading the module.
Therefore dependencies between modules must be taken into account when
modules are loaded to ensure that references can be resolved for each new module,
using either code compiled into the VxWorks image or modules that have already
been loaded into the system.
Failure to do so results in incompletely resolved code, which retains references to
undefined symbols at the end of the load process. For diagnostic purposes, the
loader prints a list of missing symbols to the console. This code should not be
executed, since the behavior when attempting to execute an improperly relocated
instruction is not predictable.
Normally, if a load fails, the partially installed code is removed. However, if the
only failure is that some symbols are unresolved, the code is not automatically
unloaded (but the API returns NULL to indicate failure programmatically). This
allows the user to examine the result of the failed load, and even to execute
portions of the code that are known to be completely resolved. Therefore, code
modules that have unresolved symbols must be removed by a separate unload
command (unld( ) with the C interpreter, or module unload with the command
interpreter).
613
12
VxWorks
Kernel Programmer's Guide, 6.6
Note that the sequential nature of the loader means that unloading a code module
which has been used to resolve another code module may leave references to code
or data which are no longer available. Execution of code holding such dangling
references may have unexpected results.
See Statically Linking Kernel Application Modules, p.63.
Resolving Common Symbols
Common symbols provide a challenge for the kernel object-module loader that is
not confronted by a traditional linker. Consider the following example:
#include
int willBeCommon;
void main (void) {}
{
...
}
The symbol willBeCommon is uninitialized, so it is technically an undefined
symbol. Many compilers will generate a common symbol in this case.
ANSI C allows multiple object modules to define uninitialized global symbols of
the same name. The linker is expected to consistently resolve the various modules
references to these symbols by linking them against a unique instance of the
symbol. If the different references specify different sizes, the linker should define
a single symbol with the size of the largest one and link all references against it.
This is not a difficult task when all of the modules are linked in the same operation,
such as when executing the host-resident linker, ldarch.
However, when VxWorks modules are loaded sequentially, the loader can only
resolve the references of the module that it is currently loading with the those of
the modules that it has already loaded, regardless of what the final, full set of
modules may be. The loadLib API functions provide three options for controlling
how common symbols are linked:
■
The default behavior is to treat common symbols as if there were no previous
matching reference (LOAD_COMMON_MATCH_NONE). The result is that
every loaded module has its own copy of the symbol. For example, for three
loads using this option with the same common symbol, three new global
symbols are created.
■
Common symbols are identified with any matching symbol in the symbol
table (LOAD_COMMON_MATCH_ALL).
614
12 Target Tools
12.3 Kernel Object-Module Loader
■
Common symbols are identified with any matching symbol that was not in the
original boot image (LOAD_COMMON_MATCH_USER).
Note that these options only control the loader’s behavior with regard to the
operation in which they are used—they only affect what happens with the symbols
of the module being loaded. For example, consider the case in which module A has
common symbols, and module B has undefined symbols that are resolved by
module A. If module A is loaded with the LOAD_COMMON_MATCH_NONE
option, this does not prevent module B from being linked against A’s symbols
when B is loaded next. That is, the load flag used with module A does not prevent
the loader from resolving undefined references in module B against module A.
The option to specify matching of common symbols may be set in each call using
the loadLib API. Extreme care should be used when mixing the different possible
common matching behaviors for the loader. It is much safer to pick a single
matching behavior and to use it for all loads. For detailed descriptions of the
matching behavior under each option, see Table 12-12.
NOTE: Note that the shell load command, ld, has a different mechanism for
controlling how common symbols are handled and different default behavior. For
details, see the reference entry for usrLib.
Resolving Weak Symbols
Some programming languages (such as C++) use the weak binding class in addition
to the global and local classes. The ELF specification mandates that a weak symbol
is ignored if there is an existing global symbol with the same name. This is the
default behavior provided by the VxWorks 6.x object module loader, specified
with the LOAD_WEAK_MATCH_ALL loader option.
Note, however, that the default behavior for VxWorks 5.x was to always register
weak symbols as globals, regardless of any existing definitions. To replicate this
behavior, use the LOAD_WEAK_MATCH_NONE loader option.
Stripping Symbols From Modules
Symbols can be removed from object files by stripping them, which is commonly
done with the GNU strip utility. The main purpose of stripping object files is to
reduce their size.
615
12
VxWorks
Kernel Programmer's Guide, 6.6
The kernel object-module loader, however, requires relocation, section, and
symbol information in order to load a module. By default the strip utility removes
the information required by the loader, and it should therefore be used with the
--strip-unneeded option, which removes all symbols that are not needed for the
relocation process.
Note that the --strip-unneeded option removes debug information.
Function Calls, Relative Branches, and Load Failures
For some architectures, function calls are performed using relative branches by
default. This causes problems if the routine that is called resides further in memory
than a relative branch instruction can access (which may occur if the board has a
large amount of memory).
In this case, a module load fails; the kernel module loader prints an error message
about relocation overflow and sets the
S_loadElfLib_RELOCATION_OFFSET_TOO_LARGE errno (kernel shell).
To deal with this problem, compilers (both GNU and Wind River) have options to
prevent the use of relative branches for function calls. See the VxWorks Architecture
Supplement for more information.
Kernel Object Modules With SDA
The loader cannot perform small data area (SDA) relocation. If a kernel module is
built with SDA, the loader will not load it, but generates the following error:
■
S_loadLib_SDA_NOT_SUPPORTED
The Wind River Compiler (diab) assembler flag -Xwarn-use-greg can be used to
generate the following warning if code accesses the SDA reserved registers:
Xwarn-use-greg=0x2004
The objdump and readelf tools can be used to see if there are any SDA relocations
in a module. The relocation types pertaining to SDA are described in the ELF
architecture ABI supplement.
The SDA_DISABLE makefile variable can be used to disable SDA, as follows:
SDA_DISABLE=TRUE
616
12 Target Tools
12.4 Kernel Symbol Tables
12.4 Kernel Symbol Tables
A symbol table is a data structure that stores information that describes the
routines, variables, and constants in all modules, and any variables created from
the shell. There is a symbol table library, which can be used to manipulate the two
different types of kernel symbol tables: a user-created symbol table and a system
symbol table, which is the most commonly used. Note that both types of symbol
tables used in the kernel are entirely independent of the symbol tables used by
applications running in user-space processes (RTPs).
A system symbol table is required for the kernel object-module loader. The only
exception being when fully-linked object modules are loaded with the
LOAD_FULLY_LINKED option. For more information see Table 12-13 and
loadModuleAt( ) in the VxWorks Kernel API Reference.
Symbol Entries
12
Each symbol in the table comprises these items:
name
The name is a character string derived from the name in the source code.
value
The value is usually the address of the element that the symbol refers to: either
the address of a routine, or the address of a variable (that is, the address of the
contents of the variable). The value is represented by a pointer.
group
The group number of the module that the symbol comes from.
symRef
The symRef is usually the module ID of the module that the symbol comes
from.
type
The type is provides additional information about the symbol. For symbols in
the system symbol table, it is one of the types defined in
installDir/vxworks-6.x/target/h/symbol.h. For example, SYM_UNDF,
SYM_TEXT, and so on. For user symbol tables, this field can be user-defined.
617
VxWorks
Kernel Programmer's Guide, 6.6
Symbol Updates
The symbol table is updated whenever modules are loaded into, or unloaded from,
the target. You can control the precise information stored in the symbol table by
using the kernel object-module loader options listed in Table 12-8.
Searching the Symbol Library
You can easily search all symbol tables for specific symbols. To search from the
shell with the C interpreter, use lkup( ). You can also use symShow( ) for general
symbol information. For details, see the API references for these commands.
To search programmatically, use the symbol library API's, which can be used to
search the symbol table by address, by name, and by type, and a function that may
be used to apply a user-supplied function to every symbol in the symbol table. For
details, see the symLib reference entry.
12.4.1 Configuring VxWorks with Symbol Tables
VxWorks can be configured with support for user symbols tables or with support
for both user symbol tables and a system symbol table.
For information about user symbol tables, see 12.4.6 Creating and Using User Symbol
Tables, p.623. For information about the system symbol table, see 12.4.4 Using the
VxWorks System Symbol Table, p.622.
Configuration for User Symbol Tables
Configuring VxWorks with the INCLUDE_SYM_TBL component provides the basic
symbol table library, symLib, and support for creating user symbol tables.
A user symbol table is created at run-time with the symTblCreate( ) routine, with
parameters for the width of the symbol table's hash table, the name clash policy
and the memory partition to use. For more information, see the VxWorks API
reference for symTblCreate( ).
618
12 Target Tools
12.4 Kernel Symbol Tables
Configuration for a System Symbol Table
To include information about the symbols present in the kernel—and therefore to
enable the shell, kernel object-module loader, and debugging facilities to function
properly—a system symbol table must be created and initialized.
To create a system symbol table, VxWorks must be configured with the
INCLUDE_SYM_TBL_INIT component and the INCLUDE_SYM_TBL component
(which also provides support for user symbol tables).
The INCLUDE_SYM_TBL_INIT component includes the configuration parameter
SYM_TBL_HASH_SIZE_LOG2 which allows you to modify the symbol table width.
This parameter defines the width of the symbol table's hash table. It takes a
positive value that is interpreted as a power of two. The default value for
SYM_TBL_HASH_SIZE_LOG2 is 8; and the default width of the symbol table is
therefore 256. Using smaller values requires less memory, but degrades lookup
performance, so the search takes longer on the average.
The system symbol table, sysSymTbl, is configured to allow name clashes. In the
event that they occur, the most recently added symbol is the one that is returned
when searching the symbol table by name.
To initialize the system symbol table (which adds VxWorks kernel symbols),
VxWorks must include a component for either a symbol table that is part of the
system image, or a component for a symbol table that is downloaded separately
from the host system:
INCLUDE_STANDALONE_SYM_TBL
Creates a built-in system symbol table, in which both the system symbol table
and the VxWorks image are contained in the same module in which the system
symbol table is contained, in the VxWorks image. This type of symbol table is
described in 12.4.2 Creating a Built-In System Symbol Table, p.620.
INCLUDE_NET_SYM_TBL
Creates an separate system symbol table as a .sym file that is downloaded to
the VxWorks system. This type of symbol table is described in 12.4.3 Creating
a Loadable System Symbol Table, p.621.
When the system symbol table is first created at system initialization time, it
contains no symbols. Symbols must be added to the table at run-time. Each of these
components handles the process of adding symbols differently.
619
12
VxWorks
Kernel Programmer's Guide, 6.6
12.4.2 Creating a Built-In System Symbol Table
A built-in system symbol table copies information into wrapper code, which is
then compiled and linked into the kernel when the system is built.
Although using a built-in symbol table can produce a larger VxWorks image file
than might otherwise be the case, it has several advantages, particularly for
production systems:
■
It requires less memory than using a loadable symbol table—as long as you are
not otherwise using the kernel object-module loader and associated
components that are required for a loadable symbol table.
■
It does not require that the target have access to a host (unlike the
downloadable symbol table).
■
It is faster to load the single image file than loading separate files for the
VxWorks image and the loadable symbol table .sym file because some remote
operations4 on a file take longer than the data transfer to memory.
■
It is useful in deployed ROM-based systems that have no network
connectivity, but require the shell as user interface.
Generating the Symbol Information
A built-in system symbol table relies on the makeSymTbl utility to obtain the
symbol information. This utility uses the gnu utility nmarch to generate
information about the symbols contained in the image. Then it processes this
information into the file symTbl.c that contains an array, standTbl, of type
SYMBOL described in Symbol Entries, p.617. Each entry in the array has the symbol
name and type fields set. The address (value) field is not filled in by makeSymTbl.
Compiling and Linking the Symbol File
The symTbl.c file is treated as a normal .c file, and is compiled and linked with the
rest of the VxWorks image. As part of the normal linking process, the toolchain
linker fills in the correct address for each global symbol in the array. When the
build completes, the symbol information is available in the image as a global array
of VxWorks symbols. After the kernel image is loaded into target memory at
4. That use open( ), seek( ), read( ), and close( ).
620
12 Target Tools
12.4 Kernel Symbol Tables
system initialization, the information from the global SYMBOL array is used to
construct the system symbol table.
The definition of the standTbl array can be found in the following files after the
VxWorks image is built:
■
installDir/vxworks-6.x/target/config/bspName/symTbl.c for images built
directly from a BSP directory.
■
installDir/vxworks-6.x/target/proj/projDir/buildDir/symTbl.c for images using
the project facility.
12.4.3 Creating a Loadable System Symbol Table
A loadable symbol table is built into a separate object module file (vxWorks.sym
file). This file is downloaded to the system separately from the system image, at
which time the information is copied into the symbol table.
12
Creating the .sym File
The loadable system symbol table uses an ELF file named vxWorks.sym file, rather
than the symTbl.c file. The file is created by using the objcopy utility to strip all
sections, except the symbol information, from the final VxWorks image.
Loading the .sym File
During boot and initialization, the vxWorks.sym file is downloaded using the
kernel object-module loader, which directly calls loadModuleAt( ). To download
the vxWorks.sym file, the loader uses the current default device, which is
described in 7.3.1 Filenames and the Default Device, p.363.
To download the VxWorks image, the loader also uses the default device, as is
current at the time of that download. Therefore, the default device used to
download the vxWorks.sym file may, or may not, be the same device. This is
because the default device can be set, or reset, by other initialization code that runs.
This modification can happen after the VxWorks image is downloaded, but before
the symbol table is downloaded.
Nevertheless, in standard VxWorks configurations, that do not include
customized system initialization code, the default device at the time of the
621
VxWorks
Kernel Programmer's Guide, 6.6
download of the vxWorks.sym, is usually set to one of the network devices, and
using either rsh or ftp as the protocol.
12.4.4 Using the VxWorks System Symbol Table
Once it is initialized, the VxWorks system symbol table includes a complete list of
the names and addresses of all global symbols in the compiled image that is
booted. This information is needed on the target to enable the full functionality of
the target tools libraries.
The target tools maintain the system symbol table with up-to-date name and
address information for all of the code statically compiled into the system or
dynamically downloaded. (The LOAD_NO_SYMBOLS option can be used to hide
loaded modules, so that their symbols do not appear in the system symbol table;
see Table 12-9).
Symbols are dynamically added to, and removed from, the system symbol table
when:
■
modules are loaded and unloaded
■
variables are dynamically created from the shell
■
the wdb agent synchronizes symbol information with the host (see
12.4.5 Synchronizing Host and Kernel Modules List and Symbol Table, p.623)
The exact dependencies between the system symbol table and the other target tools
are as follows:
■
Kernel Object-Module Loader: The kernel loader requires the system symbol
table. The system symbol table does not require the presence of the loader.
■
Debugging Facilities: The target-based symbolic debugging facilities and user
commands such as i and tt, rely on the system symbol table to provide
information about entry points of tasks, symbolic contents of call stacks, and
so on.
■
Kernel Shell: The kernel shell does not strictly require the system symbol table,
but its functionality is greatly limited without it. The kernel shell requires the
system symbol table to provide the ability to run functions using their
symbolic names. The kernel shell uses the system symbol table to execute shell
commands, to call system routines, and to edit global variables. The kernel
shell also includes the library usrLib, which contains the commands i, ti, sp,
period, and bootChange.
622
12 Target Tools
12.4 Kernel Symbol Tables
■
WDB Target Agent: The WDB target agent adds symbols to the system symbol
table as part of the symbol synchronization with the host.
If the facilities provided by the symbol table library are needed for user
(non-operating system) code, another symbol table should be created and
manipulated using the symbol library. See 12.4.6 Creating and Using User Symbol
Tables, p.623.
NOTE: If you choose to use both the host-resident and target-resident tools at the
same time, use the synchronization method to ensure that both the host and target
resident tools share the same list of symbols.
12.4.5 Synchronizing Host and Kernel Modules List and Symbol Table
If both host tools and target tools are going to be used with a target system, the
modules list and symbol table maintained on the host system must be
synchronized with the modules list and symbol table maintained on the target.
This ensures that the host and target tools share the same list of symbols.
The host tools maintain their own modules list and symbol table—the target server
modules list and symbol table— on the host. In this chapter it is referred to as the
host modules list and symbol table.
Module list and symbol table synchronization is provided automatically when
VxWorks is configured with the WDB target agent and the kernel object-module
loader (INCLUDE_WDB and INCLUDE_LOADER). To remove this feature, you
need only remove the INCLUDE_WDB_MDL_SYM_SYNC component.
Note that the modules and symbols synchronization will only work if the WDB
agent is in task mode. If the WDB agent is in system mode, the modules and
symbols added from both the host and the target will not be synchronized.
For information about WDB, see 12.6 WDB Target Agent, p.626.
12.4.6 Creating and Using User Symbol Tables
Although it is possible for user code in the kernel to manipulate symbols in the
system’s symbol table, this is not a recommended practice. Addition and removal
of symbols to and from the symbol table should only be carried out by operating
system libraries. Any other use of the system symbol table may interfere with the
proper operation of the operating system; and even simply introducing additional
623
12
VxWorks
Kernel Programmer's Guide, 6.6
symbols could have an adverse and unpredictable effect on linking any modules
that are subsequently downloaded.
Therefore, user-defined symbols should not be added programmatically to the
system symbol table. Instead, when user code in kernel space requires a symbol
table for its own purposes, a user symbol table should be created. For more
information, see the VxWorks API reference for symLib.
12.5 Show Routines
VxWorks includes system information routines that can be invoked from the
shell’s C interpreter. They should not be used programmatically.
The show routines print pertinent system status on the specified object or service;
however, they show only a snapshot of the system service at the time of the call
and may not reflect the current state of the system. To use these routines, you must
include the appropriate component when you configure VxWorks. When you
invoke them, their output is sent to the standard output device. Table 12-14 lists
common system show routines:
Table 12-14
Show Routines
Call
Description
envShow( )
Displays the environment for a given INCLUDE_TASK_SHOW
task on stdout.
memPartShow( )
Shows the partition blocks and
statistics.
INCLUDE_MEM_SHOW
memShow( )
System memory show routine.
INCLUDE_MEM_SHOW
moduleShow( )
Prints information for all loaded
modules, or an individual module.
INCLUDE_MODULE_MANAGER
msgQShow( )
Message queue show utility (for both INCLUDE_POSIX_MQ_SHOW
POSIX and native VxWorks message INCLUDE_MSG_Q_SHOW
queues).
624
Component
12 Target Tools
12.5 Show Routines
Table 12-14
Show Routines (cont’d)
Call
Description
Component
semShow( )
Semaphore show utility (for both
POSIX and native VxWorks
semaphores).
INCLUDE_SEM_SHOW,
INCLUDE_POSIX_SEM_SHOW
show( )
Generic object show utility. The
show( ) routine does not work with
modules or symbol tables; see
moduleShow( ) and symshow( ).
INCLUDE_SHOW_ROUTINES
stdioShow( )
Standard I/O file pointer show
utility.
INCLUDE_STDIO_SHOW
symShow( )
Prints symbol table information.
INCLUDE_SYM_TBL_SHOW
taskSwitchHookShow( ) Shows the list of task switch routines. INCLUDE_TASK_HOOKS_SHOW
taskCreateHookShow( )
Shows the list of task create routines. INCLUDE_TASK_HOOKS_SHOW
taskDeleteHookShow( )
Shows the list of task delete routines. INCLUDE_TASK_HOOKS_SHOW
taskShow( )
Displays the contents of a task
control block.
INCLUDE_TASK_SHOW
wdShow( )
Watchdog show utility.
INCLUDE_WATCHDOGS_SHOW
Table 12-15
Network Show Routines
Call
Description
ifShow( )
Display the attached network interfaces.
inetstatShow( )
Display all active connections for IP sockets.
ipstatShow( )
Display IP statistics.
netPoolShow( )
Show pool statistics.
netStackDataPoolShow( )
Show network stack data pool statistics.
netStackSysPoolShow( )
Show network stack system pool statistics.
mbufShow( )
Report mbuf statistics.
625
12
VxWorks
Kernel Programmer's Guide, 6.6
Table 12-15
Network Show Routines (cont’d)
Call
Description
netShowInit( )
Initialize network show routines.
arpShow( )
Display entries in the system ARP table.
arptabShow( )
Display the known ARP entries.
routestatShow( )
Display routing statistics.
routeShow( )
Display host and network routing tables.
hostShow( )
Display the host table.
mRouteShow( )
Print the entries of the routing table.
12.6 WDB Target Agent
The VxWorks WDB target agent is a target-resident, run-time facility that is
required for connecting host tools to a VxWorks target system. It is not required
(or generally useful) for deployed systems, nor is it required for development
using the kernel shell (see 12.2 Kernel Shell, p.577). The facility is also referred to as
the target agent, the WDB agent, or simply as WDB. The acronym stands for Wind
DeBug.
The WDB agent carries out requests transmitted from host-based debugging tools
and replies with the results. The WDB agent contains a compact implementation
of UDP/IP, and a proprietary RPC messaging protocol called WDB. The WDB
(Wind DeBug) protocol specifies how the target server (on the host) communicates
with the target agent (on the target). The protocol includes a compact
programming language called Gopher, which permits on-the-fly extension by
supporting programmable investigation of the target system.
The WDB protocol provides a core minimum of the services necessary to respond
to requests from the host tools. These protocol requests include memory
transactions, breakpoint/event notification services, virtual I/O support, tasking
control, and real-time process control. The WDB protocol uses the Sun
Microsystems specification for External Data Representation (XDR) for data
transfer.
626
12 Target Tools
12.6 WDB Target Agent
WDB can be configured for system mode debugging, task mode debugging, or
both (switching between the two modes under the control of host tools). In task
mode, WDB runs as a kernel task. In system mode, WDB operates independently
of the kernel, and the kernel is under WDB’s control. With system mode, WDB can
be started before VxWorks is running, which can be particularly useful in the early
stages of porting a BSP to a new board. (See Debugging Mode Options, p.632 and
12.6.6 Starting the WDB Target Agent Before the VxWorks Kernel, p.640).
The WDB agent’s interface to communications drivers avoids the run-time I/O
system, so that the WDB agent remains independent of the run-time OS. Drivers
for the WDB agent are low-level drivers that provide both interrupt-driven and
polling-mode operation. Polling mode is required to support system-level control
of the target.
The WDB agent synthesizes the target-control strategies of task-level and
system-wide debugging. The agent can execute in either mode and switch
dynamically between them, provided the appropriate drivers are present in the
Board Support Package (BSP). This permits debugging of any aspect of an
embedded application whether it is a task, an interrupt service routine, or the
kernel itself.
NOTE: If both host tools and target tools are going to be used with a target system,
the modules list and symbol table maintained on the host system must be
synchronized with the modules list and symbol table maintained on the target.
This ensures that the host and target tools share the same list of symbols. See the
discussion of the INCLUDE_WDB_MDL_SYM_SYNC component in Additional
Options, p.634, and 12.4.5 Synchronizing Host and Kernel Modules List and Symbol
Table, p.623.
12.6.1 Configuring VxWorks with the WDB Target Agent
WDB target agent functionality is provided by a suite of components, some of
which are optional, and others of which provide support for alternate modes of
connection. By default VxWorks is configured with a pipe connection for the
VxWorks simulator, and an Enhanced Network Driver (END) connection for all
hardware targets.
The INCLUDE_WDB component provides the basic target agent facilities. It allows
you to connect a target server, get basic information about the target, and load
modules.
627
12
VxWorks
Kernel Programmer's Guide, 6.6
Basic WDB Configuration
The configuration parameters for the basic INCLUDE_WDB component are:
WDB_COMM_PORT
The UDP port used by the WDB agent to connect to the host (the default is
0x4321). This is also the default port used by the target server. If you must
change this port, be sure to update the port information in Wind River
Workbench.
WDB_STACK_SIZE
The stack size of the WDB agent.
WDB_POOL_SIZE
The size of the memory pool used by the WDB agent.
To configure the WDB agent, you must also choose the appropriate
connection-type component, one or more debugging modes, one initialization
component, and any other options you may need. For information about reducing
the size of the agent, see 12.6.4 Scaling the WDB Target Agent, p.639.
Host-Target Communication Options
The WDB components required for different types of host-target connections are
described in Table 12-16. VxWorks should be configured with only one WDB
communication component.
Table 12-16
WDB Connection Components
Component
Description
INCLUDE_WDB_COMM_END
The WDB enhanced network driver
(END) connection component. The END
driver supports both system and task
mode debugging. This component is the
default.
INCLUDE_WDB_COMM_NETWORK
The WDB UDP/IP network connection
component. This communication type
only supports task mode debugging.
INCLUDE_WDB_PROXY INCLUDE_
WDB_PROXY_TIPC
The WDB TIPC components required for
the gateway system on a TIPC network.
628
12 Target Tools
12.6 WDB Target Agent
Table 12-16
!
WDB Connection Components (cont’d)
Component
Description
INCLUDE_WDB_COMM_TIPC
The WDB TIPC component for other
targets (non-gateway) on a TIPC network.
INCLUDE_WDB_COMM_SERIAL
The WDB serial connection component.
Useful when no network connections are
available.
INCLUDE_WDB_COMM_VTMD
The WDB visionICE II or visionProbe II
emulator connection component. This
communication link is useful when
debugging hardware bring up.
INCLUDE_WDB_COMM_PIPE
The WDB simulator pipe connection
component—used only for the VxWorks
simulator.
INCLUDE_WDB_COMM_CUSTOM
A WDB custom connection component,
created by the user (see 12.6.7 Creating a
Custom WDB Communication Component,
p.642).
WARNING: Both VxWorks and the host target connection must be configured for
the same type of host-target communication facilities. For example, if a serial
connection is going to be used, then VxWorks must be configured with
INCLUDE_WDB_COMM_SERIAL and the host target server must be configured
with the wdbserial back end. For more information about target connection
configuration, see the Wind River Workbench User’s Guide: New Target Server
Connections.
629
12
VxWorks
Kernel Programmer's Guide, 6.6
Enhanced Network Driver Connection Configuration
The configuration parameters for the INCLUDE_WDB_COMM_END component
are:
WDB_MTU
The maximum transfer unit (MTU). The default MTU is 1500 bytes.
WDB_END_DEVICE_NAME
By default, this parameter is set to NULL and the END driver used by the
WDB agent is the one specified with a VxWorks boot loader device parameter.
If you want to use a different device, set this parameter to the name of the
device (for example, dc).
WDB_END_DEVICE_UNIT
If WDB_END_DEVICE_NAME is specified, set this parameter to the unit
number of the END device you want to use.
Network Connection Configuration
The sole configuration parameter for the INCLUDE_WDB_COMM_NETWORK
component is WDB_MTU, which defines the MTU for a UDP/IP network
connection.
TIPC Network Connection Configuration
The configuration parameters for the INCLUDE_WDB_COMM_TIPC component
are:
WDB_TIPC_PORT_TYPE
The TIPC port type. The default is 70.
WDB_TIPC_PORT_INSTANCE
The TIPC port instance. The default is 71.
Note that the INCLUDE_WDB_COMM_TIPC component is used for the targets on
a TIPC network that you want to connect to with host tools, but not for the target
that serves as a gateway between the host system’s TCP/IP network and the
targets’ TIPC network. See 12.6.3 Using the WDB Target Agent with a TIPC Network,
p.638 for more information.
For more information about TIPC itself, see the Wind River TIPC for VxWorks 6
Programmer's Guide.
630
12 Target Tools
12.6 WDB Target Agent
Serial Connection Configuration
The configuration parameters for the INCLUDE_WDB_COMM_SERIAL component
are:
WDB_MTU
The MTU for a serial connection.
WDB_TTY_BAUD
The bps rate of the serial channel. The default is 9600 bps. For better
performance, use the highest line speed available, which is often 38400 bps.
Values higher than 34800 may not provide satisfactory performance. Try a
slower speed if you suspect data loss.
WDB_TTY_CHANNEL
The channel number. Use 0 if you have only one serial port on the target. Use
1 (the default) if you want to keep the VxWorks console on the first serial port.5
If your target has a single serial channel, you can use the target server virtual
console to share the channel between the console and the target agent. You
must configure your system with the CONSOLE_TTY parameter set to NONE
and the WDB_TTY_CHANNEL parameter set to 0.
When multiplexing the virtual console with WDB communications, excessive
output to the console may lead to target server connection failures. The
following may help resolve this problem:
■
Decrease the amount of data being transmitted to the virtual console from
your application.
■
Increase the time-out period for the target server.
■
Increase the baud rate of the target agent and the target server connection.
INCLUDE_WDB_TTY_TEST
When set to TRUE, this parameter causes words WDB READY to be displayed
on the WDB serial port on startup. By default, this parameter is set to TRUE.
WDB_TTY_ECHO
When set to TRUE, all characters received by the WDB agent are echoed on the
serial port. As a side effect, echoing stops the boot process until a target server
is attached. By default, this parameter is set to FALSE.
5. VxWorks serial channels are numbered starting at 0. Thus Channel 1 corresponds to the
second serial port if the board’s ports are labeled starting at 1. If your board has only one
serial port, you must change WDB_TTY_CHANNEL to 0 (zero).
631
12
VxWorks
Kernel Programmer's Guide, 6.6
visionICE or visionProbe Connection Configuration
The sole configuration parameter for the INCLUDE_WDB_COMM_VTMD
component is TMD_DEFAULT_POLL_DELAY, which specifies the clock tick
interval for polling data on the target.
Pipe Connection Configuration
The sole configuration parameter for the INCLUDE_WDB_COMM_PIPE component
(for the VxWorks simulator only) is WDB_MTU, which defines the MTU for a pipe
connection.
Debugging Mode Options
WDB provides two debugging mode options by default: system mode and task
mode. With system mode, the entire system is stopped when a breakpoint is hit.
This allows you to set breakpoints anywhere, including ISRs. Note that for SMP
systems, software breakpoints are persistent—that is they are retained in target
memory (for UP systems they are not).
With task mode, a task or group of tasks is stopped when a breakpoint is set, but
an exception or an interrupt does not stop if it hits a breakpoint. When the WDB
agent is configured for task mode, the tWdbTask task is used to handle all WDB
requests from the host.
You can include support for both modes, which allows tools such as the host shell
or the debugger to dynamically switch from one mode to the other.
For information about WDB behavior with SMP systems, see 12.6.2 WDB Target
Agent and VxWorks SMP, p.637. For information about the SMP configuration of
VxWorks, see 15. VxWorks SMP.
System Mode Debugging Configuration
The INCLUDE_WDB_SYS component provides support for system mode
debugging. Note that this mode is only supported when the communication type
has a polling mode for reading the device, which is not the case with the network
component INCLUDE_WDB_COMM_NETWORK.
The configuration parameters for the INCLUDE_WDB_SYS component are:
WDB_SPAWN_OPTS
The task options flag used by tasks spawned in system mode.
632
12 Target Tools
12.6 WDB Target Agent
WDB_SPAWN_PRI
The task priority used by tasks spawned in system mode.
WDB_SPAWN_STACK_SIZE
The stack size used by tasks spawned by the WDB target agent.
Task Mode Debugging Configuration
The INCLUDE_WDB_TASK component provides support for task mode. The
configuration parameters are:
WDB_MAX_RESTARTS
The maximum number of times an agent can be restarted when it gets an error
(the default is 5).
WDB_RESTART_TIME
The delay (in seconds) before restarting the WDB agent task when it gets an
error (the default is 10 seconds).
WDB_TASK_OPTIONS
12
The options parameter of the WDB task.
WDB_TASK_PRIORITY
The priority of the WDB task. The default priority is 3.
WDB_SPAWN_STACK_SIZE
The stack size used by tasks spawned by the WDB target agent.
Process Management Options
The INCLUDE_WDB_RTP component provides support for real time process (RTP)
operations (creation, deletion) and notifications (creation, deletion). This
component is automatically included if the system supports real time processes
(INCLUDE_RTP) and task debugging mode (INCLUDE_WDB_TASK).
The INCLUDE_WDB_RTP_BP component provides support for real time process
debugging. It allows use of process-wide breakpoints. This component is
automatically included when the system supports real time processes
(INCLUDE_RTP) and task breakpoints (INCLUDE_WDB_TASK_BP).
The INCLUDE_WDB_RTP_CONTROL component allows the debugger to configure
a process or kernel task such that its child processes are stopped at creation; that
is, they do not start automatically when they are spawned. By default, processes
can be spawned with an option that causes them to stop before they run, but child
processes to not inherit this characteristic.
633
VxWorks
Kernel Programmer's Guide, 6.6
Initialization Options
WDB can be configured to start either before or after kernel initialization. By
default, WDB is started after the kernel has been initialized.
The INCLUDE_WDB_POST_KERNEL_INIT component causes WDB to be started
once kernel has fully been initialized. The INCLUDE_WDB_PRE_KERNEL_INIT
component causes WDB to be started before kernel initialization has completed.
If WDB starts before kernel initialization, it is possible to perform early system
debugging. However, because the kernel has not been initialized when WDB
starts, task debugging is not supported in this mode. In addition, the END
connection cannot be used with this mode because the network has not been
initialized when WDB starts. Also see 12.6.6 Starting the WDB Target Agent Before
the VxWorks Kernel, p.640.
When WDB starts after kernel initialization, all WDB features are fully supported.
It is, of course, not possible to debug kernel initialization activity.
Additional Options
The following components provide additional optional functions. You can include
or exclude them based on your requirements.
The INCLUDE_WDB_BANNER component displays the WDB banner on the
console.
The INCLUDE_WDB_BP component provides support for breakpoints in the WDB
agent itself. This component is needed if you want to debug the target from a host
tool. The configuration parameter for this component is WDB_BP_MAX, which
specifies the maximum number of breakpoints allocated on the target at startup.
When this number of breakpoints is reached, it is still possible to allocate space for
new breakpoints in task mode. In system mode, however, it is not possible to set
additional breakpoints once the limit has been reached.
The INCLUDE_WDB_BP_SYNC component provides a breakpoint synchronization
mechanism between host tools and target system. If this component is included in
the VxWorks configuration, host tools are notified of any breakpoint creations and
deletions that are made from the kernel shell. The component is automatically
included when debug is provided for both the kernel shell (with
INCLUDE_DEBUG) and the host tools (with INCLUDE_WDB_TASK_BP).
634
12 Target Tools
12.6 WDB Target Agent
The INCLUDE_WDB_CTXT component provides support for context operations:
creation, deletion, suspension, resumption. A context can be a task, a real time
process, or the system itself.
The INCLUDE_WDB_DIRECT_CALL component allows you to call functions in the
WDB agent context directly.
The INCLUDE_WDB_EVENTPOINTS component adds support for eventpoints. An
eventpoint can be a breakpoint, an eventpoint on context creation, or an eventpoint
on context deletion. This component is the core component for all eventpoint
types. Each time an eventpoint is hit, the corresponding event is sent to the target
server.
The INCLUDE_WDB_EVENTS component adds support for asynchronous events.
Asynchronous events are sent from the target to target server, to notify host tools
about event activity on the target; for example, if a breakpoint has been hit, an
exception occurred, or a context (task or process) has started or exited. The
component is required (and is automatically included) when using breakpoints,
exception notification, context start/exit notification, and so on.
The INCLUDE_WDB_EXC_NOTIFY component adds support for exception
notification. When an exception occurs on the target, the appropriate event is sent
to the target server.
The INCLUDE_WDB_EXIT_NOTIFY component adds support for context deletion
notification. To be notified of a context exit, an eventpoint of type WDB_CTX_EXIT
must be set. Tools set this eventpoint when they must be notified. This component
supports notification for task and real time process contexts.
The INCLUDE_WDB_FUNC_CALL component handles function calls by spawning
tasks to run the functions. This service is only available in task mode.
The INCLUDE_WDB_GOPHER component provides support for the Gopher
information gathering language. It is used by many host tools and cannot be
removed from a system that uses other WDB options. The configuration
parameters for this component are:
■
WDB_GOPHER_TAPE_LEN, which defines the length of one gopher tape.
Gopher tapes are used to record and upload data processed by the gopher.
The default tape length is 1400 words, each of which are 32 bits wide.
■
WDB_GOPHER_TAPE_NB, which defines the maximal number of gopher
tapes that can be dynamically allocated. At startup, only one gopher tape
is available. As needed, more tapes can be allocated. Dynamic allocation
of tapes is only available in task mode. The default number of tapes is 10.
635
12
VxWorks
Kernel Programmer's Guide, 6.6
The INCLUDE_WDB_MEM component provides support for reading from, and
writing to, target memory.
The INCLUDE_WDB_REG component provides support for reading from, and
writing to, registers. The WDB_REGS_SIZE configuration parameter defines the
size of an internal memory buffer used by the WDB agent to store coprocessor
registers (to allow access to the registers in system mode).
The INCLUDE_WDB_START_NOTIFY component provides support for context
creation notification. To be notified of a context exit, an eventpoint of type
WDB_CTX_START must be set. Tools set this eventpoint when they must be
notified. This component supports task and real time process contexts.
The INCLUDE_WDB_TASK_BP component provides support for breakpoints in
task debugging mode. This component is automatically included when WDB
breakpoints (INCLUDE_WDB_BP) and task debugging mode
(INCLUDE_WDB_TASK) are included in the system.
The INCLUDE_WDB_TASK_HOOKS component initializes task hooks needed to
support task debugging mode. It is automatically included when task debugging
mode (INCLUDE_WDB_TASK) is included, and should never be removed
manually.
The INCLUDE_WDB_TASK_REG component provides support for task register
operations (read and write). It is automatically included when WDB supports
register operations (INCLUDE_WDB_REG) and task debugging mode
(INCLUDE_WDB_TASK).
The INCLUDE_WDB_TSFS component adds support for a virtual file system
enabled by the WDB protocol, the Target Server File System (see 8.9 Target Server
File System: TSFS, p.518). This component is automatically included when the
INCLUDE_WVUPLOAD_TSFSSOCK component is included for System Viewer
upload support.
The INCLUDE_WDB_USER_EVENT component handles user defined events. For
more information about user events, see the VxWorks wdbUserEvtLib API
reference.
The INCLUDE_WDB_VIO component provides a driver for a virtual I/O (VIO)
access.
The INCLUDE_WDB_VIO_LIB component handles VIO access (read( ) and
write( )). It requires the VIO driver component and the events component.
The INCLUDE_WDB_MDL_SYM_SYNC component handles module and symbol
synchronization between the target and the target server. It is required only if both
the WDB agent (INCLUDE_WDB) and the kernel object-module loader
636
12 Target Tools
12.6 WDB Target Agent
(INCLUDE_LOADER) are available on the target and you are using the host-based
loader (through the debugger, for example). The component synchronizes the
records of modules and symbols that are kept by the host and kernel loaders. For
more information, see the VxWorks wdbMdlSymSyncLib API reference and
12.4.5 Synchronizing Host and Kernel Modules List and Symbol Table, p.623.
12.6.2 WDB Target Agent and VxWorks SMP
This section describes the behavior of the WDB target agent when used with an
SMP configuration of VxWorks. For information about the SMP configuration of
VxWorks, see 15. VxWorks SMP.
Task Mode Debugging
With the task mode debugging agent:
■
Breakpoint exceptions that occur while holding an ISR-callable spinlock are
ignored.
■
Breakpoint exceptions that occur while holding a task-only spinlock are
ignored.
■
Breakpoint exceptions that occur while interrupts are locked are ignored.
System Mode Debugging
With the system mode debugging agent:
■
When the system is stopped either by a breakpoint or an exception, all CPUs
of the target are stopped.
■
Breakpoint cannot be set on a specific CPU, they are always system wide.
■
The maximum number of hardware breakpoints that can be installed on a SMP
system is limited to the maximum number of hardware breakpoints of one
core.
■
On a stop request (for example, a sysSuspend( ) call from the host shell), all
processors of the system stop. However, if one processor of the system is
executing a portion of code protected against interrupts (for example with an
intCpuLock( ) call), then it will stop only when the interrupts are re-enabled.
637
12
VxWorks
Kernel Programmer's Guide, 6.6
12.6.3 Using the WDB Target Agent with a TIPC Network
Wind River host tools can be used to debug VxWorks targets on a TIPC network
that do not have direct access to the host by way of TCP/IP or a serial line. In order
to do so, however, one of the VxWorks targets on the TIPC network must serve as
a gateway system.
A gateway must have access to both the host’s TCP/IP network and the targets’
TIPC network, and it must run a target agent proxy server. The proxy server
supports both networking protocols and provides the link between the host target
server and the WDB agent on the target system, thus allowing for remote
debugging of the other VxWorks targets on the TIPC network. The proxy server
can support multiple connections between the host system and different VxWorks
targets.
Note that with TIPC, WDB system mode debugging is not supported over TIPC
(see Debugging Mode Options, p.632).
For information about TIPC, see the Wind River TIPC for VxWorks 6 Programmer’s
Guide.
Target System Configuration
The VxWorks gateway target and the other VxWorks targets on the TIPC network
(to which the host tools attach) must each be configured with different WDB
components:
■
The gateway target must be configured with the INCLUDE_WDB_PROXY and
INCLUDE_WDB_PROXY_TIPC components (as well as with both TIPC and
TCP/IP support).
■
Any other targets to which host tools will attach must be configured with the
basic INCLUDE_WDB component and the INCLUDE_WDB_COMM_TIPC
component (as well as with TIPC support).
When the INCLUDE_WDB_COMM_TIPC component is included, WDB system
mode is excluded from the configuration, as it is not supported with TIPC
communication.
For information about the configuration parameters for these components, see
Basic WDB Configuration, p.628 and TIPC Network Connection Configuration, p.630.
638
12 Target Tools
12.6 WDB Target Agent
Establishing a Host-Target Connection
To establish a connection between the host and the targets on the TIPC network,
first boot the gateway and other targets.
Wind River Workbench provides options for connecting with a target running a
WDB agent proxy. See the Wind River Workbench User’s Guide for more information.
From the command line, the syntax for starting a target server connection with a
target running the WDB agent proxy is as follows:
tgtsvr -V -B wdbproxy -tipc -tgt targetTipcAddress -tipcpt tipcPortType -tipcpi
tipcPortInstance wdbProxyIpAddress/name
In this command:
■
targetTipcAddress is the TIPC address of the target to which you want to
connect.
■
tipcPortType is the TIPC port type used for the WDB connection (the default is
70).
■
tipcPortInstance is the TIPC port instance used for the WDB connection (the
default is 71).
■
wdbProxyIpAddress/name is the IP address or target name of the gateway target
that is running the WDB proxy agent.
12.6.4 Scaling the WDB Target Agent
In a memory-constrained system, you may wish to create a smaller target agent.
To reduce its size, you can remove the optional facilities listed in Table 12-17. They
are otherwise included by default.
Table 12-17
Optional WDB Agent Components
Component
Description
INCLUDE_WDB_BANNER
Prints a banner to console after the agent is
initialized.
INCLUDE_WDB_VIO
Provides the VxWorks driver for accessing
virtual I/O.
INCLUDE_WDB_USER_EVENT
Provides the ability to send user events to the
host.
639
12
VxWorks
Kernel Programmer's Guide, 6.6
You can also reduce the maximum number of WDB breakpoints with the
WDB_BP_MAX parameter of the INCLUDE_WDB_BP component. If you are using a
serial connection, you can also set the INCLUDE_WDB_TTY_TEST parameter to
FALSE.
If you are using a communication path that supports both system and task mode
agents, then by default both agents are started. Since each agent consumes target
memory (for example, each agent has a separate execution stack), you may wish to
exclude one of the agents from the target system. You can configure the target to
use only a task-mode or only a system-mode agent with the INCLUDE_WDB_TASK
or INCLUDE_WDB_SYS options.
12.6.5 WDB Target Agent and Exceptions
If an application or BSP uses excHookAdd( ) or signal handlers to handle
exceptions, WDB does not notify the host tools of the exceptions handled by those
facilities. Host tool notification can be suppressed for all other exceptions by
removing the INCLUDE_WDB_EXC_NOTIFY component from the VxWorks
configuration.
If the WDB task (tWdbTask) takes an exception, it is restarted after a
(configurable) delay. The connection between the target server and the target
agent is down during the delay period. The length of the delay can be set with the
WDB_RESTART_TIME parameter. Note that the WDB task is started in the kernel
only if WDB is set to run in task mode.
12.6.6 Starting the WDB Target Agent Before the VxWorks Kernel
By default, the WDB target agent is initialized near the end of the VxWorks
initialization sequence. This is because the default configuration calls for the agent
to run in task mode and to use the network for communication; thus, WDB is
initialized after the kernel and the network.
In some cases—such as during BSP development—you may want to start the agent
before the kernel, and initialize the kernel under the control of the host tools.
VxWorks Configuration
To be able to start WDB before the kernel, reconfigure VxWorks as follows:
640
12 Target Tools
12.6 WDB Target Agent
1.
Choose a communication path that can support a system-mode agent; a raw
serial connection. (The END communication path cannot be used as it requires
that the system be started before it is initialized.)
2.
Select only the INCLUDE_WDB_SYS component—and not the task mode
component.
By default, the task mode starts two agents: a system-mode agent and a
task-mode agent. Both agents begin executing at the same time, but the
task-mode agent requires the kernel to be running.
3.
Remove the INCLUDE_WDB_BANNER component. For some architectures,
calling this component before kernel is initialized may hang the target.
4.
Add the INCLUDE_WDB_PRE_KERNEL_INIT component and remove the
INCLUDE_WDB_POST_KERNEL_INIT component. (See Initialization Options,
p.634.)
This causes the project code generator to make the usrWdbInit( ) call earlier in
the initialization sequence. It will be called from usrInit( ) just before the
kernel is started.6
Run-time Operation
When the host target server has connected to the system-mode WDB target agent,
you can resume the system to start the kernel under the agent’s control.
After connecting to the target agent, set a breakpoint in usrRoot( ), then continue
the system. The routine kernelInit( ) starts the multi-tasking kernel with
usrRoot( ) as the entry point for the first task. Before kernelInit( ) is called,
interrupts are still locked. By the time usrRoot( ) is called, interrupts are unlocked.
Errors before reaching the breakpoint in usrRoot( ) are most often caused by a
stray interrupt: check that you have initialized the hardware properly in the BSP
sysHwInit( ) routine. Once sysHwInit( ) is working properly, you no longer need
to start the agent before the kernel.
6. The code generator for prjConfig.c is based on the component descriptor language, which
specifies when components are initialized. The component descriptor files are searched in a
specified order, with the project directory being last, and overriding the default definitions
in the generic descriptor files. For more information, see 2.8.2 CDF Precedence and CDF Installation, p.75.
641
12
VxWorks
Kernel Programmer's Guide, 6.6
NOTE: If you use a serial connection when you start WDB before the kernel, you
must modify the SIO driver so that it can properly deal with interrupts and the
order of system initialization in this context. See the VxWorks Device Driver
Developer's Guide: Additional Drivers for detailed information.
!
CAUTION: When the agent is started before the kernel, there is no way for the host
to get the agent’s attention until a breakpoint occurs. This is because only system
mode is supported and the WDB communication channel is set to work in polled
mode only. On the other hand, the host does not really need to get the agent’s
attention: you can set breakpoints in usrRoot( ) to verify that VxWorks can get
through this routine. Once usrRoot( ) is working, you can start the agent after the
kernel (that is, within usrRoot( )), after which the polling task is spawned
normally.
12.6.7 Creating a Custom WDB Communication Component
To create a custom communication component:
1.
Write a WDB packet driver. The template file
installDir/vxworks-6.x/target/src/drv/wdb/wdbTemplatePktDrv.c can be
used as a starting point.
2.
Create a configlette file in installDir/vxworks-6.x/target/config/comps/src that
contains the routine wdbCommDevInit( ) to initialize the driver. You can base
it on one of the WDB communication path configlettes in this directory
(wdbEnd.c, wdbSerial.c, and so on).
3.
Create a component descriptor file (CDF) for the custom component called
01wdbCommCustom.cdf in the directory
installDir/vxworks-6.x/target/config/comps/vxWorks. The file must identify
the driver module, the configlette, and any special parameters.
For information about creating custom components, see 2.8 Custom VxWorks
Components and CDFs, p.67, including Defining a Component, p.68.
The custom communication component can then be used by configuring VxWorks
with WDB_COMM_CUSTOM.
642
12 Target Tools
12.7 Common Problems
12.7 Common Problems
This section lists frequently encountered problems that can occur when using the
target tools.
Kernel Shell Debugging Never Hits a Breakpoint
I set a breakpoint on a function I called from the kernel shell, but the breakpoint is
not being hit. Why not?
Explanation
The kernel shell task runs with the VX_UNBREAKABLE option. Functions that are
called directly from the kernel shell command prompt, are executed within the
context of the kernel shell task. Therefore, breakpoints set within the directly called
function will not be hit.
12
Solution
Instead of running the function directly, use taskSpawn( ) with the function as the
entry point, or the shell’s C interpreter sp( ) command.
Insufficient Memory
The kernel object-module loader reports insufficient memory to load a module;
however, checking available memory indicates the amount of available memory to
be sufficient. What is happening and how do I fix it?
Explanation
The kernel loader calls the device drivers through a VxWorks’ transparent
mechanism for file management, which makes calls to open, close, and ioctl. If you
use the kernel loader to load a module over the network (as opposed to loading
from a target-system disk), the amount of memory required to load an object
module depends on what kind of access is available to the remote file system over
the network. This is because, depending on the device that is actually being used
for the load, the calls initiate very different operations.
For some devices, the I/O library makes a copy of the file in target memory.
Loading a file that is mounted over a device using such a driver requires enough
memory to hold two copies of the file simultaneously. First, the entire file is copied
643
VxWorks
Kernel Programmer's Guide, 6.6
to a buffer in local memory when opened. Second, the file resides in memory when
it is linked to VxWorks. This copy is then used to carry out various seek and read
operations. Therefore, using these drivers requires sufficient memory available to
hold two copies of the file to be downloaded, as well as a small amount of memory
for the overhead required or the load operation.
Also consider that loading a module sometimes requires additional space, as the
sections have to be aligned in memory (whereas the toolchain may compact them
all in the object file to save space). See 12.3.5 Specifying Memory Locations for Loading
Objects, p.610.
Solution
Download the file using a different device. Loading an object module from a host
file system mounted through NFS only requires enough memory for one copy of
the file (plus a small amount of overhead).
"Relocation Does Not Fit" Error Message
When downloading, the following type of error message occurs:
Relocation value does not fit in 26 bits (offset: 0x10, type: 1)
What does this error mean and what should I do?
Explanation
Some architectures have instructions that use less than 32 bits to reference a nearby
position in memory. Using these instructions can be more efficient than always
using 32 bits to refer to nearby places in memory.
The problem arises when the compiler has produced such a reference to something
that lies farther away in memory than the range that can be accessed with the
reduced number of bits. For instance, if a call to printf( ) is encoded with one of
these instructions, the load may succeed if the object code is loaded near the kernel
code, but fail if the object code is loaded farther away from the kernel image.
For additional information, see Function Calls, Relative Branches, and Load Failures,
p.616.
Solution
T he loader prints the relocation type and offset as part of the error message to
facilitate diagnostics. The offset and types can be retrieved with the readelf -r
command. Relocation types are described in the ELF architecture supplement.
644
12 Target Tools
12.7 Common Problems
Recompile the object file using -Xcode-absolute-far for the Wind River compilers,
and for GNU compilers, the appropriate long call option, -mlongcall (for PPC
architecture). See the VxWorks Architecture Supplement for the appropriate options.
Missing Symbols
Symbols in code modules downloaded from the host do not appear from the
kernel shell, and vice versa. Symbols created from the host shell are not visible
from the kernel shell, or symbols created from the kernel shell are not visible from
the host shell. Why is this happening, and how can I get them to appear?
Explanation
The symbol synchronization mechanism must be enabled separately on the host
and target.
Solution
12
Check to see if the module and symbol synchronization is enabled for the target
server as well as compiled into the image. For more information, see
12.4.5 Synchronizing Host and Kernel Modules List and Symbol Table, p.623.
Kernel Object-Module Loader is Using Too Much Memory
Including the kernel loader causes the amount of available memory to be much
smaller. How can I get more memory?
Explanation
Including the kernel loader causes the system symbol table to be included. This
symbol table contains information about every global symbol in the compiled
VxWorks image.
Using the kernel loader takes additional memory away from your application—
most significantly for the target-resident symbol table required by the kernel
loader.
Solution
Use the host tools rather than the target tools and remove all target tools from your
VxWorks image.
645
VxWorks
Kernel Programmer's Guide, 6.6
Symbol Table Unavailable
The system symbol table failed to download onto my target. How can I use the
kernel shell to debug the problem, since I cannot call functions by name?
Solution
Use addresses of functions and data, rather than using the symbolic names. The
addresses can be obtained from the VxWorks image on the host, using the nmarch
utility.
The following is an example from a UNIX host:
> nmarch vxWorks | grep memShow
0018b1e8 T memShow
0018b1ac T memShowInit
Use this information to call the function by address from the kernel shell. (The
parentheses are mandatory when calling by address.)
-> 0x0018b1e8 ()
status
bytes
blocks
avg block max block
------ --------- -------- ---------- ---------current
free
14973336
20
748666
12658120
alloc 14201864
16163
878
cumulative
alloc 21197888
value = 0 = 0x0
142523
148
-
For modules that are relocated, use nm on the module to get the function address
(which is the offset within the module's text segment) then add to that value the
starting address of the text segment of the module when it was loaded in memory.
646
13
C++ Development
13.1 Introduction 647
13.2 Configuring VxWorks for C++ 648
13.3 C++ Code Requirements 649
13.4 Using C++ in Signal Handlers and ISRs 649
13.5 Downloadable Kernel Modules in C++ 650
13.6 C++ Compiler Differences 653
13.7 Namespaces 656
13.8 C++ Demo Example 657
13.1 Introduction
This chapter provides information about C++ development for VxWorks using the
Wind River and GNU toolchains.
!
WARNING: Wind River Compiler C++ and GNU C++ binary files are not
compatible.
647
VxWorks
Kernel Programmer's Guide, 6.6
NOTE: This chapter provides information about facilities available in the VxWorks
kernel. For information about facilities available to real-time processes, see the
corresponding chapter in the VxWorks Application Programmer’s Guide.
13.2 Configuring VxWorks for C++
By default, VxWorks includes only minimal C++ support. You can add C++
functionality by including any or all of the following components:
INCLUDE_CTORS_DTORS
(included in default kernels)
Ensures that compiler-generated initialization functions, including initializers
for C++ static objects, are called at kernel start up time.
INCLUDE_CPLUS
Includes basic support for C++ applications. Typically this is used in
conjunction with INCLUDE_CPLUS_LANG.
INCLUDE_CPLUS_LANG
Includes support for C++ language features such as new, delete, and
exception handling.
INCLUDE_CPLUS_IOSTREAMS
Includes all library functionality.
INCLUDE_CPLUS_DEMANGLER
Includes the C++ demangler, which is useful if you are using the kernel shell
loader because it provides for returning demangled symbol names to kernel
shell symbol table queries. This component is added by default if both the
INCLUDE_CPLUS and INCLUDE_SYM_TBL components are included in
VxWorks.
648
13 C++ Development
13.3 C++ Code Requirements
13.3 C++ Code Requirements
Any VxWorks task that uses C++ must be spawned with the VX_FP_TASK option.
By default, tasks spawned from host tools (such as the Wind Shell) automatically
have VX_FP_TASK enabled.
!
WARNING: Failure to use the VX_FP_TASK option when spawning a task that uses
C++ can result in hard-to-debug, unpredictable floating-point register corruption
at run-time.
If you reference a (non-overloaded, global) C++ symbol from your C code you
must give it C linkage by prototyping it using extern "C":
#ifdef __cplusplus
extern "C" void myEntryPoint ();
#else
void myEntryPoint ();
#endif
You can also use this syntax to make C symbols accessible to C++ code. VxWorks
C symbols are automatically available to C++ because the VxWorks header files
use this mechanism for declarations.
Each compiler has its own C++ libraries and C++ headers (such as iostream and
new). The C++ headers are located in the compiler installation directory rather
than in installDir/vxworks-6.x/target/h. No special flags are required to enable the
compilers to find these headers.
NOTE: In releases prior to VxWorks 5.5, Wind River recommended the use of the
flag -nostdinc. This flag should not be used with the current release since it prevents
the compilers from finding headers such as stddef.h.
13.4 Using C++ in Signal Handlers and ISRs
Special care must be taken when using C++ code in signal handlers and ISRs. For
information in this regard, see 4.18.5 Signal Handlers, p.236 and 4.20.3 Writing and
Debugging ISRs, p.244.
649
13
VxWorks
Kernel Programmer's Guide, 6.6
13.5 Downloadable Kernel Modules in C++
C++ code that is downloaded into the VxWorks kernel should be linked into a
single downloadable object module. It must also be munched, and any COMDAT
or linkonce sections collapsed. VxWorks provides several strategies available for
calling static constructors and destructors in downloadable modules.
13.5.1 Use a Single C++ Module
The VxWorks loader only supports C++ modules that are self-contained. A
self-contained C++ module is one that does not use classes from other C++
modules, and whose classes are not used by other C++ modules. In particular, a
module must either contain its own copy of the standard library, or not use the
C++ standard library at all.
To produce self-contained modules, all C++ object files that are to be downloaded
should be linked into a single downloadable object module.
Unloading a C++ module that is not self-contained may result in dangling
references from objects created in other modules to data structures in the unloaded
module. Specifically, this can happen if the iostreams portion of the standard
library is initialized from a module that is later unloaded. In that case, any further
use of iostreams may fail with a kernel exception (accessing an invalid address).
!
WARNING: C++ object files must be linked into one downloadable kernel module.
For information about the kernel loader, see 12.3 Kernel Object-Module Loader,
p.603.
13.5.2 Munching a C++ Application Module
Before a C++ module can be downloaded to the VxWorks kernel, it must undergo
an additional host processing step, which for historical reasons, is called munching.
Munching performs the following tasks:
■
Initializes support for static objects.
■
Ensures that the C++ run-time support calls the correct constructors and
destructors in the correct order for all static objects.
■
For the Wind River Compiler, collapses any COMDAT sections automatically;
for the GNU compiler, collapses any linkonce automatically.
650
13 C++ Development
13.5 Downloadable Kernel Modules in C++
Munching must be performed after compilation and before download.
Munching Examples
For each toolchain, the following examples compile a C++ application source file,
hello.cpp, run munch on the .o, compile the generated ctdt.c file, and link the
application with ctdt.o to generate a downloadable module, hello.out.
Using the Wind River Toolchain
1.
Compile the source code:
$ dcc -tPPC604FH:vxworks61 -Xlocal-data-area-static-only -XO
-IinstallDir/vxworks-6.1/target/h -DCPU=PPC32 -DTOOL_FAMILY=diab
-DTOOL=diab \ -D_WRS_KERNEL -c hello.cpp
2.
\
Munch the object file:
$ ddump -Ng hello.o | tclsh \
installDir/vxworks-6.1/host/resource/hutils/tcl/munch.tcl -c ppc > ctdt.c
3.
Compile the munch output:
$ dcc -tPPC604FH:vxworks61 -Xlocal-data-area-static-only -XO
-IinstallDir/vxworks-6.1/target/h -DCPU=PPC32 -DTOOL_FAMILY=diab
-DTOOL=diab \ -D_WRS_KERNEL -c ctdt.c
4.
\
Link the original object file with the munched object file to create a
downloadable module:
$ dld -tPPC604FH:vxworks61 -X -r4 -o hello.out hello.o ctdt.o
NOTE: The -r4 option collapses any COMDAT sections contained in the input files.
Using the GNU Toolchain
1.
Compile the source code:
ccppc -mcpu=604 -mstrict-align -O2 -fno-builtin \
-IinstallDir/vxworks-6.1/target/h \
-DCPU=PPC604 -DTOOL_FAMILY=gnu -DTOOL=gnu -c hello.cpp
2.
Munch the object file:
nmppc hello.o | wtxtcl installDir/vxworks-6.1/host/src/hutils/munch.tcl \
-c ppc > ctdt.c
3.
Compile the munch output:
ccppc -mcpu=604 -mstrict-align -fdollars-in-identifiers -O2 \
-fno-builtin -IinstallDir/vxworks-6.1/target/h \
-DCPU=PPC604 -DTOOL_FAMILY=gnu -DTOOL=gnu -c ctdt.c
651
13
VxWorks
Kernel Programmer's Guide, 6.6
4.
Link the original object file with the munched object file to create a
downloadable module:
ccppc -r -nostdlib -Wl,-X \
-T installDir/vxworks-6.1/target/h/tool/gnu/ldscripts/link.OUT \
-o hello.out hello.o ctdt.o
NOTE: The VxWorks kernel object module loader does not support linkonce
sections directly. Instead, the linkonce sections must be merged and collapsed into
standard text and data sections before loading. The GNU -T option collapses any
linkonce sections contained in the input files.
Using a Generic Makefile Rule
If you use the VxWorks makefile definitions, you can write a simple munching rule
which (with appropriate definitions of CPU and TOOL) works across all
architectures for both GNU and Wind River Compiler toolchains.
CPU
TOOL
= PPC604
= gnu
TGT_DIR = $(WIND_BASE)/target
include $(TGT_DIR)/h/make/defs.bsp
default : hello.out
%.o : %.cpp
$(CXX) $(C++FLAGS) -c $<
%.out : %.o
$(NM) $*.o | $(MUNCH) > ctdt.c
$(CC) $(CFLAGS) $(OPTION_DOLLAR_SYMBOLS) -c ctdt.c
$(LD_PARTIAL) $(LD_PARTIAL_LAST_FLAGS) -o $@ $*.o ctdt.o
After munching, downloading, and linking, the static constructors and destructors
are called. This step is described next.
13.5.3 Calling Static Constructors and Destructors Interactively
The kernel loader provides both manual and automatic options for calling static
constructors and destructors.
Automatic invocation is the default strategy. Static constructors are executed just
after the module is downloaded to the target and before the module loader returns
to its caller. Static destructors are executed just prior to unloading the module.
652
13 C++ Development
13.6 C++ Compiler Differences
Manual invocation means that the user must call static constructors explicitly, after
downloading the module, but before running the application. It also requires the
user to call static destructors explicitly, after the task finishes running, but before
unloading the module.
Static constructors are called by invoking cplusCtors( ). Static destructors are
called by invoking cplusDtors( ). These routines take an individual module name
as an argument. However, you can also invoke all of the static constructors or
destructors that are currently loaded into a system by calling these routines
without an argument.
!
CAUTION: When using the manual invocation method, constructors for each
module must only be run once.
You can change the strategy for calling static constructors and destructors at
run-time with the cplusXtorSet( ) routine. To report on the current strategy, call
cplusStratShow( ).
For more information on the routines mentioned in this section, see the API entries
in the online reference manuals.
Also see 12.3.3 Summary List of Kernel Object-Module Loader Options, p.606 for
information about the C++ loader and unloader options.
13.6 C++ Compiler Differences
The Wind River C++ Compiler uses the Edison Design Group (EDG) C++ front
end. It fully complies with the ANSI C++ Standard. For complete documentation
on the Wind River Compiler and associated tools, see the Wind River C/C++
Compiler User's Guide.
The GNU compiler supports most of the language features described in the ANSI
C++ Standard. For complete documentation on the GNU compiler and on the
associated tools, see the GNU ToolKit User’s Guide.
!
WARNING: Wind River Compiler C++ and GNU C++ binary files are not
compatible.
The following sections briefly describe the differences in compiler support for
template instantiation and run-time type information.
653
13
VxWorks
Kernel Programmer's Guide, 6.6
13.6.1 Template Instantiation
In C, every function and variable used by a program must be defined in exactly one
place (more precisely one translation unit). However, in C++ there are entities
which have no clear point of definition but for which a definition is nevertheless
required. These include template specializations (specific instances of a generic
template; for example, std::vector int), out-of-line bodies for inline functions, and
virtual function tables for classes without a non-inline virtual function. For such
entities a source code definition typically appears in a header file and is included
in multiple translation units.
To handle this situation, both the Wind River Compiler and the GNU compiler
generate a definition in every file that needs it and put each such definition in its
own section. The Wind River compiler uses COMDAT sections for this purpose,
while the GNU compiler uses linkonce sections. In each case the linker removes
duplicate sections, with the effect that the final executable contains exactly one
copy of each needed entity.
NOTE: Only the WRC linker can be used to process files containing COMDAT
sections. Similarly only the GNU linker can be used on files containing linkonce
sections. Furthermore the VxWorks target and host loaders are not able to process
COMDAT and linkonce sections. A fully linked VxWorks image will not contain
any COMDAT or linkonce sections. However intermediate object files compiled
from C++ code may contain such sections. To build a downloadable C++ module,
or a file that can be processed by any linker, you must perform an intermediate link
step using the -r5 option (WRC) or specifying the link.OUT linker script (GCC).
See 10.4.1 Munching C++ Application Modules for full details. (Note that while the
the -r5 and -r4 options—the latter referred to elsewhere in this chapter—both
collapse COMDAT files, their overall purpose is different, and their use is
mutually exclusive in a single linker command.)
It is highly recommended that you use the default settings for template
instantiation, since these combine ease-of-use with minimal code size. However it
is possible to change the template instantiation algorithm; see the compiler
documentation for details.
Wind River Compiler
The Wind River Compiler C++ options controlling multiple instantiation of
templates are:
654
13 C++ Development
13.6 C++ Compiler Differences
-Xcomdat
This option is the default. When templates are instantiated implicitly, the
generated code or data section are marked as comdat. The linker then
collapses identical instances marked as such, into a single instance in memory.
!
CAUTION: If code is going to be used as downloadable kernel modules, the -r4
option must be used to collapse any COMDAT sections contained in the input files.
See 13.5.2 Munching a C++ Application Module, p.650.
-Xcomdat-off
Generate template instantiations and inline functions as static entities in the
resulting object file. Can result in multiple instances of static member-function
or class variables.
For greater control of template instantiation, the -Ximplicit-templates-off option
tells the compiler to instantiate templates only where explicitly called for in source
code; for example:
template class A;
template int f1(int);
// Instantiate A and all member functions.
// Instantiate function int f1{int).
13
GNU Compiler
The GNU C++ compiler options controlling multiple instantiation of templates
are:
-fimplicit-templates
This option is the default. Template instantiations and out-of-line copies of
inline functions are put into special linkonce sections. Duplicate sections are
merged by the linker, so that each instantiated template appears only once in
the output file.
!
CAUTION: The VxWorks dynamic loader does not support linkonce sections
directly. Instead, the linkonce sections must be merged and collapsed into
standard text and data sections before loading. This is done with a special link step
described in 13.5.2 Munching a C++ Application Module, p.650.
-fno-implicit-templates
This is the option for explicit instantiation. Using this strategy explicitly
instantiates any templates that you require.
655
VxWorks
Kernel Programmer's Guide, 6.6
13.6.2 Run-Time Type Information
Both compilers support Run-time Type Information (RTTI), and the feature is
enabled by default. This feature adds a small overhead to any C++ program
containing classes with virtual functions.
For the Wind River Compiler, the RTTI language feature can be disabled with the
-Xrtti-off flag.
For the GNU compiler, the RTTI language feature can be disabled with the
-fno-rtti flag.
13.7 Namespaces
Both the Wind River and GNU C++ compilers supports namespaces. You can use
namespaces for your own code, according to the C++ standard.
The C++ standard also defines names from system header files in a namespace
called std. The standard requires that you specify which names in a standard
header file you will be using.
The following code is technically invalid under the latest standard, and will not
work with this release. It compiled with a previous release of the GNU compiler,
but will not compile under the current releases of either the Wind River or GNU
C++ compilers:
#include
int main()
{
cout << "Hello, world!" << endl;
}
The following examples provide three correct alternatives illustrating how the
C++ standard would now represent this code. The examples compile with either
the Wind River or the GNU C++ compiler:
// Example 1
#include
int main()
{
std::cout << "Hello, world!" << std::endl;
}
656
13 C++ Development
13.8 C++ Demo Example
// Example 2
#include
using std::cout;
using std::endl;
int main()
{
cout << "Hello, world!" << endl;
}
// Example 3
#include
using namespace std;
int main()
{
cout << "Hello, world!" << endl;
}
13.8 C++ Demo Example
The factory demo example demonstrates various C++ features in the kernel,
including the Standard Template Library, user-defined templates, run-time type
information, and exception handling. This demo is located in
installDir/vxworks-6.x/target/usr/apps/samples/cplusplus/factory.
To create, compile, build, and run this demo program you can use either
Workbench or the command line, as shown below.
For the factory demo, your kernel must include the following components in
VxWorks:
■
■
■
INCLUDE_CPLUS
INCLUDE_CPLUS_LANG
INCLUDE_CPLUS_IOSTREAMS
In addition, for GNU only, include the following components:
■
■
INCLUDE_CPLUS_STRING
INCLUDE_CPLUS_STRING_IO
To build factory from the command line, simply copy the factory sources to the
BSP directory, as shown below:
cd installDir/vxworks-6.1/target/config/bspDir
cp installDir/vxworks-6.1/target/src/demo/cplusplus/factory/factory.* .
657
13
VxWorks
Kernel Programmer's Guide, 6.6
Then, to build a bootable image containing the factory example, run make as
shown below:
make ADDED_MODULES=factory.o
and boot the target.
To build a downloadable image containing the factory example, run make as
shown below:
make factory.out
Then, from the WindSh, load the factory module, as shown below:
ld < factory.out
Finally, to run the factory demo example, type at the shell:
-> testFactory
Full documentation on what you should expect to see is provided in the source
code comments for the demo program.
658
PART II
Multiprocessing Technologies
14
Overview of Multiprocessing Technologies .... 661
15
VxWorks SMP ..................................................... 665
16
Shared-Memory Objects: VxMP ........................ 717
17
Distributed Shared Memory: DSHM .................. 755
18
Message Channels ............................................. 787
659
VxWorks
Kernel Programmer's Guide, 6.6
660
14
Overview of Multiprocessing
Technologies
14.1 Introduction 661
14.2 VxWorks SMP 662
14.3 Shared Memory Objects 662
14.4 Distributed Shared Memory 662
14.5 TIPC Over Distributed Shared Memory 662
14.6 Message Channels 663
14.1 Introduction
VxWorks provides various multiprocessor technologies, for asymmetric
multiprocessing (AMP) and symmetric multiprocessing (SMP) systems. These
include VxWorks SMP (an optional product), shared memory objects (VxMP),
distributed shared memory (DSHM), TIPC over DSHM, and message channels.
661
VxWorks
Kernel Programmer's Guide, 6.6
14.2 VxWorks SMP
VxWorks SMP is a configuration of VxWorks designed for symmetric
multiprocessing (SMP). It provides the same distinguishing RTOS characteristics
of performance and determinism as the uniprocessor (UP) configuration. The
differences between the SMP and UP configurations are limited, and strictly
related to support for multiprocessing. For detailed information, see 15. VxWorks
SMP.
14.3 Shared Memory Objects
VxMP is a VxWorks component that provides shared-memory objects dedicated
to high-speed synchronization and communication between tasks running in
separate instances of VxWorks. For detailed information, see 16. Shared-Memory
Objects: VxMP.
14.4 Distributed Shared Memory
The VxWorks distributed shared memory (DSHM) facility is a middleware
subsystem that allows multiple services to communicate over different types of
buses that support shared-memory communication. For detailed information, see
17. Distributed Shared Memory: DSHM.
14.5 TIPC Over Distributed Shared Memory
For information about TIPC, and about using TIPC with DSHM, see the Wind River
TIPC for VxWorks 6 Programmer’s Guide.
662
14 Overview of Multiprocessing Technologies
14.6 Message Channels
14.6 Message Channels
Message channels are a socket-based facility that provides for inter-task
communication within a memory boundary, between memory boundaries (kernel
and processes), between nodes (processors) in a multi-node cluster, and between
between multiple clusters. For detailed information, see 18. Message Channels.
14
663
VxWorks
Kernel Programmer's Guide, 6.6
664
15
VxWorks SMP
Optional Product
15.1 Introduction 666
15.2 Technology Overview 666
15.3 VxWorks SMP Configuration and Build 674
15.4 Booting VxWorks SMP 676
15.5 Programming for VxWorks SMP 676
15.6 Spinlocks for Mutual Exclusion and Synchronization 679
15.7 CPU-Specific Mutual Exclusion 685
15.8 Memory Barriers 687
15.9 Atomic Memory Operations 690
15.10 CPU Affinity 691
15.11 CPU Information and Management 694
15.12 Debugging SMP Code 698
15.13 Optimizing SMP Performance 699
15.14 Sample Programs 702
15.15 Migrating Code to VxWorks SMP 702
665
VxWorks
Kernel Programmer's Guide, 6.6
15.1 Introduction
VxWorks SMP is a configuration of VxWorks designed for symmetric
multiprocessing (SMP). It provides the same distinguishing RTOS characteristics
of performance and determinism as the uniprocessor (UP) configuration. The
differences between the SMP and UP configurations are limited, and strictly
related to support for multiprocessing.
This chapter describes the features provided by VxWorks to support symmetric
multiprocessing. It discusses the features that are unique to the SMP configuration,
as well as the differences in operating system facilities and programming practices
used for the UP configuration and the SMP configuration. It also provides
guidelines for migrating UP code to SMP code. In this chapter, the terms VxWorks
SMP and VxWorks UP is used to identify the uniprocessor and symmetric
multiprocessing configurations of VxWorks, respectively.
For information about features that are common to both the VxWorks SMP and
VxWorks UP configurations—such as multitasking, I/O, file systems, and so on—
see Part ICore Technologies, p.1.
NOTE: SMP support for VxWorks is available as an optional product. However,
default SMP system images for the simulators (including the WDB target agent,
kernel shell, object module loader, and so on) are provided with the standard
VxWorks installation as an introduction to the SMP product. For location
information, see Default VxWorks SMP Images, p.674.
15.2 Technology Overview
Multiprocessing systems include two or more processors in a single system.
Symmetric multiprocessing (SMP) is a variant of multiprocessing technology in
which one instance of an operating system controls all processors, and in which
memory is shared. SMP differs from asymmetric multiprocessing (AMP) in that an
AMP system has a separate instance of an operating system executing on each
processor (and each instance may or may not be the same type of operating
system).
666
15 VxWorks SMP
15.2 Technology Overview
15.2.1 Terminology
The terms CPU and processor are often used interchangeably in computer
documentation. However, it is useful to distinguish between the two for hardware
that supports SMP. In this guide, particularly in the context of VxWorks SMP, the
terms are used as follows:
CPU
A single processing entity capable of executing program instructions and
processing data (also referred to as a core, as in multicore).
processor
A silicon unit that contains one or more CPUs.
multiprocessor
A single hardware system with two or more processors.
uniprocessor
A silicon unit that contains a single CPU.
For example, a dual-core MPC8641D would be described as a processor with two
CPUs. A quad-core Broadcom 1480 would be described as a processor with four
CPUs.
Uniprocessor code may not always execute properly on an SMP system, and code
that has been adapted to execute properly on an SMP system may still not make
optimal use of symmetric multiprocessing. The following terms are therefore used
to clarify the state of code in relation to SMP:
SMP-ready
Runs correctly on an SMP operating system, although it may not make use of
more than one CPU (that is, does not take full advantage of concurrent
execution for better performance).
SMP-optimized
Runs correctly on an SMP operating system, uses more than one CPU, and
takes sufficient advantage of multitasking and concurrent execution to
provide performance gains over a uniprocessor implementation.
667
15
VxWorks
Kernel Programmer's Guide, 6.6
15.2.2 VxWorks SMP Operating System Features
With few exceptions, the SMP and uniprocessor (UP) configurations of VxWorks
share the same API—the difference amounts to only a few routines. There is binary
compatibility for both kernel and RTP applications between UP and SMP
configurations of VxWorks (for the same VxWorks release), as long as the
applications are based on the subset of APIs used by VxWorks SMP. A few
uniprocessor APIs are not suitable for an SMP system, and they are therefore not
provided. Similarly, SMP-specific APIs are not relevant to a uniprocessor system—
but default to appropriate uniprocessor behaviors (such as task spinlocks
defaulting to task locking), or have no effect.
VxWorks SMP is designed for symmetric target hardware. That is, each CPU has
equivalent access to all memory and all devices. VxWorks SMP can therefore run
on targets with multiple single-core processors or with multicore processors, as
long as they provide a uniform memory access (UMA) architecture with
hardware-managed cache-coherency.
This section provides a brief overview of areas in which VxWorks offers alternate
or additional features designed for symmetric multiprocessing. The topics are
covered in detail later in this chapter.
Multitasking
SMP changes the conventional uniprocessor paradigm of priority-based
preemptive multitasking programming, because it allows true concurrent
execution of tasks and handling of interrupts. This is possible because multiple
tasks can run on multiple CPUs, while being controlled by a single instance of an
operating system.
Uniprocessor multitasking environments are often described as ones in which
multiple tasks can run at the same time, but the reality is that the CPU only executes
one task at a time, switching from one task to the another based on the
characteristics of the scheduler and the arrival of interrupts. In an SMP system
concurrent execution is a fact and not an illusion.
Scheduling
VxWorks SMP provides a priority-based preemptive scheduler, like VxWorks UP.
In both VxWorks UP and VxWorks SMP, tasks are scheduled—and real-time
processes (RTPs) are not. However, the VxWorks SMP scheduler is different from
the uniprocessor scheduler in that it also manages the concurrent execution of
tasks on different CPUs.
668
15 VxWorks SMP
15.2 Technology Overview
Mutual Exclusion
Because SMP systems allow for truly concurrent execution, the uniprocessor
mechanisms for disabling (masking) interrupts and for suspending task
preemption in order to protect critical regions are inappropriate for—and not
available in—an SMP operating system. Enforcing interrupt masking or
suspending task preemption across all CPUs would defeat the advantages of truly
concurrent execution and drag multiprocessing performance down towards the
level of a uniprocessor system.
VxWorks SMP therefore provides specialized mechanisms for mutual exclusion
between tasks and interrupts executing and being received (respectively)
simultaneously on different CPUs. In place of uniprocessor task and interrupt
locking routines—such as taskLock( ) and intLock( )—VxWorks SMP provides
spinlocks, atomic memory operations, and CPU-specific mutual exclusion
facilities.
CPU Affinity
By default, any task can run on any of the CPUs in the system (which generally
provides the best load balancing) and interrupts are routed to CPU 0 (the bootstrap
CPU). There are instances, however, in which it is useful to assign specific tasks or
interrupts to a specific CPU. VxWorks SMP provides this capability, which is
referred to a as CPU affinity.
15.2.3 VxWorks SMP Hardware
The hardware required for use with VxWorks SMP must consist of symmetric
multiprocessors—either multicore processors or hardware systems with multiple
single CPUs. The processors must be identical, all memory must be shared
between the CPUs (none may be local to a CPU), and all devices must be equally
accessible from all CPUs.That is, targets for VxWorks SMP must adhere to the
uniform memory access (UMA) architecture.
Figure 15-1 illustrates the typical target hardware for a dual CPU SMP system.
669
15
VxWorks
Kernel Programmer's Guide, 6.6
Figure 15-1
SMP Hardware
Devices
Interrupts
Programmable
Interrupt Controller
Interrupts
Interrupts
CPU 0
CPU 1
Cache
Cache
Snooping
High Bandwidth Bus
Shared Memory
670
15 VxWorks SMP
15.2 Technology Overview
Regardless of the number of CPUs (typically 2, 4 or 8) in an SMP system, the
important characteristics are the same:
■
Each CPU accesses the very same physical memory subsystem; there is no
memory local to a CPU. This means it is irrelevant which CPU executes code.
■
Each CPU has its own memory management unit that allows concurrent
execution of tasks with different virtual memory contexts. For example, CPU
0 can execute a task in RTP 1 while CPU 1 executes a task in RTP 2.
■
Each CPU has access to all devices. Interrupts from these devices can be routed
to any one of the CPUs through an a programmable interrupt controller. This
means that it is irrelevant which CPU executes interrupt service routines (ISRs)
when handling interrupts.
■
Tasks and ISRs can be synchronized across CPUs and mutual exclusion
enforced by using spinlocks.
■
Bus snooping logic ensures the data caches between CPUs are always
coherent. This means that the operating system does not normally need to
perform special data cache operations order to maintain coherent caches.
However, this implies that only memory access attributes that allow bus
snooping are used in the system. Restrictions in terms of memory access
modes allowed in an SMP system, if any, are specific to a hardware
architecture.
15.2.4 Comparison of VxWorks SMP and AMP
The features of VxWorks SMP may be highlighted by comparison with the way
VxWorks is used in asymmetric multiprocessing (AMP), using the same target
hardware in both cases. VxWorks AMP technologies include VxMP, TIPC (over
shared memory), and distributed shared memory (DSHM). The relationship
between CPUs and basic uses of memory in SMP and AMP systems are illustrated
in Figure 15-2 and Figure 15-3.
671
15
VxWorks
Kernel Programmer's Guide, 6.6
Figure 15-2
SMP System
Memory
VxWorks
RTP A
CPU 0
RTP B
Memory shared between CPUs
Figure 15-3
CPU 1
Access
AMP System
Memory
VxWorks A
VxWorks B
CPU 0
RTP X
CPU 1
RTP Y
Shared
Memory
Memory shared between CPUs
672
Access
15 VxWorks SMP
15.2 Technology Overview
In an SMP configuration the entire physical memory space is shared between the
CPUs. This memory space is used to store a single VxWorks SMP image (text, data,
bss, heap). It is also used to store any real-time processes (RTPs) that are created
during the lifetime of the system. Because both CPUs can potentially read from,
write to and execute any memory location, any kernel task or user (RTP) task can
be executed by either CPU.
In an AMP configuration there is one copy of the VxWorks image in memory for
each CPU. Each operating system image can only be accessed by the CPU to which
it belongs. It is therefore impossible for CPU 1 to execute kernel tasks residing in
VxWorks CPU 0's memory, or the reverse. The same situation applies for RTPs. An
RTP can only be accessed and executed by the instance of VxWorks from which it
was started.
In an AMP system some memory is shared, but typically the sharing is restricted
to reading and writing data. For example, for passing messages between two
instances of VxWorks. Hardware resources are mostly divided between instances
of the operating system, so that coordination between CPUs is only required when
accessing shared memory.
With an SMP system, both memory and devices are shared between CPUs, which
requires coordination within the operating system to prevent concurrent access to
shared resources.
15
673
VxWorks
Kernel Programmer's Guide, 6.6
15.3 VxWorks SMP Configuration and Build
VxWorks SMP can be configured and built using the standard Wind River
Workbench and vxprj facilities. VxWorks SMP images cannot be created with the
legacy BSP directory configuration and build method (using config.h and make).
Workbench provides a VxWorks Image Project (VIP) option for creating VxWorks
SMP projects. The vxprj provides the -smp option for configuring and building
VxWorks SMP from the command line. For example:
vxprj create -smp hpcNet8641 diab
!
CAUTION: VxWorks SMP does not support MMU-less configurations.
!
CAUTION: Boot loaders for VxWorks SMP must not be built with the SMP build
option—neither with Workbench nor with vxprj. For more information about boot
loaders for VxWorks SMP, see 15.4 Booting VxWorks SMP, p.676.
Default VxWorks SMP Images
Default VxWorks SMP images are provided in project directories parallel to those
for VxWorks UP images. For example, for the hpcNet8641 BSP, the directories are
as follows:
■
installDir/vxworks-6.x/target/proj/hpcNet8641_diab_smp
■
installDir/vxworks-6.x/target/proj/hpcNet8641_gnu_smp
■
installDir/vxworks-6.x/target/proj/hpcNet8641_diab
■
installDir/vxworks-6.x/target/proj/hpcNet8641_gnu
Debug Versions of Spinlock Components
The INCLUDE_SPINLOCK_DEBUG component provides versions of spinlocks that
are useful for debugging SMP applications. By default the standard
INCLUDE_SPINLOCK component is included in VxWorks SMP; if
INCLUDE_SPINLOCK_DEBUG is included, it is removed automatically. For more
information, see Debug Versions of Spinlocks, p.681.
CPU Configuration Parameters
There are several configuration parameters that are specific to VxWorks SMP,
which are provided by the INCLUDE_KERNEL component. These parameters are
as follows:
674
15 VxWorks SMP
15.3 VxWorks SMP Configuration and Build
VX_SMP_NUM_CPUS
Defines the number of CPUs that should be enabled for VxWorks SMP. The
maximum number of CPUs for each architecture is as follows: ARM = 4,
IA32 = 8, MIPS = 16, PowerPC = 8, VxWorks Simulator = 32.
ENABLE_ALL_CPUS
Enables all CPUs that have been configured for the system’s use with
VX_SMP_NUM_CPUS. The default is TRUE, in which case VxWorks boots with
all CPUs enabled and running. The parameter can be set to FALSE for
debugging purposes, in which case only CPU 0 (the bootstrap CPU) will be
enabled by the VxWorks initialization code. The kernelCpuEnable( ) routine
can then be used to enable a specific CPU once the system has booted.
VX_ENABLE_CPU_TIMEOUT
The time-out value (in seconds) for the period during which additional cores
may be enabled. When kernelCpuEnable( ) is called, it waits for the time
defined by VX_ENABLE_CPU_TIMEOUT for the additional core to come up. If
ENABLE_ALL_CPUS is set to TRUE, the value of VX_ENABLE_CPU_TIMEOUT
is used as the time-out period for enabling all CPUs.
Idle Task Configuration Parameters
The INCLUDE_PROTECT_IDLE_TASK_STACK component provides a set of
parameters for configuring the exception stack for the CPU idle task. For
information about idle tasks, see CPU Idle Tasks, p.678.
IDLE_TASK_EXCEPTION_STACK_SIZE
Size (in bytes) of the idle tasks’ exception stacks.
IDLE_TASK_EXC_STACK_OVERFLOW_SIZE
Size (in bytes) of the overflow protection area adjacent to the idle task's
exception stack.
IDLE_TASK_EXC_STACK_UNDERFLOW_SIZE.
Size (in bytes) of the underflow protection area adjacent to the idle task's
exception stack.
675
15
VxWorks
Kernel Programmer's Guide, 6.6
15.4 Booting VxWorks SMP
Booting VxWorks SMP is essentially the same operation as booting VxWorks UP.
The boot loader is simply responsible for booting the bootstrap CPU (referred to as
CPU 0). The boot loader has no knowledge of any other CPUs. Once the VxWorks
SMP image is loaded on CPU 0 and started, that instance of the operating system
enables the other CPUs in the system.
VxWorks SMP can, however, be configured so that the image loaded onto CPU 0
does not automatically enable the other CPUs, which can then be enabled
interactively or programmatically. For more information in this regard, see
ENABLE_ALL_CPUS, p.675).
!
CAUTION: Boot loaders for VxWorks SMP must not be built with the SMP build
option—neither with the SMP selection for a Workbench VxWorks Image Project
(VIP), nor with the -smp option for vxprj. Boot loaders built with the SMP build
option will not function properly.
For detailed information about VxWorks boot loaders, see 3. Boot Loader.
15.5 Programming for VxWorks SMP
Programming for VxWorks SMP and VxWorks UP is in many respects the same.
With few exceptions, the SMP and uniprocessor (UP) configurations of VxWorks
share the same API—the difference amounts to only a few routines. There is binary
compatibility for both kernel and RTP applications between UP and SMP
configurations of VxWorks (for the same VxWorks release), as long as the
applications are based on the subset of APIs used by VxWorks SMP. A few
uniprocessor APIs are not suitable for an SMP system, and they are therefore not
provided. Similarly, SMP-specific APIs are not relevant to a uniprocessor system—
but default to appropriate uniprocessor behaviors (such as task spinlocks
defaulting to task locking), or have no effect.
However, because of the nature of SMP systems, SMP programming requires
special attention to the mechanisms of mutual exclusion, and to design
considerations that allow for full exploitation of the capabilities of a
676
15 VxWorks SMP
15.5 Programming for VxWorks SMP
multiprocessing system. Also note that VxWorks SMP maintains an idle task for
each CPU, and that idle tasks must not be interfered with.
SMP and Mutual Exclusion
The use of mutual exclusion facilities is one of the critical differences between
uniprocessor and SMP programming. While some facilities are the same for
VxWorks UP and VxWorks SMP, others are necessarily different. In addition,
reliance on implicit synchronization techniques—such as relying on task priority
instead of explicit locking—do not work in an SMP system (for more information
on this topic, see 15.15.4 Implicit Synchronization of Tasks, p.706).
Unlike uniprocessor systems, SMP systems allow for truly concurrent execution,
in which multiple tasks may execute, and multiple interrupts may be received and
serviced, all at the same time. In most cases, the same mechanisms—semaphores,
message queues, and so on—can be used in both uniprocessor and SMP systems
for mutual exclusion and coordination of tasks (see 4.8 Intertask and Interprocess
Communication, p.193).
However, the specialized uniprocessor mechanisms for disabling (masking)
interrupts and for suspending task preemption in order to protect critical regions
are inappropriate for—and not available in—an SMP system. This is because they
would defeat the advantages of truly concurrent execution by enforcing masking
or preemption across all CPUs, and thus drag a multiprocessing system down
towards the performance level of uniprocessor system.
The most basic differences for SMP programming therefore have to do with the
mechanisms available for mutual exclusion between tasks and interrupts
executing and being received (respectively) on different CPUs. In place of
uniprocessor task and interrupt locking routines—such as taskLock( ) and
intLock( )—VxWorks SMP provides the following facilities:
■
spinlocks for tasks and ISRs
■
CPU-specific mutual exclusion for tasks and ISRs
■
atomic memory operations
■
memory barriers
As with the uniprocessor mechanisms used for protecting critical regions,
spinlocks and CPU-specific mutual exclusion facilities should only used when
they are guaranteed to be in effect for very short periods of time. The appropriate
use of these facilities is critical to making an application SMP-ready (see
15.2.1 Terminology, p.667).
677
15
VxWorks
Kernel Programmer's Guide, 6.6
Note that both spinlocks and semaphores provide full memory barriers (in
addition to the memory barrier macros themselves).
For more information about these topics, see 15.6 Spinlocks for Mutual Exclusion and
Synchronization, p.679, 15.7 CPU-Specific Mutual Exclusion, p.685, 15.8 Memory
Barriers, p.687, and 15.9 Atomic Memory Operations, p.690.
CPU Affinity for Interrupts and Tasks
By default, any task can run on any of the CPUs in the system (which generally
provides the best load balancing) and interrupts are routed to CPU 0. There are
cases, however, in which it is useful to assign tasks or interrupts to a specific CPU.
VxWorks SMP provides this capability, which is referred to a as CPU affinity.
For more information about interrupt and CPU affinity, see 15.10 CPU Affinity,
p.691.
CPU Idle Tasks
VxWorks SMP includes a per-CPU idle task that does not exist in VxWorks UP.
The idle task has the lowest priority in the system, below the range permitted for
application use (for more information, see 4.3.1 Task Priorities, p.166). Idle tasks
make an SMP system more efficient by providing task context when a CPU enters
and exits an idle state.
The existence of idle tasks does not affect the ability of a CPU to go to sleep (when
power management is enabled) if there is no work to perform. Do not perform any
operations that affect the execution of an idle task.
The kernelIsCpuIdle( ) and kernelIsSystemIdle( ) routines provide information
about whether a specific CPU is executing an idle task, or whether all CPUs are
executing idle tasks (respectively).
For information about configuration options for idle tasks, see 15.3 VxWorks SMP
Configuration and Build, p.674.
!
WARNING: Do not suspend, stop, change the priority, attempt a task trace, or any
similar operations on an idle task. Deleting, suspending, or stopping an idle task
causes the system to crash due to an exception in the scheduler. Changing the
priority of an idle task to a higher priority puts the CPU into a low power mode
prematurely. That is, simply do not use the task ID (tid) of an idle task as a
parameter to any VxWorks routine except taskShow( ).
678
15 VxWorks SMP
15.6 Spinlocks for Mutual Exclusion and Synchronization
RTP Applications
As in VxWorks UP systems, RTP (user mode) applications have a more limited set
of mutual exclusion and synchronization mechanisms available to them than
kernel code or kernel applications. In VxWorks SMP, they can make use of
semaphores and atomic operations, but not spinlocks, memory barriers, or
CPU-specific mutual exclusion mechanisms. In addition, the semExchange( )
routine provides for an atomic give and exchange of semaphores.
Optimization for SMP
Using the appropriate facilities in the appropriate manner alone allows an
application to execute properly on an SMP system, but does not necessarily take
full advantage of the multiprocessing capabilities of the hardware. In order to do
so, the design of the application must be geared to exploiting the advantages
offered by SMP.
For information in this regard, see 15.13 Optimizing SMP Performance, p.699.
15.6 Spinlocks for Mutual Exclusion and Synchronization
Spinlocks provide a facility for short-term mutual exclusion and synchronization
in an SMP system. Spinlocks must be explicitly taken and released. While
semaphores can also be used for mutual exclusion and synchronization, spinlocks
are designed for use in situations comparable to those in which taskLock( ) and
intLock( ) are used in VxWorks UP. Semaphores should be used in an SMP system
for the same purposes as in a uniprocessor system. (Note that both spinlocks and
semaphores provide full memory barriers.)
One of the unique characteristics of VxWorks spinlocks is that they are
implemented with algorithms that ensure that they are fair, meaning that they are
deterministic in the time between the request and take, and they operate in as close
to FIFO order as possible. (For information on the different types of spinlocks, see
Types of Spinlocks, p.680.)
For information about why uniprocessor mechanisms are not supported on
VxWorks SMP for interrupt locking and suspension of task preemption, and SMP
alternatives, see 15.15.5 Synchronization and Mutual Exclusion Facilities, p.707,
Interrupt Locking: intLock( ) and intUnlock( ), p.710, Task Locking: taskLock( ) and
679
15
VxWorks
Kernel Programmer's Guide, 6.6
taskUnlock( ), p.711, and Task Locking in RTPs: taskRtpLock( ) and taskRtpUnlock( ),
p.711.
NOTE: Spinlocks are not available to RTP (user-mode) applications.
Spinlocks as Full Memory Barriers
VxWorks spinlocks operate as full memory barriers between acquisition and
release. A full memory barrier forces both read and write memory access
operations to be performed in strict order. The process of updating data structures
is therefore fully completed between the time a spinlock is acquired and released.
Types of Spinlocks
VxWorks SMP provides two types of spinlocks:
■
ISR-callable spinlocks, which are used to address contention between ISRs—or
between a task and other tasks and ISRs. They disable (mask) interrupts on the
local CPU. When called by tasks they suspend task preemption on the local
CPU as well.
■
Task-only spinlocks, which are used to address contention between tasks (and
not ISRs). They suspend task preemption on the local CPU.
The local CPU is the one on which the spinlock call is performed. For detailed
information about spinlocks, see 15.6.1 ISR-Callable Spinlocks, p.682 and
15.6.2 Task-Only Spinlocks, p.682.
Spinlock Behavior and Usage Guidelines
Unlike the behavior associated with semaphores, a task that attempts to take a
spinlock that is already held by another task does not pend; instead it continues
executing, simply spinning in a tight loop waiting for the spinlock to be freed.
The terms spinning and busy waiting—which are both used to describe this
activity—provide insight into both the advantages and disadvantages of
spinlocks. Because a task (or ISR) continues execution while attempting to take a
spinlock, the overhead of rescheduling and context switching can be avoided
(which is not the case with a semaphore). On the other hand, spinning does no
useful work, and ties up one or more of the CPUs.
Spinlocks should therefore only be used when they are likely to be efficient; that is,
when they are going to be held for very short periods of time (as with taskLock( )
and intLock( ) in a uniprocessor system). If a spinlock is held for a long period of
680
15 VxWorks SMP
15.6 Spinlocks for Mutual Exclusion and Synchronization
time, the drawbacks are similar to intLock( ) and taskLock( ) being held for a long
time in VxWorks UP—increased interrupt and task latency.
Acquisition of a spinlock on one CPU does not affect the processing of interrupts
or scheduling of tasks on other CPUs. Tasks cannot be deleted while they hold a
spinlock.
For detailed cautionary information about spinlock use, see 15.6.3 Caveats With
Regard to Spinlock Use, p.683 and 15.6.4 Routines Restricted by Spinlock Use, p.683.
Debug Versions of Spinlocks
The debug version of VxWorks spinlocks (provided with the
INCLUDE_SPINLOCK_DEBUG component) is designed for use while developing
applications that use spinlocks. It allows for catching violations of guidelines for
appropriate spinlock use (for information in this regard, see 15.6.3 Caveats With
Regard to Spinlock Use, p.683).
The following is a list of checks that are performed by the debug version of
spinlocks:
■
■
■
■
task-only take with spinLockTaskTake( )
■
Calling from an ISR context results in an error.
■
Recursive taking of a spinlock results in an error.
■
Nested taking of spinlocks results in an error.
15
task-only give with spinLockTaskGive( )
■
Calling from an ISR context results in an error.
■
Attempting to give up a spinlock without first acquiring a spinlock results
in an error.
ISR-only take with spinLockIsrTake( )
■
Recursive taking of a spinlock results in an error.
■
Nested taking of spinlocks result in an error.
ISR-only give with spinLockIsrGive( )
■
Attempting to give up a spinlock without first acquiring a spinlock results
in an error.
Errors are handled by the error detection and reporting facility, in the form of a
fatal kernel error with an appropriate error string (for information about the error
detection and reporting facility, see 11. Error Detection and Reporting).
681
VxWorks
Kernel Programmer's Guide, 6.6
15.6.1 ISR-Callable Spinlocks
Spinlocks that are used to address contention between ISRs—or between a task
and other tasks and ISRs—are referred to as ISR-callable spinlocks.
These spinlocks can be acquired by both tasks and ISRs. They disable (mask)
interrupts on the local CPU, which prevents the caller from being preempted while
it holds the spinlock (which could otherwise lead to a livelock). If a task acquires
an ISR-callable spinlock, task preemption is also suspended on the local CPU while
that task holds the spinlock. This allows the task to execute the critical section that
the spinlock is protecting. Interrupts and tasks on other CPUs are not affected. The
routines used for ISR-callable spinlocks are listed in Table 15-1.
For VxWorks UP, ISR-callable spinlocks are implemented with the same behavior
as the interrupt locking routines intLock( ) and intUnlock( ).
Table 15-1
ISR-Callable Spinlock Routines
Routine
Description
spinLockIsrInit( )
Initializes an ISR-callable spinlock.
spinLockIsrTake( )
Acquires an ISR-callable spinlock.
spinLockIsrGive( )
Relinquishes ownership of an ISR-callable
spinlock.
15.6.2 Task-Only Spinlocks
Spinlocks that are used to address contention between tasks alone (and not ISRs)
are called task-only spinlocks. These spinlocks disable task preemption on the local
CPU while the caller holds the lock (which could otherwise lead to a livelock
situation). This prevents the caller from being preempted by other tasks and allows
it to execute the critical section that the lock is protecting. Interrupts are not
disabled and task preemption on other CPUs is not affected. The routines used for
task-only spinlocks are listed in Table 15-2.
For VxWorks UP, task-only spinlocks are implemented with the same behavior as
the task locking routines taskLock( ) and taskUnlock( ).
682
15 VxWorks SMP
15.6 Spinlocks for Mutual Exclusion and Synchronization
Table 15-2
Task-Only Spinlock Routines
Routine
Description
spinLockTaskInit( )
Initializes a task-only spinlock.
spinLockTaskTake( )
Acquires a task-only spinlock.
spinLockTaskGive( )
Relinquishes ownership of a task-only spinlock.
15.6.3 Caveats With Regard to Spinlock Use
Because of the concurrency of execution inherent in SMP systems, spinlocks
should be used with care. The following prescriptions should be adhered to avoid
problems with spinlocks:
■
A spinlock should only be held for a short and deterministic period of time.
■
A task or ISR must not take more than one spinlock at a time. Livelocks may
result when an entity that already holds a spinlock takes another spinlock.
Livelocks are similar to the deadlocks that occur with semaphore use. With
spinlocks, however, the entity does not pend or block; it spins without ever
acquiring the spinlock and the CPU appears to be hung. Because interrupts are
masked or task preemption is disabled, the state cannot be remedied.
■
A task or ISR must not take a spinlock that it already holds. That is, recursive
takes of a spinlock should not be made. A livelock will occur.
■
In order to prevent a task or ISR from entering a kernel critical region while it
already holds a spinlock—and cause the system to enter a livelock state—a
task or ISR must not call specified routines while it holds a spinlock. The
VxWorks SMP kernel itself uses spinlocks to protect its critical regions. For
information about these routines, see 15.6.4 Routines Restricted by Spinlock Use,
p.683.
A debug version of spinlocks can be used to catch these problems. For information,
see Debug Versions of Spinlocks, p.681.
15.6.4 Routines Restricted by Spinlock Use
Certain routines should not be called while the calling entity (task or ISR) holds a
spinlock. This restriction serves to prevent a task or ISR from entering a kernel
critical region while it already holds a spinlock—and cause the system to enter a
683
15
VxWorks
Kernel Programmer's Guide, 6.6
livelock state (for more information, see 15.6.3 Caveats With Regard to Spinlock Use,
p.683). The routine restriction also apply to intCpuLock( ) (for more information
about this routine see 15.7.1 CPU-Specific Mutual Exclusion for Interrupts, p.685).
This restriction applies because the kernel requires interrupts to be enabled to
implement its multi-CPU scheduling algorithm.
It is outside the scope of this document to list all the VxWorks spinlock restricted
routines. However, generally speaking these are routines related to the creation,
destruction and manipulation of kernel objects (semaphores, tasks, message
queues, and so on) as well as any routine that can cause a scheduling event.
While the restriction imposed by spinlock use may seem to be a hindrance, it really
should not be. Spinlocks are meant for very fast synchronization between
processors. Holding a spinlock and attempting to perform notable amounts of
work, including calling into the kernel, results in poor performance on an SMP
system, because either task preemption or interrupts, or both, are disabled when a
CPU owns a spinlock.
Table 16 identifies some of the routines restricted by spinlock and CPU lock use.
Table 16
Routines Restricted by Spinlock and CPU Lock Use
Library
Routines
taskLib
taskExit( ), taskDelete( ), taskDeleteForce( ), taskInitExcStk( ),
taskUnsafe( ), exit( ), taskSuspend( ), taskResume( ),
taskPrioritySet( ), taskDelay( ), taskStackAllot( ), taskRestart( ),
taskCpuLock( ), taskCpuUnlock( ), taskCreateLibInit( ),
taskCreate( ), taskActivate( ), taskCpuAffinitySet( ),
taskCpuAffinityGet( ), taskSpawn( ), taskInit( )
msgQLib
msgQCreate( ), msgQDelete( ), msgQSend( ), msgQReceive( ),
msgQInitialize( ), msgQNumMsgs( ),
msgQEvLib
msgQEvStart( ), msgQEvStop( )
semLib
semTake( ), semGive( ), semFlush( ), semDelete( )
semBLib
semBInitialize( ), semBCreate( )
semCLib
semCInitialize( ), semCCreate( )
semMLib
semMInitialize( ), semMGiveForce( ), semMCreate( )
semEvLib
semEvStart( ), semEvStop( )
684
15 VxWorks SMP
15.7 CPU-Specific Mutual Exclusion
Table 16
Routines Restricted by Spinlock and CPU Lock Use
Library
Routines
wdLib
wdCreate( ), wdDelete( ), wdinitialise( ), wdStart( ),
wdCancel( )
kernelLib
kernelTimeSlice( ), kernelCpuEnable( )
intLib
intDisconnect( )
intArchLib
intConnect( ), intHandlerCreate( ), intVecTableWriteProtect( )
eventLib
eventSend( ), eventReceive( )
15.7 CPU-Specific Mutual Exclusion
VxWorks SMP provides facilities for CPU-specific mutual exclusion, that is for
mutual exclusion operations whose scope is entirely restricted to the CPU on
which the call is made (the local CPU). These facilities are designed to facilitate
porting uniprocessor code to an SMP system.
15.7.1 CPU-Specific Mutual Exclusion for Interrupts
CPU-specific mutual exclusion for interrupts allows for disabling (masking)
interrupts on the CPU on which the calling task or ISR is running. For example if
task A, running on CPU 0, performs a local CPU interrupt lock operation, no
interrupts can be processed by CPU 0 until the lock is released by task A.
Execution of interrupts on other CPUs in the SMP system is not affected. In order
to be an effective means of mutual exclusion, therefore, all tasks and ISRs that
should participate in the mutual exclusion scenario should have CPU affinity set
for the local CPU (for information, see 15.10.1 Task CPU Affinity, p.691).
Note that some routines should not be used if the calling task or ISR has locked
interrupts on the local CPU—similar to the case of holding spinlocks (see
15.6.3 Caveats With Regard to Spinlock Use, p.683). The restricted routines are
described in 15.6.4 Routines Restricted by Spinlock Use, p.683.
685
15
VxWorks
Kernel Programmer's Guide, 6.6
The routines listed in Table 15-3 are used for disabling and enabling interrupts on
the local CPU.
Note that in a uniprocessor system they default to the behavior of intLock( ) and
intUnlock( ).
Table 15-3
CPU-Specific Mutual Exclusion Routines for Interrupts
Routine
Description
intCpuLock( )
Disables interrupts on the CPU on which the
calling task or ISR is running.
intCpuUnlock( )
Enables interrupts on the CPU on which the
calling task or ISR is running.
For more information about these routines, see the intLib entry in the VxWorks
API references.
15.7.2 CPU-Specific Mutual Exclusion for Tasks
CPU-specific mutual exclusion for tasks allows for suspending task preemption on
the CPU on which the calling task is running. That is, it provides for local CPU task
locking, and effectively prevents any other task from running on the local CPU. For
example, task A running on CPU 0 can perform a local CPU task lock operation so
that no other task can run on CPU 0 until it releases the lock or makes a blocking
call.
The calling task is also prevented from migrating to another CPU until the lock is
released.
Execution on other CPUs in the SMP system is not affected. In order to be an
effective means of mutual exclusion, therefore, all tasks that should participate in
the mutual exclusion scenario should have CPU affinity set for the local CPU (for
information, see 15.10.1 Task CPU Affinity, p.691).
The routines listed in Table 15-4 are used for suspending and resuming task
preemption on the local CPU.
Note that in a uniprocessor system they default to the behavior of taskLock( ) and
taskUnlock( ).
686
15 VxWorks SMP
15.8 Memory Barriers
Table 15-4
CPU-Specific Mutual Exclusion Routines for Tasks
Routine
Description
taskCpuLock( )
Disables task preemption for the CPU on which
the calling task is running.
taskCpuUnlock( )
Enables context task switching on the CPU on
which the calling task is running.
For more information about these routines, see the taskLib entry in the VxWorks
API references.
15.8 Memory Barriers
In modern multiprocessing architectures, individual CPUs can reorder both read
and write operations in order to improve overall system efficiency. From the
perspective of a single CPU in the system, this reordering is completely
transparent because the CPU ensures that any read operation gets data that was
previously written, regardless of the order in which the read and write operations
are actually committed to system memory. The reordering occurs in the
background, and is never visible to the programmer.
In an multiprocessor system, an individual CPU can execute a series of write
operations to memory, and these write operations can be queued between the CPU
and system memory. The CPU is allowed to commit these queued operations to
system memory in any order, regardless of the order in which the operations arrive
in the CPU's write queue. Similarly, a CPU is free to issue more than one read
operation in parallel, whether as the result of speculative execution, or because the
program has requested more than one independent read operation.
Because of this reordering, two tasks that share data should never assume that the
order in which an operation is performed on one CPU will be the same as the order
in which the operations are written to or read from memory. A classic example of
this ordering problem involves two CPUs, in which one CPU prepares an item of
work to be performed, and then sets a boolean flag to announce the availability of
the work unit to a second CPU that is waiting for it. The code in this case would
look like the following:
687
15
VxWorks
Kernel Programmer's Guide, 6.6
/* CPU 0 - announce the availability of work */
pWork = &work_item;
workAvailable = 1;
/* store pointer to work item to be performed */
/* CPU 1 - wait for work to be performed */
while (!workAvailable);
doWork (pWork);
/* error - pWork might not be visible to this CPU yet */
It is very likely that the pWork pointer used by CPU 1 will contain incorrect data
because CPU 0 reorders its write operations to system memory, which causes CPU
1 to observe the change to the workAvailable variable before the value of the
pWork variable has been updated. In a case like this, the likely result is a system
crash due to de-referencing an invalid pointer.
To solve the memory ordering problem, VxWorks provides a set of memory
barrier operations. The sole purpose of memory barrier operations is to provide a
way to guarantee the ordering of operations between cooperating CPUs. Memory
barriers fall into three general classes:
■
■
■
read memory barrier
write memory barrier
full (read/write) memory barrier
NOTE: VxWorks SMP provides a set of synchronization primitives to protect
access to shared resources. These primitives include semaphores, message queues,
and spinlocks. These primitives include full memory barrier functionality.
Additional memory barrier operations are not required with these facilities to
protect shared resources.
NOTE: Memory barriers are not available to RTP (user-mode) applications.
Read Memory Barrier
The VX_MEM_BARRIER_R( ) macro provides a read memory barrier.
VX_MEM_BARRIER_R( ) enforces ordering between all of the read operations
that have occurred prior to the barrier, and all of the read operations that occur
after the barrier. Without this barrier, a CPU is free to reorder its pending read
operations in any way that does not affect program correctness from a
uniprocessor perspective. For example, a CPU is free to reorder the following
independent reads:
a = *pAvalue;
b = *pBvalue;
688
/* read may occur _after_ read of *pBvalue */
/* read may occur _before read of *pAValue */
15 VxWorks SMP
15.8 Memory Barriers
By inserting a memory barrier between the read operations, you can guarantee
that the reads occur in the appropriate order:
a = *pAvalue;
VX_MEM_BARRIER_R();
b = *pBvalue;
/* will occur before read of *pBvalue */
/* will occur after read of *pAvalue */
While VX_MEM_BARRIER_R( ) can ensure that the read operations occur in the
correct order, this guarantee is not helpful unless the writer of the shared data also
ensures that the writes of the shared data also occur in the correct order. For this
reason, the VX_MEM_BARRIER_R( ) and VX_MEM_BARRIER_W( ) macros
should always be used together.
Write Memory Barrier
The VX_MEM_BARRIER_W( ) macro provides a write memory barrier.
VX_MEM_BARRIER_W( ) enforces the ordering between all of the write
operations that have occurred prior to the barrier, and all of the write operations
that occur after the barrier. The following code fragment is taken from a preceding
example, but modified to take advantage of a memory barrier:
pWork = &work_item;
VX_MEM_BARRIER_W();
workAvailable = 1;
15
Inserting a barrier between the update of *pWork and the update of
workAvailable ensures that the value of workAvailable in system memory is
updated after the value of pWork has been updated in system memory. Note that
VX_MEM_BARRIER_W( ) does not actually force the writing of these values to
system memory. Instead, it merely enforces the order in which these values are
written. Note that VX_MEM_BARRIER_W( ) should always be used with
VX_MEM_BARRIER_R( ) or VX_MEM_BARRIER_RW( ).
Read/Write Memory Barrier
The VX_MEM_BARRIER_RW( ) macro provides a read/write memory barrier.
This is also referred to as a full fence memory barrier. VX_MEM_BARRIER_RW( )
combines the effects of both the VX_MEM_BARRIER_R( ) and
VX_MEM_BARRIER_W( ) operations into a single primitive. On some systems,
VX_MEM_BARRIER_RW( ) may be substantially more expensive than either
VX_MEM_BARRIER_R( ) or VX_MEM_BARRIER_W( ). Unless both read and
write ordering is required, Wind River does not recommend the use of
VX_MEM_BARRIER_RW( ).
689
VxWorks
Kernel Programmer's Guide, 6.6
15.9 Atomic Memory Operations
Atomic operations make use of CPU support for atomically accessing memory.
They combine a set of (architecture-specific) operations into what is effectively a
single operation that cannot be interrupted by any other operation on the memory
location in question. Atomic operations thereby provide mutual exclusion for a
simple set of operations (such as incrementing and decrementing variables).
Atomic operations can be useful as a simpler alternative to spinlocks, such as for
updating a single data element. For example, you can update the next pointer in a
singly-linked list from NULL to non-NULL (without interrupts locked) using an
atomic operation, which allows you to create lock-less algorithms.
Because the atomic operations are performed on a memory location supplied by
the caller, users must ensure the location has memory access attributes and an
alignment that allows atomic memory access—otherwise an access exception will
occur. Restrictions, if any, are specific to the CPU architecture. For more
information, see 15.15.9 Memory-Access Attributes, p.714.
The vxAtomicLib library provides a number of routines that perform atomic
operations. They are described in Table 15-5.
Atomic operation routines are available in user space (for RTP applications) as
well as in the kernel.
Table 15-5
Atomic Memory Operation Routines
Routine
Description
vxAtomicAdd( )
Adds two values atomically.
vxAtomicSub( )
Subtracts one value from another atomically.
vxAtomicInc( )
Increments a value atomically.
vxAtomicDec( )
Decrements a value atomically.
vxAtomicOr( )
Performs a bitwise OR operation on two values
atomically.
vxAtomicXor( )
Performs a bitwise XOR operation on two
values atomically.
vxAtomicAnd( )
Performs a bitwise AND operation on two
values atomically.
690
15 VxWorks SMP
15.10 CPU Affinity
Table 15-5
Atomic Memory Operation Routines
Routine
Description
vxAtomicNand( )
Performs a bitwise NAND operation on two
values atomically.
vxAtomicSet( )
Sets one value to another atomically.
vxAtomicClear( )
Clears a value atomically.
vxCas( )
Performs an atomic compare-and-swap of two
values atomically.
15.10 CPU Affinity
VxWorks SMP provides facilities for CPU affinity; that is, for assigning specific
interrupts or tasks to specific CPUs.
15
15.10.1 Task CPU Affinity
VxWorks SMP provides the ability to assign tasks to a specific CPU, after which
the scheduler ensures the tasks are only executed on that CPU. This assignment is
referred to as task CPU affinity.
While the default SMP operation in which any task can run on any CPU often
provides the best overall load balancing, there are cases in which assigning a
specific set of tasks to a specific CPU can be useful. For example, if a CPU is
dedicated to signal processing and does no other work, the cache remains filled
with the code and data required for that activity. This saves the cost of moving to
another CPU—which is incurred even within single piece of silicon, as the L1 cache
is bound to a single CPU, and the L1 must be refilled with new text and data if the
task migrates to a different CPU.
Another example is a case in which profiling an application reveals that some of its
tasks are frequently contending for the same spinlock, and a fair amount of
execution time is wasted waiting for a spinlock to become available. Overall
performance could be improved by setting task CPU affinity such that all tasks
691
VxWorks
Kernel Programmer's Guide, 6.6
involved in spinlock contention run on the same CPU. This would free up more
time other CPUs for other tasks.
Task CPU affinity can be set in the following manner:
■
A task can set its own CPU affinity or the CPU affinity of another task by
calling taskCpuAffinitySet( ).
■
A newly created task inherits the CPU affinity (if any) of the parent task. A task
created or initialized by any of the following routines inherits the CPU affinity
of the calling task: taskSpawn( ), taskCreate( ), taskInit( ), taskOpen( ), and
taskInitExcStk( ).
The creating task’s CPU affinity is not inherited, however, when the task that
is created is an RTPs initial task. For example, if a task invokes rtpSpawn( ),
the initialization task of the resulting RTP does not inherit the CPU affinity of
the caller.
The taskLib library provides routines for managing task CPU affinity. They are
described in Table 15-6.
Table 15-6
Task CPU Affinity Routines
Routine
Description
taskCpuAffinitySet( )
Sets the CPU affinity for a task.
taskCpuAffinityGet( )
Returns the CPU affinity for a task.
The routine taskCpuAffinitySet( ) takes a CPU set variable (of type cpuset_t) to
identify the CPU to which the task should be assigned. Similarly, the
taskCpuAffinityGet( ) routine takes a pointer to a cpuset_t variable for the
purpose of recording the CPU affinity for a given task.
In both cases the CPUSET_ZERO( ) macro must be used to clear the cpuset_t
variable before the call is made. For taskCpuAffinitySet( ), the CPUSET_SET( )
macro must be used after CPUSET_ZERO( ) and before the routine itself is called.
To remove task CPU affinity, use the CPUSET_ZERO( ) macro to clear the
cpuset_t variable, and then make the taskCpuAffinitySet( ) call again.
For more information about using these routines and macros see Task CPU Affinity
Examples, p.693 and CPU Set Variables and Macros, p.696
692
15 VxWorks SMP
15.10 CPU Affinity
RTP Tasks and CPU Affinity
By default, real-time process (RTP) tasks inherit the CPU affinity setting of the task
that created the RTP. If the parent task has no specific CPU affinity (that is, it can
execute on any available CPU and may migrate from one CPU to the other during
its lifetime), then the RTP's tasks have no specific CPU affinity either. If the parent
task has its affinity set to a given CPU, then by default, the RTP tasks inherit this
affinity and execute only on the same CPU as the parent task.
The RTP_CPU_AFFINITY_NONE option for rtpSpawn( ) can be used to create an
RTP in which tasks have no CPU affinity, despite the fact that the RTP’s parent task
may have itself have had one.
Task CPU Affinity Examples
The following sample code illustrates the sequence to set the affinity of a newly
created task to CPU 1.
STATUS affinitySetExample (void)
{
cpuset_t affinity;
int tid;
/* Create the task but only activate it after setting its affinity */
tid = taskCreate ("myCpu1Task", 100, 0, 5000, printf,
(int) "myCpu1Task executed on CPU 1 !", 0, 0, 0,
0, 0, 0, 0, 0, 0);
if (tid == NULL)
return (ERROR);
/* Clear the affinity CPU set and set index for CPU 1 */
CPUSET_ZERO (affinity);
CPUSET_SET (affinity, 1);
if (taskCpuAffinitySet (tid, affinity) == ERROR)
{
/* Either CPUs are not enabled or we are in UP mode */
taskDelete (tid);
return (ERROR);
}
/* Now let the task run on CPU 1 */
taskActivate (tid);
return (OK);
}
The next example shows how a task can remove its affinity to a CPU:
693
15
VxWorks
Kernel Programmer's Guide, 6.6
{
cpuset_t affinity;
CPUSET_ZERO (affinity);
taskCpuAffinitySet (0, affinity);
}
15.10.2 Interrupt CPU Affinity
SMP hardware requires programmable interrupt controller devices (for more
information see 15.2.3 VxWorks SMP Hardware, p.669). VxWorks SMP makes use
of this hardware to allow assignment interrupts to a specific CPU. By default,
interrupts are routed to the bootstrap CPU (CPU 0).
Interrupt CPU affinity can be useful for load balancing (for example, if there is too
much total interrupt traffic for one CPU to handle). It can also be used as an aid in
migrating code from VxWorks UP (for more information, see Interrupt Locking:
intLock( ) and intUnlock( ), p.710).
Runtime assignment of interrupts to a specific CPU occurs at boot time, when the
system reads interrupt configuration information from the BSP. The interrupt
controller then receives a command directing that a given interrupt be routed to a
specific CPU. For information about the mechanism involved, see the VxWorks
Device Driver Developer's Guide.
15.11 CPU Information and Management
VxWorks SMP provides several routines and macros for getting and manipulating
information about CPUs, as well as for managing their operation.
CPU Information and Management Routines
The kernelLib and vxCpuLib libraries provide routines for getting information
about, and for managing, CPUs. They are described in Table 15-7 and Table 15-8.
694
15 VxWorks SMP
15.11 CPU Information and Management
Table 15-7
kernelLib CPU Routines
Routine
Description
kernelIsCpuIdle( )
Returns TRUE if the specified CPU is idle.
kernelIsSystemIdle( )
Returns TRUE if all enabled CPUs are idle
kernelCpuEnable( )
Enables the CPU with the specified index.
The kernelCpuEnable( ) routine allows you to enable a specific CPU. Once a CPU
is enabled, it starts dispatching tasks as directed by the scheduler. All CPUs are
enabled by default, but the ENABLE_ALL_CPUS component parameter can be used
to boot VxWorks SMP with just CPU 0 enabled (for more information see
ENABLE_ALL_CPUS, p.675). Then, kernelCpuEnable( ) can be used to selectively
enable individual CPUs.
Table 15-8
vxCpuLib CPU Routines
Routine
Description
vxCpuConfiguredGet( )
Returns the number of CPUs that have been
statically configured into the system with the
BSP.
vxCpuEnabledGet( )
Returns the set of CPUs that are enabled in the
system.
vxCpuIndexGet( )
Returns the mapped CPU number as provided by
the operating system.
vxCpuIdGet( )
Returns the true CPU number, as defined by the
architecture variant (and not the operating
system).
The vxCpuConfiguredGet( ) routine returns the number of CPUs configured into
a VxWorks SMP system with the BSP, which may not be the same as the number
of CPUs available in the hardware platform.
The vxCpuEnabledGet( ) routine returns the set of CPUs enabled (running) on the
system, of which the total may be different from the number of CPUs configured
into the system with the BSP, or available in the hardware platform. As noted
above, all CPUs are enabled by default, but the ENABLE_ALL_CPUS configuration
parameter can be set so that VxWorks SMP boots with just CPU 0 enabled (for
695
15
VxWorks
Kernel Programmer's Guide, 6.6
more information see ENABLE_ALL_CPUS, p.675). Then the kernelCpuEnable( )
routine can be used to selectively enable individual CPUs.
The return type used by vxCpuEnabledGet( ) to identify a CPU set is cpuset_t.
Note that the CPUSET_ZERO( ) macro must be used to clear the cpuset_t variable
before the vxCpuEnabledGet( ) call is made. For information about the cpuset_t
variable type, and the macros used to manipulate cpuset_t variables, see CPU Set
Variables and Macros, p.696.
The routine vxCpuIndexGet( ) returns the index of the calling task’s CPU. The
index of a CPU is a number between 0 and N-1, where N is the number of CPUs
configured into the SMP system. Note, however, that tasks can migrate from one
CPU to another (by default), so there is no guarantee that the index returned by
vxCpuIndexGet( ) identifies the CPU on which the task is running after the
return—unless the calling task is prevented from migrating to another CPU with
taskCpuLock( ) or intCpuLock( ).
CPU Set Variables and Macros
VxWorks SMP provides a CPU set variable type, and CPU set macros for
manipulating variables defined by that type. The variable and macros must be
used in conjunction with various routines—such as taskCpuAffinitySet( )—for
getting information about CPUs and managing their use.
The cpuset_t variable type is used for identifying the CPUs that have been
configured into a VxWorks SMP system with the target BSP, which may be a
subset of the CPUs in the hardware platform.
Each bit in a cpuset_t variable corresponds to a specific CPU, or CPU index, with
the first bit representing CPU 0 (the bootstrap CPU). The first bit corresponds to
index 0, the second to 1, the third to 2, and so on (regardless of the physical location
of the CPUs in the hardware).
As an example, for an eight CPU hardware system, for which the BSP configures
four CPUs for VxWorks SMP, the CPUSET_ZERO( ) macro would clear all the bits
in a cpuset_t variable, and then a call to vxCpuIndexGet( ) would set the first four.
CPU set macros must be used to set and unset CPU indices (change the bits of
cpuset_t variables). These macros are described in Table 15-9CPU Set Macros,
p.697. In order to use these macros, include the cpuset.h header file.
!
CAUTION: Do not manipulate cpuset_t type variables directly. Use CPU set
macros.
696
15 VxWorks SMP
15.11 CPU Information and Management
Table 15-9
CPU Set Macros
Macro
Description
CPUSET_SET( )
Sets a specific CPU index (one specific
cpuset_t variable bit).
CPUSET_SETALL( )
Sets CPU indices (all cpuset_t variable bits)
for all CPUs that are configured into the
system.
CPUSET_SETALL_BUT_SELF( ) Sets indices (cpuset_t variable bits) for all
CPUs that are configured into the system,
except for the index of the CPU on which the
macro is called.
CPUSET_CLR( )
Clears a specific CPU index (one specific
cpuset_t variable bit).
CPUSET_ZERO( )
Clears all CPU indices (all cpuset_t variable
bits).
CPUSET_ISSET( )
Returns TRUE if the specified index (cpuset_t
variable bit) is set in the cpuset_t variable.
CPUSET_ISZERO( )
Returns TRUE if the no indices (cpuset_t
variable bits) are set in the cpuset_t variable.
CPUSET_ATOMICSET( )
Atomically sets a specific CPU index (one
specific cpuset_t variable bit).
CPUSET_ATOMICCLR( )
Atomically clears a specific CPU index (one
specific cpuset_t variable bit).
For an example of how CPU set macros are used, see 15.10.1 Task CPU Affinity,
p.691. For more information about the macros, see the entry for cpuset in the
VxWorks API references.
697
15
VxWorks
Kernel Programmer's Guide, 6.6
15.12 Debugging SMP Code
Debugging and system monitoring tools such as System Viewer, and the
Workbench debugger provide support for debugging SMP code.
For information about debugging problems related to spinlock use, see Debug
Versions of Spinlocks, p.681.
For information about target tool behavior with VxWorks SMP, see 12. Target
Tools.
As appropriate, debugging facilities provide CPU-specific information. For
example, checkStack( ) displays the interrupt stack of all CPUs. The output from
the kernel shell looks like the following:
-> checkStack
NAME
ENTRY
------------ -----------tJobTask
0x60056a50
(Exception Stack)
tExcTask
0x60055c60
(Exception Stack)
tLogTask
logTask
(Exception Stack)
tNbioLog
0x60057720
(Exception Stack)
tShell0
shellTask
(Exception Stack)
tWdbTask
wdbTask
(Exception Stack)
tAioIoTask1 aioIoTask
(Exception Stack)
tAioIoTask0 aioIoTask
(Exception Stack)
tNet0
ipcomNetTask
(Exception Stack)
ipcom_syslog 0x60108350
(Exception Stack)
ipnetd
0x6010c630
(Exception Stack)
tAioWait
aioWaitTask
(Exception Stack)
tIdleTask0
idleTaskEntr
(Exception Stack)
tIdleTask1
idleTaskEntr
(Exception Stack)
INTERRUPT CPU 0
INTERRUPT CPU 1
value = 2 = 0x2
->
698
TID
SIZE
CUR HIGH MARGIN
---------- ----- ----- ----- -----0x603ce228 24576
104
832 23744
12072
0
0 12072
0x601a3b30 24576
184
520 24056
12216
0
0 12216
0x603d7c38 24576
228
564 24012
12216
0
0 12216
0x603d2010 24576
188
524 24052
12072
0
0 12072
0x6051ce90 81920 6780 9344 72576
12072
0
0 12072
0x603c7b88 24576
196
532 24044
12072
0
0 12072
0x60461740 40960
116
452 40508
12216
0
0 12216
0x60461a60 40960
116
452 40508
12216
0
0 12216
0x60473af0 24576
156
652 23924
12216
0
0 12216
0x60485558 16384
352 1092 15292
12216
0
0 12216
0x604a58c8 24576
128 2012 22564
12216
0
0 12216
0x60451020 40960
244
880 40080
12216
0
0 12216
0x60389a20 12288
396
900 11388
12200
0
772 11428
0x603a2000 12288
396
900 11388
12200
0 1156 11044
57344
0 1796 55548
57344
0 1272 56072
15 VxWorks SMP
15.13 Optimizing SMP Performance
And spy( ) reports the number of ticks spent in kernel, interrupt, idle, and task
code for each CPU. The output looks like the following:
-> spy
value = 1634761696 = 0x61707be0
->
NAME
ENTRY
TID
------------ ------------ ---------tJobTask
0x60056ae0
0x603d2010
tExcTask
0x60055cf0
0x601a3b30
tLogTask
logTask
0x603d7c38
tNbioLog
0x600577b0
0x603db110
tShell0
shellTask
0x6051cec8
tWdbTask
wdbTask
0x603c7840
tSpyTask
spyComTask
0x61707be0
tAioIoTask1 aioIoTask
0x60443888
tAioIoTask0 aioIoTask
0x60443c88
tNet0
ipcomNetTask 0x60485020
ipcom_syslog 0x60109060
0x60485c78
ipnetd
0x6010d340
0x603bf0c8
tAioWait
aioWaitTask 0x60443590
tIdleTask0
idleTaskEntr 0x60389a38
tIdleTask1
idleTaskEntr 0x603a2000
KERNEL
INTERRUPT
TOTAL
PRI
--0
0
0
0
1
3
5
50
50
50
50
50
51
287
287
total % (ticks)
--------------0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
100% (
1000)
100% (
1000)
0% (
0)
0% (
0)
200% (
1000)
delta % (ticks)
--------------0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
0% (
0)
100% (
500)
100% (
500)
0% (
0)
0% (
0)
200% (
500)
CPU
KERNEL
INTERRUPT
IDLE
TASK
TOTAL
--- --------------- --------------- --------------- --------------- ----0
0% (
0)
0% (
0) 100% (
1000)
0% (
0) 100%
1
0% (
0)
0% (
0) 100% (
1000)
0% (
0) 100%
Note that while timexLib can avoid precision errors by auto-calibrating itself and
doing several calls of the functions being monitored, it suffers from the lack of
scheduling management during the calls. The tasks can move between CPUs while
the measurements take place. Depending on how often this occurs, this is likely to
have an impact the precision of the measurement.
15.13 Optimizing SMP Performance
The purpose of SMP systems is to increase performance. Simply making code
SMP-ready does not necessarily exploit the performance potential available with
multiple CPUs. Additional work is necessary to make code SMP-optimized. (For
definitions of these terms, see 15.2.1 Terminology, p.667.)
699
15
VxWorks
Kernel Programmer's Guide, 6.6
The performance improvement of an SMP algorithm is almost completely
dependent on the amount of parallelism in the algorithm and the quality of the
multi-threaded implementation. Some algorithms are highly parallel in nature,
and take good advantage of multiple CPUs. A good example is an image
compressor, which can compress separate bands of data independently on
separate threads of execution. Since contention is low, the utilization of the CPU
can be very high, resulting in good SMP performance.
With a poorly-designed SMP algorithm, on the other hand, the cost of
synchronizing two threads of execution can completely negate the benefits of
using more than one CPU. Similarly, if you have a data-dependent algorithm and
both CPUs are competing for the same data, the system bus can be swamped with
competing bus transactions, slowing the throughput to the point that the CPUs are
data-starved, and overall throughput suffers.
In the worst case, SMP will actually slow down an algorithm, resulting in worse
performance than on a uniprocessor system. In the best case, taking advantage of
the fact that there is twice as much L1 cache in a dual-processor system, might
allow algorithms to run twice as fast, simply because the working set of the
algorithm fits better in the twice-as-large cache. These types of algorithms are,
however, fairly rare.
Threading
Threading involves turning a single-thread application into a multi-threaded one
by replicating tasks. A typical example involves a worker task that fetches work
from a queue that is being filled by another task or an ISR. Assuming the
bottleneck in the application is the worker task, performance can be increased by
replicating the worker task. Threading is not a new concept—it was introduced
when multitasking operating systems were created. However, in a uniprocessor
system threading only increases throughput of the application when its threads
are subject to wait periods. That is, when one thread waits for a resource, the
scheduler can dispatch another thread, and so on. In cases where the bottleneck is
the CPU itself, threading cannot help performance. For example, compute
intensive applications typically do not benefit from threading on a uniprocessor
system. However, this is not the case on a SMP system. Because of the presence of
additional CPUs, threading increases performance particularly when the
bottleneck is the CPU.
Using Spinlocks
Spinlocks affect interrupt and task-preemption latency and should therefore be
used sparingly and only for very short periods of time. For more information, see
15.6 Spinlocks for Mutual Exclusion and Synchronization, p.679.
700
15 VxWorks SMP
15.13 Optimizing SMP Performance
Using Floating Point and Other Coprocessors
For reasons of efficiency, coprocessor task creation options (such as VP_FP_TASK)
should be used carefully—that is, only with tasks that will actually make use of
coprocessors. When a task is created with a coprocessor option, the state of the
coprocessor is saved and restored with each context switch, which is unnecessary
overhead if the coprocessor is never used. VxWorks SMP does not support the
VxWorks UP option of lazy state save-and-restore, because tasks are not guaranteed
to resume execution on the same CPU on which they were last scheduled to run.
Using vmBaseLib
The vmBaseLib library is the VxWorks MMU management library that allows
kernel applications and drivers to manage the MMU. An important task of an SMP
operating system is to ensure the coherency of the translation look aside buffers
(TLBs) of the MMU contained in each CPU. Some CPUs, like the MPC8641D, have
hardware that ensures TLBs are always coherent. Other CPUs, such as the
BCM1480 and Intel Dual Core Xeon LV, do not have this capability. In these cases
the operating system is responsible for propagating MMU events that affect TLB
coherency to all CPUs in the system.
While not all events require propagation—it is generally limited to events that
modify an existing page mapping such as with vmStateSet( )—the propagation
that must be performed has a negative impact on some VxWorks SMP vmBaseLib
routines. To reduce the negative impact on your system’s performance, minimize
the number of calls to vmStateSet( ), and so on. For example, if a region with
special settings is needed from time to time during system operation, it is better to
set it up once during startup, and then reuse it as needed, rather than creating and
destroying a region for each use.
Interrupt and Task CPU Affinity
For some applications and systems, assigning specific interrupts or specific tasks
to designated CPUs can provide performance advantages. For more information,
see 15.10.2 Interrupt CPU Affinity, p.694 and 15.10.1 Task CPU Affinity, p.691.
701
15
VxWorks
Kernel Programmer's Guide, 6.6
15.14 Sample Programs
Sample programs are provided for VxWorks SMP, which demonstrates VxWorks
SMP features and performance.
The following applications illustrate I/O intensive activity and system-call
intensive activity:
philDemo
Dijkstra's Dining Philosophers Problem.
smpLockDemo
Demonstrates VxWorks SMP’s synchronization mechanism for sharing data
across multiple CPUs.
The next set of demos illustrate computation-intense activity:
primesDemo
Prime number computation.
rawPerf
Calculation of pi using floating-point arithmetic.
The demo applications can be linked to a VxWorks SMP image by configuring
VxWorks with the INCLUDE_SMP_DEMO component. The source code is
provided in installDir/vxworks-6.x/target/src/demo/smp, if VxWorks source code
has been installed. For more information about the demos, refer to their entries in
the VxWorks Kernel API Reference.
15.15 Migrating Code to VxWorks SMP
The key issue to consider in migrating code to an SMP system is that symmetric
multiprocessing allows for concurrent execution of tasks with other tasks, of tasks
with ISRs, and of ISRs with other ISRs.
Concurrent execution in SMP requires the use of different facilities for mutual
exclusion and synchronization, it precludes some routines that are available for a
uniprocessor system, and it makes the practice of relying on implicit
synchronization of tasks dangerous (if not disastrous). For example, the
concurrent execution of several of the highest priority tasks in a system can
uncover unprotected race conditions that were hidden in a uniprocessor system.
702
15 VxWorks SMP
15.15 Migrating Code to VxWorks SMP
In addition, multiple CPUs introduce complexities with regard to objects that are
global in a uniprocessor system, but must be CPU-specific in an SMP system.
Migrating code from VxWorks UP to VxWorks SMP necessarily involves several
steps between uniprocessor code and hardware to SMP code and hardware.
The migration process also involves using different multitasking facilities,
different BSP support, and so on. Some parts of migration activity involve
replacing a uniprocessor technology with an SMP one—such as replacing
taskLock( ) with spinLockTaskTake( )—while others involve changing the use of
features that have different behaviors in the VxWorks UP and VxWorks SMP (for
example, some vmBaseLib routines).
This section provides an overview of the migration process, a summary of the
operating system facilities that need to be taken into account in migration, and
more detailed information about individual migration issues. It does not provide
a completely self-contained discussion of migration to SMP. It is necessarily a
supplement to the preceding material in this chapter, which provides information
about the core features of VxWorks SMP. Incorporation of these features naturally
forms the basis for migrating code from VxWorks UP to VxWorks SMP. The
material following the discussion of general issues—15.15.1 Code Migration Path,
p.703 and 15.15.2 Overview of Migration Issues , p.705—therefore covers some of the
less tidy aspects of migration.
15
15.15.1 Code Migration Path
This section describes the migration model and the recommended path for
migrating applications from VxWorks UP to VxWorks SMP.
Wind River recommends that you approach migrating code designed for an earlier
version of VxWorks UP to the current version of VxWorks SMP with the following
steps:
703
VxWorks
Kernel Programmer's Guide, 6.6
Step 1:
Migrate Uniprocessor Code from Previous to Current Version of VxWorks
Migrating the VxWorks UP 6.previous code to a VxWorks UP 6.current system
involves eliminating or replacing any routines and coding practices that are
unsupported or incompatible with VxWorks SMP. For an overview of what must
be replaced or changed, see 15.15.2 Overview of Migration Issues , p.705. Note that
this step might include migrating from uniprocessor hardware to a single
processor on SMP hardware; that is, the hardware used for the end-point of the
first and second migration steps shown in Figure 15-4 may therefore be the same.
Step 2:
Migrate Code from Current VxWorks UP to VxWorks SMP
Migrating the code from the VxWorks UP 6.current system to a VxWorks SMP
6.current system involves correcting any concurrent execution bugs (such as
contention issues and deadlocks) that appear in an SMP environment. This step
includes migrating to multiprocessor use of SMP hardware.
Step 3:
Optimize Code for SMP Performance
Optimizing the code on VxWorks SMP allows it to make the fullest use of
symmetric multiprocessing. For more information about this topic, see
15.13 Optimizing SMP Performance, p.699.
For definitions SMP-ready and SMP-optimized see 15.2.1 Terminology, p.667.
Figure 15-4 illustrates the recommended application migration path.
Figure 15-4
VxWorks SMP Migration
UP App
UP App
SMPReady
SMP App
SMPReady
SMP App
SMPOptimized
VxWorks UP
6.previous
VxWorks UP
6.current
VxWorks SMP
6.current
VxWorks SMP
6.current
UP Hardware
UP Hardware
SMP Hardware
SMP Hardware
704
15 VxWorks SMP
15.15 Migrating Code to VxWorks SMP
15.15.2 Overview of Migration Issues
Table 15-10 provides an overview of the uniprocessor features or programming
practices that are incompatible with (or unsupported by) VxWorks SMP, the
appropriate SMP alternatives, and references to these topics. All code designed for
VxWorks UP should be examined carefully with regard to these issues as part of
the migration process.
Table 15-10
VxWorks SMP Migration Issues
Incompatible Uniprocessor
Features and Practices
SMP Features and Practices
Reference
Coding practices relying on
implicit synchronization.
Use of explicit
synchronization facilities,
such as semaphores and
spinlocks.
15.15.4 Implicit Synchronization of
Tasks, p.706
Various cacheLib routines.
Revise use of routines.
cacheLib Restrictions, p.708
Various vmBaseLib routines. Restrict use of routines.
vmBaseLib Restrictions, p.709
taskLock( ), intLock( )
spinLockLib,
taskCpuLock( ),
intCpuLock( ), atomic
operators
15.15.5 Synchronization and Mutual
Exclusion Facilities, p.707 and Task
Locking: taskLock( ) and
taskUnlock( ), p.711
taskRtpLock( ),
taskRtpUnlock( )
semaphores, atomic
operators
15.15.5 Synchronization and Mutual
Exclusion Facilities, p.707 and Task
Locking in RTPs: taskRtpLock( ) and
taskRtpUnlock( ), p.711
task variables, taskVarLib
routines
__thread storage class
Task Variable Management:
taskVarLib , p.712
tlsLib routines
__thread storage class
Task Local Storage: tlsLib, p.712
Replace with CPU-specific
Accessing global variables
that are CPU-specific variables variable routines and
or inaccessible in SMP.
practices.
15.15.8 SMP CPU-Specific Variables
and Uniprocessor Global Variables,
p.712
705
15
VxWorks
Kernel Programmer's Guide, 6.6
Table 15-10
VxWorks SMP Migration Issues
Incompatible Uniprocessor
Features and Practices
SMP Features and Practices
Reference
Memory-access attributes
unsuited for SMP memory
coherency.
15.15.9 Memory-Access Attributes,
Review calls that directly
p.714
manipulate coherency
protocols and caching
modes. Adhere to restrictions
as documented in
architecture supplements.
Drivers that are not VxBus
compliant.
VxBus-compliant drivers.
15.15.10 Drivers and BSPs, p.715
Uniprocessor BSP.
SMP BSP.
15.15.10 Drivers and BSPs, p.715
Uniprocessor boot loader.
Boot loader that supports
VxWorks SMP.
Also note that the custom scheduler framework is not supported for VxWorks
SMP.
15.15.3 RTP Applications and SMP
As in VxWorks UP systems, RTP (user mode) applications have a more limited set
of mutual exclusion and synchronization mechanisms available to them than
kernel code or kernel applications. In VxWorks SMP, they can make use of
semaphores and atomic operations, but not spinlocks, memory barriers, or
CPU-specific mutual exclusion mechanisms. In addition, the semExchange( )
routine provides for an atomic give and exchange of semaphores.
15.15.4 Implicit Synchronization of Tasks
VxWorks is a multitasking operating system, and VxWorks and its applications
are re-entrant; therefore migrating to a system in which tasks run concurrently is
not normally a problem as long as tasks are explicitly synchronized. For example,
Task A giving a semaphore to Task B to allow it to run is a form of explicit
synchronization. On the other hand, implicit synchronization techniques—such as
those that rely on task priority—cannot be relied on in VxWorks SMP. For
example, if high priority Task A spawns low priority Task B, expecting that Task
706
15 VxWorks SMP
15.15 Migrating Code to VxWorks SMP
B will not run until Task A releases the CPU is an invalid assumption on an SMP
system.
Implicit synchronization based on task priority is not easy to detect. Careful
review of all code that causes a task to become ready to run would be a useful
approach. For example, review code that uses the following types of routines:
■
Routines That Create Tasks
The taskSpawn( ), rtpSpawn( ), and other routines create a new task. In an
SMP system, the new task may have already started running on another CPU
by the time the procedure call returns—regardless of the relative priority of the
new task compared to the creating task. If the creator and the created tasks
interact using semaphores, message queues or other objects, these objects must
be created or initialized before creating the new task.
■
Routines that Unpend a Waiting Task
The semGive( ), msgQSend( ), eventSend( ) and other routines can unpend a
waiting task, which may begin running before the procedure call returns—
even though the waiting task has a lower priority than the calling task.
For example, in a VxWorks UP system, a task can protect a critical section of code
by using intLock( ), which prevents all interrupts from being processed and
thereby prevents ISRs from entering the critical section. The ISR does not use
explicit mutual exclusion when accessing the critical section because the task
cannot be running when the ISR is running on a uniprocessor system. (This is
likely to be a common occurrence in drivers where a portion of a driver runs in an
ISR and queues up work for a task to perform.) The assumption that tasks do not
run when ISRs do is simply not true in an SMP system. Therefore ISRs must use
explicit mutual exclusion in cases such as the one described above. The preferred
mechanism is the ISR-callable spinlock as described in 15.6 Spinlocks for Mutual
Exclusion and Synchronization, p.679.
15.15.5 Synchronization and Mutual Exclusion Facilities
Because of concurrent execution on an SMP system, there are necessarily
differences in facilities available for explicit synchronization and mutual exclusion
for VxWorks UP and VxWorks SMP.
Semaphores are appropriate for both environments, but uniprocessor interrupt
and task locking mechanisms are not appropriate and not available for SMP—
spinlocks and other mechanisms should be used instead.
707
15
VxWorks
Kernel Programmer's Guide, 6.6
In VxWorks SMP it is, for example, possible for a task to be running at the same
time that an ISR is executing. This is not possible in VxWorks UP, and therefore
requires changes to the way mutual exclusion between a task and an ISR is done.
A common synchronization method between an ISR and a task in VxWorks is the
binary semaphore. This mechanism works equally well in VxWorks SMP, and
therefore code that uses binary semaphores in this manner need not be modified
for VxWorks SMP—provided the ISR is running with interrupts enabled when it
calls semGive( ). This is also true of other messaging and synchronization
routines, such as message queues and VxWorks events. Note, however, that when
an ISR wakes up a task (by giving a binary semaphore, sending a VxWorks event,
sending a message to a message queue, etc.), the awakened task may start running
immediately on another CPU.
For more information about uniprocessor synchronization mechanisms and the
SMP alternatives, see 15.15.7 Unsupported Uniprocessor Routines and SMP
Alternatives , p.709.
15.15.6 VxWorks SMP Variants of Uniprocessor Routines
While the routines provided in VxWorks UP and VxWorks SMP are largely the
same, there are a few that have different behaviors in the VxWorks SMP due to the
requirements of multiprocessor systems, and their use has restrictions.
cacheLib Restrictions
The VxWorks UP cache routines are designed around a uniprocessor system.
Enabling and disabling the caches, invalidating, flushing or clearing elements of
the cache all have a CPU-specific nature, as they refer to the local CPU’s cache. In
an SMP system, this CPU-specific nature is less meaningful. The systems that are
supported by VxWorks SMP all provide hardware cache coherency, both between
the individual CPUs in the SMP system and between the memory subsystem and
the device address space. Given these characteristics, the cache restrictions and
behavior modifications described below apply to VxWorks SMP.
cacheEnable( ) and cacheDisable( )
The only way for the hardware cache coherency to be effective is to have the caches
turned on at all times. VxWorks SMP therefore turns on the caches of each CPU as
it is enabled, and never allows them to be disabled. Calling cacheEnable( ) in
VxWorks SMP always returns OK. Calling cacheDisable( ) in VxWorks SMP
always returns ERROR, with errno set to S_cacheLib_FUNCTION_UNSUPPORTED.
708
15 VxWorks SMP
15.15 Migrating Code to VxWorks SMP
cacheClear( ), cacheFlush( ), and cacheInvalidate( )
Because of the hardware cache coherency of SMP-capable platforms, these
routines are not necessary. If these functions are called in VxWorks SMP, they
perform no function (are NOOPs) and simply return OK.
cacheLock( ) and cacheUnlock( )
These routines are not supported in VxWorks SMP. If they are called, they return
ERROR and set errno to S_cacheLib_FUNCTION_UNSUPPORTED
vmBaseLib Restrictions
VxWorks SMP does not provide APIs for changing memory page attributes. On an
SMP system it is essential that the RAM regions that are shared between the CPUs
never be allowed to get out of coherency with one another. If a single page in
system RAM were to have its attributes changed so that it no longer correctly
participates in the hardware coherency protocol, any operating system use of that
page (for spinlocks, shared data structures, and so on) would be at risk of
unpredictable behavior. This unpredictable behavior might even occur long after
the offending change in the page attributes has occurred. This type of problem
would be extremely difficult to debug, because of the underlying assumption that
the hardware coherency in SMP simply works.
vmBaseStateSet( ) and vmStateSet( )
These routines are called to modify the attributes of a single page of virtual
memory. In an SMP system, the caching attributes of a page cannot be modified.
Attempting to do so causes these routines to return ERROR with errno set to
S_vmLib_BAD_STATE_PARAM.
15.15.7 Unsupported Uniprocessor Routines and SMP Alternatives
Some of the routines available in VxWorks UP are not supported in VxWorks SMP
because their functionality is at odds with truely concurrent execution of tasks and
ISRs, or because they would degrade performance to an unacceptible extent. SMP
alternatives provide comparable functionality that is designed for symmetric
multiprocessing.
709
15
VxWorks
Kernel Programmer's Guide, 6.6
Interrupt Locking: intLock( ) and intUnlock( )
In VxWorks UP, the intLock( ) routine is used by a task or ISR to prevent VxWorks
from processing interrupts. The typical use of this routine is to guarantee mutually
exclusive access to a critical section of code between tasks, between tasks and ISRs,
or between ISRs (as with nested ISRs—when ISR can be preempted by an ISR of
higher priority).
This mechanism would be inappropriate for a multiprocessor system, and
VxWorks SMP provides the following alternatives for interrupt locking:
■
If interrupt locking is used to make a simple pseudo-atomic operation on a
piece of memory, atomic operations may be a suitable alternative.
■
If interrupt locking is used as a mutual exclusion mechanism between tasks
only, semaphores or task-only spinlocks are a suitable replacements. Spinlock
acquisition and release operations are faster than semaphore operations, so
they would be suitable to protect a short critical section that needs to be fast.
Semaphores are suitable for longer critical sections.
■
If interrupt locking is used as a mutual exclusion mechanism between tasks
and ISRs, or between ISRs, ISR-callable spinlocks are a suitable replacement.
■
If interrupt locking is used as a mutual exclusion mechanism between tasks
only, taskCpuLock( ) can be used instead as long as all tasks taking part in the
mutual exclusion scenario have the same CPU affinity. This alternative should
not be used in custom extensions to the operating system other than as a
temporary measure when migrating code the from VxWorks UP to VxWorks
SMP.
■
If interrupt locking is used as a mutual exclusion mechanism between tasks,
between tasks and ISRs, or between ISRs, then intCpuLock( ) can be used as
long as all tasks and ISRs taking part in the mutual exclusion scenario have the
same CPU affinity. This alternative should not to be used in custom extensions
to the operating system other than as a temporary measure when migrating
the code from VxWorks UP to VxWorks SMP.
Note that for VxWorks SMP, ISR-callable spinlocks are implemented with the
same behavior as the interrupt locking routines intLock( ) and intUnlock( ).
For information about SMP mutual exclusion facilities, see 15.6 Spinlocks for Mutual
Exclusion and Synchronization, p.679, 15.7 CPU-Specific Mutual Exclusion, p.685, and
15.9 Atomic Memory Operations, p.690.
710
15 VxWorks SMP
15.15 Migrating Code to VxWorks SMP
Task Locking: taskLock( ) and taskUnlock( )
In VxWorks UP, task locking routines are used by a task to prevent the scheduling
of any other task in the system, until it calls the corresponding unlock routine. The
typical use of these routines is to guarantee mutually exclusive access to a critical
section of code.
With VxWorks UP, the kernel routine taskLock( ) is used to lock out all other tasks
in the system by suspending task preemption (also see Task Locking in RTPs:
taskRtpLock( ) and taskRtpUnlock( ), p.711). This mechanism would be
inappropriate for a multiprocessor system, and VxWorks SMP provides the
following alternatives:
■
Semaphores.
■
Atomic operations.
■
task-only spinlocks. Spinlock acquisition and release operations are faster than
semaphore operations (the other alternative in this case) so they would be
suitable to protect a short critical section that needs to be fast.
■
The taskCpuLock( ) routines for situations where all tasks taking part in the
task-locking scenario have the same CPU affinity. This alternative should not
be used in custom extensions to the operating system other than as a
temporary measure when migrating the code from VxWorks UP to VxWorks
SMP.
Note that for VxWorks UP, task-only spinlocks are implemented with the same
behavior as the task locking routines taskLock( ) and taskUnlock( ).
For information about SMP mutual exclusion facilities, see 15.6 Spinlocks for Mutual
Exclusion and Synchronization, p.679, 15.7 CPU-Specific Mutual Exclusion, p.685, and
15.9 Atomic Memory Operations, p.690.
Task Locking in RTPs: taskRtpLock( ) and taskRtpUnlock( )
The taskRtpLock( ) routine is used in RTP applications to prevent scheduling of
any other tasks in the process of the calling task. As with taskLock( ), the
taskRtpLock( ) routine is not appropriate for an SMP system (also see Task Locking:
taskLock( ) and taskUnlock( ), p.711).
The taskRtpLock( ) routine is provided with VxWorks SMP, but it generates a fatal
error when called, and the process terminates. Semaphores or atomic operators
should be used instead.
711
15
VxWorks
Kernel Programmer's Guide, 6.6
Task Variable Management: taskVarLib
The VxWorks UP task variable facility provided by taskVarLib is not compatible
with an SMP environment, as more than one task using the same task variable
location could be executing concurrently. Therefore task variables and the
taskVarAdd( ) and taskVarDelete( ) routines are not available in VxWorks SMP.
The __thread storage class should be used instead. For more information, see
4.7.3 Task-Specific Variables, p.190.
Task Local Storage: tlsLib
The VxWorks UP task local storage routines provided by tlsLib for user-mode
(RTP) applications are not compatible with an SMP environment, as more than one
task using the same task variable location could be executing concurrently. The
tlsLib routines are as follows:
■
■
■
■
■
tlsKeyCreate( )
tlsValueGet( )
tlsValueSet( )
tlsValueOfTaskGet( )
tlsValueOfTaskSet( )
The __thread storage class should be used instead. For more information, see
4.7.3 Task-Specific Variables, p.190.
15.15.8 SMP CPU-Specific Variables and Uniprocessor Global Variables
Some objects that are global in a uniprocessor system (such as errno) are
CPU-specific entities in VxWorks SMP, and others are inaccessible or non-existent
in VxWorks SMP.
!
CAUTION: Wind River recommends that you do not manipulate any CPU-specific
or global variables directly. Using the appropriate API is recommended to prevent
unpredictable behavior and to ensure compatibility with future versions of
VxWorks.
712
15 VxWorks SMP
15.15 Migrating Code to VxWorks SMP
SMP Per-CPU Variables
The SMP CPU-specific variables that can be accessed indirectly with appropriate
routine are as follows:
■
■
■
■
errno
taskIdCurrent
intCnt
isrIdCurrent
errno
From a programming perspective errno behaves like a global variable that
contains the error value of the currently running task or ISR. VxWorks SMP has
mechanism that allows it to manage errno as a CPU-specific variable in a
transparent manner.
Wind River recommends that you use errnoLib routines to work with errno.
However, you may access the errno variable directly from C and C++ code that
includes errno.h. Do not access the errno variable directly from assembly code.
!
CAUTION: Do not access errno directly from assembly code. Do not access errno
directly from C or C++ code that does not include errno.h.
15
taskIdCurrent
The uniprocessor taskIdCurrent global variable (declared in taskLib.h) does not
exist in VxWorks, because of concurrent execution on multiple CPUs. Any
uniprocessor code that reads taskIdCurrent should make calls to taskIdSelf( )
instead.
intCnt
In an SMP system, specific interrupts are dedicated to a specific CPU. The intCnt
variable is used to track the number of nested interrupts that exist on a specific
CPU. Code that references this variable should be changed to use the intCount( )
routine instead.
isrIdCurrent
The isrIdCurrent variable is used to identify the ISR executing on the specific CPU.
This global is only available if the INCLUDE_ISR_OBJECTS component is included
in VxWorks.Code that accesses isrIdCurrent must be changed to use the
isrIdSelf( ) routine instead.
713
VxWorks
Kernel Programmer's Guide, 6.6
Uniprocessor-Only Global Variables
The VxWorks UP variables that do not exist in VxWorks SMP—or that must not be
accessed by user code in any way—are as follows:
■
■
■
■
vxIntStackBase
vxIntStackEnd
kernelIsIdle
windPwrOffCpuState
vxIntStackBase
The vxIntStackBase variable identifies base of the interrupt stack used for
processing interrupts. For VxWorks SMP, each CPU has a vxIntStackBase to
process interrupts since interrupts may be processed by multiple CPUs
simultaneously. There is no routine for accessing this variable and it must not be
accessed by user code.
vxIntStackEnd
The vxIntStackEnd variable identifies the end of the interrupt stack for each CPU.
There is no routine for accessing this variable and it must not be accessed by user
code.
kernelIsIdle
In VxWorks UP the kernelIsIdle variable indicates whether or not the system is
idle. This variable does not exist in VxWorks SMP. CPU Information and
Management Routines, p.694 describes routines that can be used instead.
windPwrOffCpuState
The windPwrOffCpuState variable identifies power management state on the
specific CPU. There is no routine for accessing this variable and it must not be
accessed by user code.
15.15.9 Memory-Access Attributes
In an SMP system memory coherency is required to ensure that each CPU sees the
same memory contents. Depending on the CPU architecture, some memory access
attributes may not be suitable for a system where memory coherency is required.
For information in this regard, see the VxWorks Architecture Supplement.
714
15 VxWorks SMP
15.15 Migrating Code to VxWorks SMP
15.15.10 Drivers and BSPs
Both drivers and BSPs developed for VxWorks SMP must adhere to the
programming practices described throughout this chapter. Drivers must also
conform to the VxBus driver model. BSPs, in addition to providing support for
VxBus, must provide facilities different from the VxWorks UP for reboot handling,
CPU enumeration, interrupt routing and assignment, and so on. For more
information, see VxWorks Device Driver Developer’s Guide and VxWorks BSP
Developer’s Guide.
15
715
VxWorks
Kernel Programmer's Guide, 6.6
716
16
Shared-Memory Objects:
VxMP
16.1 Introduction 717
16.2 Using Shared-Memory Objects 718
16.3 System Requirements 741
16.4 Performance Considerations 742
16.5 Configuring VxWorks for Shared Memory Objects 744
16.6 Displaying Information About Shared Memory Objects 752
16.7 Troubleshooting 752
16.1 Introduction
VxMP is a VxWorks component that provides shared-memory objects dedicated
to high-speed synchronization and communication between tasks running in
separate instances of VxWorks.
Shared-memory objects are a class of system objects that can be accessed by tasks
running on different processors. The object’s data structures reside in memory
accessible by all processors. Shared-memory objects are an extension of local
VxWorks objects. Local objects are only available to tasks on a single processor.
VxMP supplies the following types of shared-memory objects:
■
shared semaphores (binary and counting)
717
VxWorks
Kernel Programmer's Guide, 6.6
■
shared message queues
■
shared-memory partitions (system- and user-created partitions)
Shared-memory objects provide the following advantages:
■
A transparent interface that allows shared-memory objects to be manipulated
with the same routines that are used for manipulating local objects.
■
High-speed inter-processor communication—no going through an
unnecessary network stack.
■
The shared memory can reside either in dual-ported RAM or on a separate
memory board.
VxMP consists of the following facilities: a name database (smNameLib), task
synchronization and resource tracking with semaphores (semSmLib), messaging
with message queues (msgQSmLib) to build a custom protocol, and a
shared-memory allocator (smMemLib).
NOTE: VxMP can only be used in kernel space. It cannot be used in user space
(real-time processes).
NOTE: VxMP is currently not supported for SMP systems.
16.2 Using Shared-Memory Objects
VxMP provides creation APIs specifically for shared memory objects—shared
semaphores, message queues, and memory partitions. As with standard kernel
objects, an object ID is returned when an object is created, and the ID is
subsequently used to identify that object. For any given shared-memory object, the
ID is global to the VxMP system. That is, the same ID is used to access that object,
regardless of the CPU from which it is accessed.
After a shared-memory object is created, kernel tasks can operate on it with the
same routines used for the corresponding local object. Shared semaphores, shared
message queues, and shared-memory partitions have the same syntax and
interface as their local counterparts, which allows routines such as semGive( ),
semTake( ), msgQSend( ), msgQReceive( ), memPartAlloc( ), and
memPartFree( ) to operate on both types of objects.
718
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
Kernel tasks running on different CPUs can provide and obtain the object ID of
shared memory objects in a variety of ways, including shared message queues and
data structures in shared memory. The most convenient method, however, is by
using the VxMP name database to publish and access the object ID.
After the shared-memory facilities are initialized at run-time, all processors are
treated alike. Kernel tasks on any CPU can create and use shared-memory objects.
No processor has priority over another from a shared-memory object’s point of
view.1
There are few restrictions on shared-memory object use (they cannot, for example,
be used at interrupt level), and they are easily portable between uniprocessor and
multiprocessor systems, which can be advantageous in the development process.
Note that throughout the remainder of this chapter, system objects under
discussion refer to shared objects unless otherwise indicated.
16.2.1 Multiprocessor-Uniprocessor Portability
VxMP provides a transparent interface that makes it easy to execute code using
shared-memory objects on both a multiprocessor system and a uniprocessor
system.
Only the object creation routines are different for shared-memory objects. After
creation, the same routines as used for operations on local objects can be used for
the shared memory objects. This allows an application to run in either a
uniprocessor or a multiprocessor environment with only minor changes to system
configuration, initialization, and object creation.
Using shared-memory objects on a uniprocessor system is useful for testing an
application before porting it to a multiprocessor configuration. However, for
objects that are used only locally, local objects always provide the best
performance.
16.2.2 Multiprocessing and Byte Ordering
Systems making use of shared memory can include a combination of supported
architectures. This enables applications to take advantage of different processor
types and still have them communicate.
1. Do not confuse this type of priority with the CPU priorities associated with VMEbus access.
719
16
VxWorks
Kernel Programmer's Guide, 6.6
Note, however, on systems where the processors have different byte ordering, you
must call the ntohl and htonl macros to byte-swap the application shared data that
is passed with message queues and so on. VxMP handles the endianness of all
system data structures and IDs internally. Names are byte-streams (strings) so
they are not subject to endianness issues. The IDs returned by the name database
are converted internally and the ID obtained by the user has the correct
endianness.
16.2.3 Restrictions on Shared Memory Object Use
Shared-memory objects are only available to kernel tasks. Unlike local semaphores
and message queues, shared-memory objects cannot be used at interrupt level. No
routines that use shared-memory objects can be called from ISRs. An ISR is
dedicated to handle time-critical processing associated with an external event;
therefore, using shared-memory objects at interrupt time is not appropriate. On a
multiprocessor system, run event-related, time-critical processing on the CPU on
which the time-related interrupt occurred.
Note that shared-memory objects are allocated from dedicated shared-memory
pools, and cannot be deleted.
When using shared-memory objects, the maximum number of each object type
must be specified; see 16.5.7 Dual-Port or External Memory, p.748. If applications are
creating more than the specified maximum number of objects, it is possible to run
out of memory. For more information in this regard, see 16.7 Troubleshooting,
p.752.
16.2.4 Publishing Objects With the Name Database
The VxMP name database allows the association of any value to any name, such as
a shared-memory object’s ID with a unique name. It can communicate or publish a
shared-memory block’s address and object type. The name database provides
name-to-value and value-to-name translation, allowing objects in the database to
be accessed either by name or by value.
While other methods exist for making an object’s ID known to other nodes (such
as with message queues, by being written to a shared memory block at a
pre-determined offset, and so on), the name database is the most convenient
method for doing so—it is simpler and it allows any node access to the information
at will.
720
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
Typically, the kernel task that creates an object also publishes the object’s ID by
means of the name database. By adding the new object to the database, the task
associates the object’s ID with a name. Tasks on other processors can look up the
name in the database to get the object’s ID. After the task has the ID, it can use it to
access the object. For example, task t1 on CPU 1 creates an object. The object ID is
returned by the creation routine and entered in the name database with the name
myObj. For task t2 on CPU 0 to operate on this object, it first finds the ID by
looking up the string myObj in the name database.
Table 16-1
Name Service Routines
Routine
Description
smNameAdd( )
Adds a name to the name database.
smNameRemove( )
Removes a name from the name database.
smNameFind( )
Finds a shared symbol by name.
smNameFindByValue( )
Finds a shared symbol by value.
smNameShow( )
Displays the name database to the standard output
device.a
a. Automatically included if INCLUDE_SM_OBJ is selected.
This same technique can be used to publish a shared-memory address. For
example, task t1 on CPU 0 allocates a portion of memory and adds the address to
the database with the name mySharedMem. Task t2 on CPU 1 can find the address
of this shared memory by looking up the address in the name database using the
string mySharedMem.
Tasks on different processors can use an agreed-upon name to get a newly created
object’s value. See Table 16-1 for a list of name service routines. Note that
retrieving an ID from the name database need occur only one time for each task,
and usually occurs during application initialization. An ID can simply be retrieved
on a per-processor basis, if it is stored in a global variable (for example). However,
it is generally a good practice to retrieve IDs on a per-task basis.
The name database service routines automatically convert to or from network-byte
order; do not call htonl( ) or ntohl( ) C macros explicitly for values provided by the
name database. These C macros must, however be used on application shared data
that is passed between processors with different byte orders using message queues
and so on. For more information in this regard, see 16.2.2 Multiprocessing and Byte
Ordering, p.719.
721
16
VxWorks
Kernel Programmer's Guide, 6.6
The object types listed in Table 16-2 are defined in smNameLib.h.
Table 16-2
Shared-Memory Object Types
Constant
Hex Value
Description
T_SM_SEM_B
0
Shared binary semaphore.
T_SM_SEM_C
1
Shared counting semaphore.
T_SM_MSG_Q
2
Shared message queue.
T_SM_PART_ID
3
Shared memory partition ID.
T_SM_BLOCK
4
Shared memory block.
The following example shows the name database as displayed by
smNameShow( ), which is automatically included if VxWorks is configured with
the INCLUDE_SM_OBJ component. The parameter to smNameShow( ) specifies
the level of information displayed; in this case, 1 indicates that all information is
shown. For additional information, see the smNameShow( ) API reference.
-> smNameShow 1
value = 0 = 0x0
The output is sent to the standard output device, and looks like the following:
Name in Database Max : 100 Current : 5 Free : 95
Name
Value
Type
----------------- ------------- ------------myMemory
0x3835a0
SM_BLOCK
myMemPart
0x3659f9
SM_PART_ID
myBuff
0x383564
SM_BLOCK
mySmSemaphore
0x36431d
SM_SEM_B
myMsgQ
0x365899
SM_MSG_Q
16.2.5 Shared Semaphores
Like local semaphores, shared semaphores provide synchronization by means of
atomic updates of semaphore state information. See 4. Multitasking and the API
reference for semLib for a complete discussion of semaphores. Shared semaphores
can be given and taken by tasks executing in the kernel on any CPU with access to
the shared memory. They can be used for either synchronization of tasks running
on different CPUs or mutual exclusion for shared resources.
To use a shared semaphore, a task creates the semaphore and announces its ID to
other nodes. This can be done, for example, by adding it to the name database. A
722
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
task on any CPU in the system can use the semaphore by first getting the
semaphore ID (for example, from the name database). When it has the ID, it can
then take or give the semaphore.
In the case of employing shared semaphores for mutual exclusion, typically there
is a system resource that is shared between tasks on different CPUs and the
semaphore is used to prevent concurrent access. Any time a task requires exclusive
access to the resource, it takes the semaphore. When the task is finished with the
resource, it gives the semaphore.
For example, there are two tasks, t1 on CPU 0 and t2 on CPU 1. Task t1 creates the
semaphore and publishes the semaphore’s ID by adding it to the database and
assigning the name myMutexSem. Task t2 looks up the string myMutexSem in
the database to get the semaphore’s ID. Whenever a task wants to access the
resource, it first takes the semaphore by using the semaphore ID. When a task is
done using the resource, it gives the semaphore.
In the case of employing shared semaphores for synchronization, assume a task on
one CPU must notify a task on another CPU that some event has occurred. The task
being synchronized pends on the semaphore waiting for the event to occur. When
the event occurs, the task doing the synchronizing gives the semaphore.
For example, there are two tasks, t1 on CPU 0 and t2 on CPU 1. Both t1 and t2 are
monitoring robotic arms. The robotic arm that is controlled by t1 is passing a
physical object to the robotic arm controlled by t2. Task t2 moves the arm into
position but must then wait until t1 indicates that it is ready for t2 to take the object.
Task t1 creates the shared semaphore and publishes the semaphore’s ID by adding
it to the database and assigning the name objReadySem. Task t2 looks up the
string objReadySem in the database to get the semaphore’s ID. It then takes the
semaphore by using the semaphore ID. If the semaphore is unavailable, t2 pends,
waiting for t1 to indicate that the object is ready for t2. When t1 is ready to transfer
control of the object to t2, it gives the semaphore, readying t2 on CPU1.
Table 16-3
Shared Semaphore Create Routines
Create Routine
Description
semBSmCreate( )
Creates a shared binary semaphore.
semCSmCreate( )
Creates a shared counting semaphore.
There are two types of shared semaphores, binary and counting. Shared
semaphores have their own create routines and return a SEM_ID. Table 16-3 lists
723
16
VxWorks
Kernel Programmer's Guide, 6.6
the create routines. All other semaphore routines, except semDelete( ), operate
transparently on the created shared semaphore.
Figure 16-1
Shared Semaphore Queues
Executes on CPU 2 after
task1 is put on queue:
task2 ( )
{
...
semTake (semSmId,t);
...
}
Pend Queue
task2
Semaphore
State
EMPTY
task1
Executes on CPU 1
before task2:
task1 ( )
{
...
semTake (semSmId,t);
...
}
Binary Shared Semaphore
SHARED MEMORY
The use of shared semaphores and local semaphores differs in several ways:
■
The shared semaphore queuing order specified when the semaphore is created
must be FIFO. If it is not, an error is generated, and errno is set to
S_msgQLib_INVALID_QUEUE_TYPE.
Figure 16-1 shows two tasks executing on different CPUs, both trying to take
the same semaphore. Task 1 executes first, and is put at the front of the queue
because the semaphore is unavailable (empty). Task 2 (executing on a different
CPU) tries to take the semaphore after task 1’s attempt and is put on the queue
behind task 1.
■
Shared semaphores cannot be given from interrupt level. If they are, an error is
generated, and errno is set to S_intLib_NOT_ISR_CALLABLE.
■
Shared semaphores cannot be deleted. Attempts to delete a shared semaphore
return ERROR and set errno to S_smObjLib_NO_OBJECT_DESTROY.
Use semInfo( ) to get the shared task control block of tasks pended on a shared
semaphore. Use semShow( ) to display the status of the shared semaphore and a
724
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
list of pended tasks. (VxWorks must be configured with the
INCLUDE_SEM_SHOW component.)
The following example displays detailed information on the shared semaphore,
using the variable mySmSemaphoreId, which holds the ID. The level of
information is specified by the second argument (0 = summary, 1 = details).
-> semShow mySmSemaphoreId, 1
value = 0 = 0x0
The output is sent to the standard output device, and looks like the following:
Semaphore Id
: 0x36431d
Semaphore Type : SHARED BINARY
Task Queuing
: FIFO
Pended Tasks
: 2
State
: EMPTY
TID
CPU Number
Shared TCB
------------- ------------- -------------0xd0618
1
0x364204
0x3be924
0
0x36421c
16
725
VxWorks
Kernel Programmer's Guide, 6.6
Example 16-1
Shared Semaphores
The following code example depicts two tasks executing on different CPUs and
using shared semaphores. The routine semTask1( ) creates the shared semaphore,
initializing the state to full. It adds the semaphore to the name database (to enable
the task on the other CPU to access it), takes the semaphore, does some processing,
and gives the semaphore. The routine semTask2( ) gets the semaphore ID from the
database, takes the semaphore, does some processing, and gives the semaphore.
/* semExample.h - shared semaphore example header file */
#define SEM_NAME "mySmSemaphore"
/* semTask1.c - shared semaphore example */
/* This code is executed by a task on CPU #1 */
#include
#include
#include
#include
#include
#include
#include "semExample.h"
/*
* semTask1 - shared semaphore user
*/
STATUS semTask1 (void)
{
SEM_ID semSmId;
/* create shared semaphore */
if ((semSmId = semBSmCreate (SEM_Q_FIFO, SEM_FULL)) == NULL)
return (ERROR);
/* add object to name database */
if (smNameAdd (SEM_NAME, semSmId, T_SM_SEM_B) == ERROR)
return (ERROR);
/* grab shared semaphore and hold it for awhile */
semTake (semSmId, WAIT_FOREVER);
/* normally do something useful */
printf ("Task1 has the shared semaphore\n");
taskDelay (sysClkRateGet () * 5);
printf ("Task1 is releasing the shared semaphore\n");
726
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
/* release shared semaphore */
semGive (semSmId);
return (OK);
}
/* semTask2.c - shared semaphore example */
/* This code is executed by a task on CPU #2. */
#include
#include
#include
#include
#include
#include "semExample.h"
/*
* semTask2 - shared semaphore user
*/
STATUS semTask2 (void)
{
SEM_ID semSmId;
int
objType;
/* find object in name database */
if (smNameFind (SEM_NAME, (void **) &semSmId, &objType, WAIT_FOREVER)
== ERROR)
return (ERROR);
/* take the shared semaphore */
printf ("semTask2 is now going to take the shared semaphore\n");
semTake (semSmId, WAIT_FOREVER);
/* normally do something useful */
printf ("Task2 got the shared semaphore!!\n");
/* release shared semaphore */
semGive (semSmId);
printf ("Task2 has released the shared semaphore\n");
return (OK);
}
727
16
VxWorks
Kernel Programmer's Guide, 6.6
16.2.6 Shared Message Queues
Shared message queues are FIFO queues used by kernel tasks to send and receive
variable-length messages on any of the CPUs that have access to the shared
memory. They can be used either to synchronize tasks or to exchange data between
kernel tasks running on different CPUs. See 4. Multitasking and the API reference
for msgQLib for a complete discussion of message queues.
To use a shared message queue, a task creates the message queue and publishes its
ID. A task that wants to send or receive a message with this message queue first
gets the message queue’s ID. It then uses this ID to access the message queue.
For example, consider a typical server/client scenario where a server task t1 (on
CPU 1) reads requests from one message queue and replies to these requests with
a different message queue. Task t1 creates the request queue and publishes its ID
by adding it to the name database assigning the name requestQue. If task t2 (on
CPU 0) wants to send a request to t1, it first gets the message queue ID by looking
up the string requestQue in the name database. Before sending its first request,
task t2 creates a reply message queue. Instead of adding its ID to the database, it
publishes the ID by sending it as part of the request message. When t1 receives the
request from the client, it finds in the message the ID of the queue to use when
replying to that client. Task t1 then sends the reply to the client by using this ID.
To pass messages between kernel tasks on different CPUs, first create the message
queue by calling msgQSmCreate( ). This routine returns a MSG_Q_ID. This ID is
used for sending and receiving messages on the shared message queue.
Like their local counterparts, shared message queues can send both urgent or
normal priority messages.
728
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
Figure 16-2
Shared Message Queues
Executes on CPU 2 after task1:
task2 ( )
{
...
msgQReceive (smMsgQId,...);
...
}
Message
Queue
Pend Queue
task2
EMPTY
task1
Executes on CPU 1 before task2:
task1 ( )
{
...
msgQReceive (smMsgQId,...);
...
}
Shared Message Queue
SHARED MEMORY
The use of shared message queues and local message queues differs in several
ways:
■
The shared message queue task queueing order specified when a message
queue is created must be FIFO. If it is not, an error is generated and errno is set
to S_msgQLib_INVALID_QUEUE_TYPE.
Figure 16-2 shows two tasks executing on different CPUs, both trying to
receive a message from the same shared message queue. Task 1 executes first,
and is put at the front of the queue because there are no messages in the
message queue. Task 2 (executing on a different CPU) tries to receive a
message from the message queue after task 1’s attempt and is put on the queue
behind task 1.
■
Messages cannot be sent on a shared message queue at interrupt level. (This is
true even in NO_WAIT mode.) If they are, an error is generated, and errno is
set to S_intLib_NOT_ISR_CALLABLE.
■
Shared message queues cannot be deleted. Attempts to delete a shared
message queue return ERROR and sets errno to
S_smObjLib_NO_OBJECT_DESTROY.
729
16
VxWorks
Kernel Programmer's Guide, 6.6
To achieve optimum performance with shared message queues, align send and
receive buffers on 4-byte boundaries.
To display the status of the shared message queue as well as a list of tasks pended
on the queue, call msgQShow( ) (VxWorks must be configured with the
INCLUDE_MSG_Q_SHOW component.) The following example displays detailed
information on the shared message queue 0x7f8c21 as indicated by the second
argument (0 = summary display, 1 = detailed display).
-> msgQShow 0x7f8c21, 1
value = 0 = 0x0
The output is sent to the standard output device, and looks like the following:
Message Queue Id : 0x7f8c21
Task Queuing
: FIFO
Message Byte Len : 128
Messages Max
: 10
Messages Queued
: 0
Receivers Blocked : 1
Send timeouts
: 0
Receive timeouts : 0
Receivers blocked :
TID
CPU Number
Shared TCB
---------- -------------------- -------------0xd0618
1
0x1364204
Example 16-2
Shared Message Queues
In the following code example, two tasks executing on different CPUs use shared
message queues to pass data to each other. The server task creates the request
message queue, adds it to the name database, and reads a message from the queue.
The client task gets the smRequestQId from the name database, creates a reply
message queue, bundles the ID of the reply queue as part of the message, and
sends the message to the server. The server gets the ID of the reply queue and uses
it to send a message back to the client. This technique requires the use of the
network byte-order conversion C macros htonl( ) and ntohl( ), because the
numeric queue ID is passed over the network in a data field.
/* msgExample.h - shared message queue example header file */
#define MAX_MSG
(10)
#define MAX_MSG_LEN (100)
#define REQUEST_Q
"requestQue"
typedef struct message
{
MSG_Q_ID replyQId;
char
clientRequest[MAX_MSG_LEN];
} REQUEST_MSG;
730
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
/* server.c - shared message queue example server */
/* This file contains the code for the message queue server task. */
#include
#include
#include
#include
#include
#include
#include
"msgExample.h"
"netinet/in.h"
#define REPLY_TEXT "Server received your request"
/*
* serverTask - receive and process a request from a shared message queue
*/
STATUS serverTask (void)
{
MSG_Q_ID
smRequestQId;
REQUEST_MSG request;
/* request shared message queue */
/* request text */
/* create a shared message queue to handle requests */
if ((smRequestQId = msgQSmCreate (MAX_MSG, sizeof (REQUEST_MSG),
MSG_Q_FIFO)) == NULL)
return (ERROR);
/* add newly created request message queue to name database */
if (smNameAdd (REQUEST_Q, smRequestQId, T_SM_MSG_Q) == ERROR)
return (ERROR);
/* read messages from request queue */
FOREVER
{
if (msgQReceive (smRequestQId, (char *) &request, sizeof (REQUEST_MSG),
WAIT_FOREVER) == ERROR)
return (ERROR);
/* process request - in this case simply print it */
printf ("Server received the following message:\n%s\n",
request.clientRequest);
/* send a reply using ID specified in client’s request message */
if (msgQSend ((MSG_Q_ID) ntohl ((int) request.replyQId),
REPLY_TEXT, sizeof (REPLY_TEXT),
WAIT_FOREVER, MSG_PRI_NORMAL) == ERROR)
return (ERROR);
}
}
731
16
VxWorks
Kernel Programmer's Guide, 6.6
/* client.c - shared message queue example client */
/* This file contains the code for the message queue client task. */
#include
#include
#include
#include
#include
#include
#include
"msgExample.h"
"netinet/in.h"
/*
* clientTask - sends request to server and reads reply
*/
STATUS clientTask
(
char * pRequestToServer
/* request to send to the server */
/* limited to 100 chars */
)
{
MSG_Q_ID
smRequestQId; /* request message queue */
MSG_Q_ID smReplyQId;
/* reply message queue */
REQUEST_MSG request;
/* request text */
int
objType;
/* dummy variable for smNameFind */
char
serverReply[MAX_MSG_LEN]; /*buffer for server’s reply */
/* get request queue ID using its name */
if (smNameFind (REQUEST_Q, (void **) &smRequestQId, &objType,
WAIT_FOREVER) == ERROR)
return (ERROR);
/* create reply queue, build request and send it to server */
if ((smReplyQId = msgQSmCreate (MAX_MSG, MAX_MSG_LEN,
MSG_Q_FIFO)) == NULL)
return (ERROR);
request.replyQId = (MSG_Q_ID) htonl ((int) smReplyQId);
strcpy (request.clientRequest, pRequestToServer);
if (msgQSend (smRequestQId, (char *) &request, sizeof (REQUEST_MSG),
WAIT_FOREVER, MSG_PRI_NORMAL) == ERROR)
return (ERROR);
/* read reply and print it */
if (msgQReceive (request.replyQId, serverReply, MAX_MSG_LEN,
WAIT_FOREVER) == ERROR)
return (ERROR);
732
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
printf ("Client received the following message:\n%s\n", serverReply);
return (OK);
}
16.2.7 Shared-Memory Allocator
The shared-memory allocator allows kernel tasks on different CPUs to allocate and
release variable size portions of memory that are accessible from all CPUs with
access to the shared-memory system. Two sets of routines are provided: low-level
routines for manipulating user-created shared-memory partitions, and high-level
routines for manipulating a shared-memory partition dedicated to the
shared-memory system pool. (This organization is similar to that used by the
local-memory manager, memPartLib.)
Shared-memory blocks can be allocated from different partitions. Both a
shared-memory system partition and user-created partitions are available.
User-created partitions can be created and used for allocating data blocks of a
particular size. Memory fragmentation is avoided when fixed-sized blocks are
allocated from user-created partitions dedicated to a particular block size.
Shared-Memory System Partition
16
To use the shared-memory system partition, a task allocates a shared-memory
block and announces its ID to other nodes. The most convenient way is to add the
address to the name database. The routine used to allocate a block from the
shared-memory system partition returns a local address. Before the address is
published to tasks on other CPUs, this local address must be converted to a global
address. Any task that must use the shared memory must first get the address of
the memory block and convert the global address to a local address. When the task
has the address, it can use the memory.
However, to address issues of mutual exclusion, typically a shared semaphore is
used to protect the data in the shared memory. Thus in a more common scenario,
the task that creates the shared memory (and adds it to the database) also creates
a shared semaphore. The shared semaphore ID is typically published by storing it
in a field in the shared data structure residing in the shared-memory block. The
first time a task must access the shared data structure, it looks up the address of
the memory in the database and gets the semaphore ID from a field in the shared
data structure. Whenever a task must access the shared data, it must first take the
733
VxWorks
Kernel Programmer's Guide, 6.6
semaphore. Whenever a task is finished with the shared data, it must give the
semaphore.
For example, assume two tasks executing on two different CPUs must share data.
Task t1 executing on CPU 1 allocates a memory block from the shared-memory
system partition and converts the local address to a global address. It then adds the
global address of the shared data to the name database with the name
mySharedData. Task t1 also creates a shared semaphore and stores the ID in the
first field of the data structure residing in the shared memory. Task t2 executing on
CPU 2 looks up the string mySharedData in the name database to get the address
of the shared memory. It then converts this address to a local address. Before
accessing the data in the shared memory, t2 gets the shared semaphore ID from the
first field of the data structure residing in the shared-memory block. It then takes
the semaphore before using the data and gives the semaphore when it is done
using the data.
User-Created Partitions
To make use of user-created shared-memory partitions, a task creates a
shared-memory partition and adds it to the name database. Before a task can use
the shared-memory partition, it must first look in the name database to get the
partition ID. When the task has the partition ID, it can access the memory in the
shared-memory partition.
For example, task t1 creates a shared-memory partition and adds it to the name
database using the name myMemPartition. Task t2 executing on another CPU
wants to allocate memory from the new partition. Task t2 first looks up the string
myMemPartition in the name database to get the partition ID. It can then allocate
memory from it, using the ID.
734
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
Using the Shared-Memory System Partition
The shared-memory system partition is analogous to the system partition for local
memory. Table 16-4 lists routines for manipulating the shared-memory system
partition.
Table 16-4
Shared-Memory System Partition Routines
Call
Description
smMemMalloc( )
Allocates a block of shared system memory.
smMemCalloc( )
Allocates a block of shared system memory for an array.
smMemRealloc( )
Resizes a block of shared system memory.
smMemFree( )
Frees a block of shared system memory.
smMemShow( )
Displays usage statistics of the shared-memory system
partition on the standard output device. This routine is
automatically included if VxWorks is configured with the
INCLUDE_SM_OBJ component.
smMemOptionsSet( )
Sets the debugging options for the shared-memory
system partition.
smMemAddToPool( ) Adds memory to the shared-memory system pool.
smMemFindMax( )
Finds the size of the largest free block in the
shared-memory system partition.
Routines that return a pointer to allocated memory return a local address (that is,
an address suitable for use from the local CPU). To share this memory across
processors, this address must be converted to a global address before it is
announced to tasks on other CPUs. Before a task on another CPU uses the memory,
it must convert the global address to a local address. Macros and routines are
provided to convert between local addresses and global addresses; see the header
file smObjLib.h and the API reference for smObjLib.
Example 16-3
Shared-Memory System Partition
The following code example uses memory from the shared-memory system
partition to share data between kernel tasks on different CPUs. The first member
of the data structure is a shared semaphore that is used for mutual exclusion. The
735
16
VxWorks
Kernel Programmer's Guide, 6.6
send task creates and initializes the structure, then the receive task accesses the
data and displays it.
/* buffProtocol.h - simple buffer exchange protocol header file */
#define BUFFER_SIZE
#define BUFF_NAME
200
"myMemory"
/* shared data buffer size */
/* name of data buffer in database */
typedef struct shared_buff
{
SEM_ID semSmId;
char
buff [BUFFER_SIZE];
} SHARED_BUFF;
/* buffSend.c - simple buffer exchange protocol send side */
/* This file writes to the shared memory. */
#include
#include
#include
#include
#include
#include
#include
"buffProtocol.h"
/*
* buffSend - write to shared semaphore protected buffer
*/
STATUS buffSend (void)
{
SHARED_BUFF * pSharedBuff;
SEM_ID
mySemSmId;
/* grab shared system memory */
pSharedBuff = (SHARED_BUFF *) smMemMalloc (sizeof (SHARED_BUFF));
/*
* Initialize shared buffer structure before adding to database. The
* protection semaphore is initially unavailable and the receiver blocks.
*/
if ((mySemSmId = semBSmCreate (SEM_Q_FIFO, SEM_EMPTY)) == NULL)
return (ERROR);
pSharedBuff->semSmId = (SEM_ID) htonl ((int) mySemSmId);
736
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
/*
* Convert address of shared buffer to a global address and add to
* database.
*/
if (smNameAdd (BUFF_NAME, (void *) smObjLocalToGlobal (pSharedBuff),
T_SM_BLOCK) == ERROR)
return (ERROR);
/* put data into shared buffer */
sprintf (pSharedBuff->buff,"Hello from sender\n");
/* allow receiver to read data by giving protection semaphore */
if (semGive (mySemSmId) != OK)
return (ERROR);
return (OK);
}
/* buffReceive.c - simple buffer exchange protocol receive side */
/* This file reads the shared memory. */
#include
#include
#include
#include
#include
#include
#include
"buffProtocol.h"
16
/*
* buffReceive - receive shared semaphore protected buffer
*/
STATUS buffReceive (void)
{
SHARED_BUFF * pSharedBuff;
SEM_ID
mySemSmId;
int
objType;
/* get shared buffer address from name database */
if (smNameFind (BUFF_NAME, (void **) &pSharedBuff,
&objType, WAIT_FOREVER) == ERROR)
return (ERROR);
/* convert global address of buff to its local value */
pSharedBuff = (SHARED_BUFF *) smObjGlobalToLocal (pSharedBuff);
/* convert shared semaphore ID to host (local) byte order */
mySemSmId = (SEM_ID) ntohl ((int) pSharedBuff->semSmId);
737
VxWorks
Kernel Programmer's Guide, 6.6
/* take shared semaphore before reading the data buffer */
if (semTake (mySemSmId,WAIT_FOREVER) != OK)
return (ERROR);
/* read data buffer and print it */
printf ("Receiver reading from shared memory: %s\n", pSharedBuff->buff);
/* give back the data buffer semaphore */
if (semGive (mySemSmId) != OK)
return (ERROR);
return (OK);
}
Using User-Created Partitions
Shared-memory partitions have a separate create routine, memPartSmCreate( ),
that returns a MEM_PART_ID. After a user-defined shared-memory partition is
created, routines in memPartLib operate on it transparently. Note that the address
of the shared-memory area passed to memPartSmCreate( ) (or
memPartAddToPool( )) must be the global address.
Example 16-4
User-Created Partition
This example is similar to Example 16-3, which uses the shared-memory system
partition. This example creates a user-defined partition and stores the shared data
in this new partition. A shared semaphore is used to protect the data.
/* memPartExample.h - shared memory partition example header file */
#define
#define
#define
#define
CHUNK_SIZE
MEM_PART_NAME
PART_BUFF_NAME
BUFFER_SIZE
(2400)
"myMemPart"
"myBuff"
(40)
typedef struct shared_buff
{
SEM_ID semSmId;
char
buff [BUFFER_SIZE];
} SHARED_BUFF;
/* memPartSend.c - shared memory partition example send side */
/* This file writes to the user-defined shared memory partition. */
#include
#include
738
16 Shared-Memory Objects: VxMP
16.2 Using Shared-Memory Objects
#include
#include
#include
#include
#include
#include
#include
"memPartExample.h"
/*
* memPartSend - send shared memory partition buffer
*/
STATUS memPartSend
{
char *
PART_ID
SEM_ID
SHARED_BUFF *
(void)
pMem;
smMemPartId;
mySemSmId;
pSharedBuff;
/* allocate shared system memory to use for partition */
pMem = smMemMalloc (CHUNK_SIZE);
/* Create user defined partition using the previously allocated
* block of memory.
* WARNING: memPartSmCreate uses the global address of a memory
* pool as first parameter.
*/
if ((smMemPartId = memPartSmCreate (smObjLocalToGlobal (pMem), CHUNK_SIZE))
== NULL)
return (ERROR);
/* allocate memory from partition */
pSharedBuff = (SHARED_BUFF *) memPartAlloc ( smMemPartId,
sizeof (SHARED_BUFF));
if (pSharedBuff == 0)
return (ERROR);
/* initialize structure before adding to database */
if ((mySemSmId = semBSmCreate (SEM_Q_FIFO, SEM_EMPTY)) == NULL)
return (ERROR);
pSharedBuff->semSmId = (SEM_ID) htonl ((int) mySemSmId);
/* enter shared partition ID in name database */
if (smNameAdd (MEM_PART_NAME, (void *) smMemPartId, T_SM_PART_ID) == ERROR)
return (ERROR);
/* convert shared buffer address to a global address and add to database */
if (smNameAdd (PART_BUFF_NAME, (void *) smObjLocalToGlobal(pSharedBuff),
T_SM_BLOCK) == ERROR)
return (ERROR);
739
16
VxWorks
Kernel Programmer's Guide, 6.6
/* send data using shared buffer */
sprintf (pSharedBuff->buff,"Hello from sender\n");
if (semGive (mySemSmId) != OK)
return (ERROR);
return (OK);
}
/* memPartReceive.c - shared memory partition example receive side */
/* This file reads from the user-defined shared memory partition. */
#include
#include
#include
#include
#include
#include
#include
"memPartExample.h"
/*
* memPartReceive - receive shared memory partition buffer
*
* execute on CPU 1 - use a shared semaphore to protect shared memory
*/
STATUS memPartReceive (void)
{
SHARED_BUFF * pBuff;
SEM_ID
mySemSmId;
int
objType;
/* get shared buffer address from name database */
if (smNameFind (PART_BUFF_NAME, (void **) &pBuff, &objType,
WAIT_FOREVER) == ERROR)
return (ERROR);
/* convert global address of buffer to its local value */
pBuff = (SHARED_BUFF *) smObjGlobalToLocal (pBuff);
/* Grab shared semaphore before using the shared memory */
mySemSmId = (SEM_ID) ntohl ((int) pBuff->semSmId);
semTake (mySemSmId,WAIT_FOREVER);
printf ("Receiver reading from shared memory: %s\n", pBuff->buff);
semGive (mySemSmId);
return (OK);
}
740
16 Shared-Memory Objects: VxMP
16.3 System Requirements
Side Effects of Shared-Memory Partition Options
Like their local counterparts, shared-memory partitions (both system- and
user-created) can have different options set for error handling; see the API
references for memPartOptionsSet( ) and smMemOptionsSet( ).
If the MEM_BLOCK_CHECK option is used in the following situation, the system
can get into a state where the memory partition is no longer available. If a task
attempts to free a bad block and a bus error occurs, the task is suspended. Because
shared semaphores are used internally for mutual exclusion, the suspended task
still has the semaphore, and no other task has access to the memory partition. By
default, shared-memory partitions are created without the MEM_BLOCK_CHECK
option.
16.3 System Requirements
The system requirements for VxMP include the following:
■
shared memory visible to all CPUs
■
test-and-set (TAS) operations across CPUs
■
inter-CPU notification facilities
■
a maximum of 20 CPUs (10 is the default)
16
Shared Memory Visibility
The shared-memory region used by shared-memory objects must be visible to all
CPUs in the system. Either dual-ported memory on the master CPU (CPU 0) or a
separate memory board can be used. The shared-memory objects’ anchor must be
in the same address space as the shared-memory region. Note that the memory
does not have to appear at the same local address for all CPUs. For information
about the shared memory anchor, see 16.5.4 Shared-Memory Anchor, p.745.
Test-and-Set Cycle
All CPUs in the system must support indivisible test-and-set operations across the
(VME) bus. The indivisible test-and-set operation is used by the spinlock
mechanism to gain exclusive access to internal shared data structures. Because all
741
VxWorks
Kernel Programmer's Guide, 6.6
the boards must support hardware test-and-set, the parameter SM_TAS_TYPE
must be set to SM_TAS_HARD.
!
CAUTION: Boards that make use of VxMP must support hardware test-and-set
(indivisible read-modify-write cycle). PowerPC is an exception; see the VxWorks
Architecture Supplement.
Inter-CPU Notification
CPUs must be notified of any event that affects them. The preferred method is for
the CPU initiating the event to interrupt the affected CPU. The use of interrupts is
dependent on the capabilities of the hardware. If interrupts cannot be used, a
polling scheme can be employed, although it generally results in a significant
performance penalty. For information about configuration for different
CPU-notification facilities, see 16.5.3 Mailbox Interrupts and Bus Interrupts, p.745
Maximum Number of CPUs
The maximum number of CPUs that can use shared-memory objects is 20 (CPUs
numbered 0 through 19; the default is 10). For information about configuring
VxMP for the number of CPUs in a system, see 16.5.1 Maximum Number of CPUs,
p.744.
16.4 Performance Considerations
The performance of a multi-processor system using shared memory objects can be
affected by the operation of VxMP spinlocks and well as by the use of both VxMP
and shared-memory network facilities in the same system.
Spinlock Operation
The performance of a system using shared memory objects can be affected by the
operation of spinlocks, which are used internally for cross-processor
synchronization. The spinlocks may need to be tuned for proper operation, and
interrupt latency is increased while spinlocks are held. However, spinlocks are
used only for very short periods of time to protect critical regions (in a manner
similar to the use of interrupt locking on uniprocessor systems.
742
16 Shared-Memory Objects: VxMP
16.4 Performance Considerations
Internal shared-memory object data structures are protected against concurrent
access by a spinlock mechanism. The spinlock mechanism operates as a loop in
which an attempt is made to gain exclusive access to a resource (in this case an
internal data structure). An indivisible hardware test-and-set operation is used for
this mutual exclusion. If the first attempt to take the lock fails, multiple attempts
are made.
For the duration of the spinlock, interrupts are disabled to avoid the possibility of
a task being preempted while holding the spinlock. As a result, the interrupt
latency of each processor in the system is increased. However, the interrupt
latency added by shared-memory objects is constant for a particular CPU.
For more information about and spinlocks and performance tuning, see
16.7 Troubleshooting, p.752.
Shared-Memory Objects and Shared-Memory Network Driver
Shared-memory objects and the shared-memory network2 use the same memory
region, anchor address, and interrupt mechanism. Both facilities make use of the
same shared memory region, the same shared-memory anchor, the same interrupt,
some data structures, their traffic goes over the same bus, and so on. While their
software does not effectively interact, using them together can result in reduced
performance.
If the two facilities are used together, the shared-memory anchor must be
configured in the same way for each. The shared-memory anchor is a location
accessible to all CPUs on the system. It stores a pointer to the shared-memory
header, a pointer to the shared-memory packet header (used by the
shared-memory network driver), and a pointer to the shared-memory object
header. For information about using the shared memory anchor with
shared-memory objects, see 16.5.4 Shared-Memory Anchor, p.745.
For information about shared-memory network, see the Wind River Network Stack
for VxWorks 6 Programmer’s Guide.
2. Also known as the backplane network.
743
16
VxWorks
Kernel Programmer's Guide, 6.6
16.5 Configuring VxWorks for Shared Memory Objects
To include VxMP shared-memory objects in VxWorks, configure the operating
system with the INCLUDE_SM_OBJ component.
The configuration of VxMP itself involves the following elements:
■
number of CPUs—all nodes
■
cache coherency—on a node-by-node basis
■
mailbox and bus interrupts—on a node-by-node basis
■
shared memory anchor—all nodes
■
shared memory region—master node only
■
number of shared memory objects—master node only
■
dual-port or external memory—master node only
Detailed descriptions of each of these elements, as well as examples of system
configurations, are provided in the following sections.
16.5.1 Maximum Number of CPUs
The maximum number of CPUs that can use shared-memory objects is 20 (CPUs
numbered 0 through 19). This limitation is imposed by the VMEbus hardware
itself. The practical maximum is usually a smaller number that depends on the
CPU, bus bandwidth, and application. The number is set with the SM_CPUS_MAX
configuration parameter of the INCLUDE_SM_COMMON component. By default it
is set to 10. Note that if the number is set higher than the number of boards that are
actually going to be used, it will waste memory.
16.5.2 Cache Coherency
When dual-ported memory is used on some boards without MMU or bus
snooping mechanisms, the data cache must be disabled for the shared-memory
region on the master CPU.
If you see the following runtime error message, make sure that the
INCLUDE_CACHE_ENABLE component is not included in the VxWorks
configuration:
usrSmObjInit - cache coherent buffer not available. Giving up.
744
16 Shared-Memory Objects: VxMP
16.5 Configuring VxWorks for Shared Memory Objects
Additional configuration is sometimes required to make the shared memory
non-cacheable, because the shared-memory pool is accessed by all processors on
the backplane. By default, boards with an MMU have the MMU turned on. With
the MMU on, memory that is off-board must be made non-cacheable. This is done
using the data structure sysPhysMemDesc in sysLib.c. This data structure must
contain a virtual-to-physical mapping for the VME address space used for the
shared-memory pool, and mark the memory as non-cacheable. (Most BSPs include
this mapping by default.) For more information, see 6.9.1 Configuring Virtual
Memory Management, p.344.
!
CAUTION: For the MC68K in general, if the MMU is off, data caching must be
turned off globally; see the API reference for cacheLib.
16.5.3 Mailbox Interrupts and Bus Interrupts
Two types of interrupts are supported for inter-CPU communication: mailbox
interrupts and bus interrupts. The interrupt type is specified with the
INCLUDE_SM_COMMON component parameter SM_INT_TYPE. Mailbox
interrupts (SM_INT_MAILBOX) are the preferred method, and bus interrupts
(SM_INT_BUS) are the second choice.
If interrupts cannot be used, a polling scheme can be employed (SM_INT_NONE),
but it is much less efficient.
When a CPU initializes its shared-memory objects, it defines the interrupt type as
well as three interrupt arguments. These describe how the CPU is notified of
events. These values can be obtained for any attached CPU by calling
smCpuInfoGet( ).
The default interrupt method for a target is defined with the SM_INT_TYPE,
SM_INT_ARG1, SM_INT_ARG2, and SM_INT_ARG3 parameters.
16.5.4 Shared-Memory Anchor
One of the most important aspects of configuring shared-memory objects is
defining the address of the shared-memory anchor.
Determining the address is specific to the target architecture. For example,
PowerPC leaves space at 0x4100 in its memory map to store the anchor on the
master. Most of the other architectures do the same thing at 0x600. This is defined
745
16
VxWorks
Kernel Programmer's Guide, 6.6
in VxWorks architecture-dependent header files, which ensure that the kernel
reserves memory at that location.
On PowerPC, assume for a dual-ported system that the VMEbus is seen at address
0x10000 on the master, and 0x20000 on a slave. The master can either see the
anchor at 0x4100 or 0x14100, and the slave sees it at 0x24100.
The shared-memory anchor is a location accessible to all CPUs on the system. The
anchor stores a pointer to the shared-memory header, a pointer to the
shared-memory packet header (used by the shared-memory network driver), and
a pointer to the shared-memory object header.
If the default value for the shared-memory anchor address is modified, the anchor
must be on a 256-byte boundary.
The address of the shared memory anchor can be defined statically with the
SM_ANCHOR_ADRS configuration parameter, or dynamically with the sm boot
loader parameter (sm=anchorAddress). If the boot loader parameter is defined, it
takes precedence over the static configuration.
NOTE: The shared memory anchor is used by both VxMP and the shared-memory
network driver (if both are included in the system). For information about using
VxMP and the shared memory network driver at the same time, see Shared-Memory
Objects and Shared-Memory Network Driver, p.743.
16.5.5 Shared-Memory Region
Shared-memory objects rely on a shared-memory region that is visible to all
processors. This region is used to store internal shared-memory object data
structures and the shared-memory system partition.
The shared-memory region is usually in dual-ported RAM on the master CPU, but
it can also be located on a separate memory card. The shared-memory region
address is defined as an offset from the shared-memory anchor address
(SM_ANCHOR_ADRS), as illustrated in Figure 16-3. The default is 0x600 for all
architectures but PowerPC, which is 0x4100.
746
16 Shared-Memory Objects: VxMP
16.5 Configuring VxWorks for Shared Memory Objects
Figure 16-3
Shared-Memory Layout
SHARED MEMORY
SM_ANCHOR_ADRS
0x600 (default)
.
.
.
Shared-Memory
Anchor
pointer to shared-memory
objects’ shared-memory region
~
~
~
~
shared-memory objects
Shared-Memory
Region
16.5.6 Numbers of Shared Memory Objects
The configuration parameters of the INCLUDE_SM_OBJ component are used to set
the maximum number of different types of objects. These constants are described
in Table 16-5.
Table 16-5
Configuration Constants for Shared-Memory Objects
16
Configuration Parameter
Default
Value
Description
SM_OBJ_MAX_TASK
40
Maximum number of tasks using
shared-memory objects.
SM_OBJ_MAX_SEM
60
Maximum number of shared semaphores
(counting and binary).
SM_OBJ_MAX_NAME
100
Maximum number of names in the name
database.
SM_OBJ_MAX_MSG_Q
10
Maximum number of shared message
queues.
SM_OBJ_MAX_MEM_PART
4
Maximum number of user-created
shared-memory partitions.
If the size of the objects created exceeds the shared-memory region, an error
message is displayed on CPU 0 during initialization.
747
VxWorks
Kernel Programmer's Guide, 6.6
At runtime, VxMP sets aside memory for the configured number of objects, and
then uses what is left over for the shared memory system partition.
16.5.7 Dual-Port or External Memory
The key distinction between configuration for dual-port and for external memory
is as follows:
■
For dual-port memory, the SM_OFF_BOARD parameter is set to FALSE for the
master CPU and to TRUE for all slave CPUs.
■
For external memory, the SM_OFF_BOARD parameter is set to TRUE for all
CPUs (master and slave CPUs).
The following sections describe configurations for each type of memory use.
Configuration for Dual-Port Memory
The configuration illustrated in Figure 16-4 uses the shared memory in the master
CPU’s dual-ported RAM.
Figure 16-4
Dual-Ported Memory
CPU 0
CPU 1
sm=0x1800600
RAM
anchor
0x600
allocated
pool
Local address of
VMEbus address 0
is 0x1000000
VMEbus address of dual
ported RAM = 0x800000
In this example, the settings for the master (CPU 0) are as follows: the
SM_OFF_BOARD parameter is FALSE and SM_ANCHOR_ADRS is 0x600 (the value
is specific to the processor architecture), SM_OBJ_MEM_ADRS is set to NONE,
748
16 Shared-Memory Objects: VxMP
16.5 Configuring VxWorks for Shared Memory Objects
because on-board memory is used (it is dynamically allocated at runtime);
SM_OBJ_MEM_SIZE is set to 0x20000.
NOTE: When using dual-ported memory, the shared memory can be allocated
from the master’s kernel heap by setting SM_OBJ_MEM_ADRS to NONE. Note,
however, the following caveats in this regard:
■
The entire kernel heap must be mapped onto the shared bus address space
(that is, the VMEbus address space), since the memory can be allocated from
anywhere within it. The memory space of the shared bus mapped on the slaves
must also be large enough to see the whole heap.
■
Mapping the entire kernel heap might mean mapping the entire kernel or the
entire local RAM of the master. In this case, there is the potential risk of a
malfunctioning remote target overwriting critical kernel text and data
structures on the master.
Wind River recommends the alternative of assigning a static address to
SM_OBJ_MEM_ADRS and mapping only SM_OBJ_MEM_SIZE bytes of local RAM
onto the shared bus memory space.
For the slave (CPU 0) in this example, the board maps the base of the VME bus to
the address 0x1000000. SM_OFF_BOARD is TRUE and the anchor address is
0x1800600. This is calculated by taking the VMEbus address (0x800000) and
adding it to the anchor address (0x600). Many boards require further address
translation, depending on where the board maps VME memory. In this example,
the anchor address for the slave is 0x1800600, because the board maps the base of
the VME bus to the address 0x1000000.
Configuration for External Memory
In the configuration illustrated in Figure 16-5, the shared memory is on a separate
memory board.
749
16
VxWorks
Kernel Programmer's Guide, 6.6
Figure 16-5
External Memory Board
CPU 0
CPU 1
anchor = 0x3000000
sm=0x2100000
Local address of
VMEbus address 0
is 0x1000000
Local address of
VMEbus address 0
is 0x100000
External RAM
Board (1MB)
anchor
shared-memory
pool
VMEbus address
of RAM on external
board = 0x2000000
For the master (CPU 0) in this example, the SM_OFF_BOARD parameter is TRUE,
SM_ANCHOR_ADRS is 0x3000000, SM_OBJ_MEM_ADRS is set to
SM_ANCHOR_ADRS, and SM_OBJ_MEM_SIZE is set to 0x100000.
For the slave (CPU 1), SM_OFF_BOARD is TRUE and the anchor address is
0x2100000. This is calculated by taking the VMEbus address of the memory board
(0x2000000) and adding it to the local VMEbus address (0x100000).
16.5.8 Configuration Example
This section describes the configuration settings for a multiprocessor system with
three CPUs and dual-ported memory.
The master is CPU 0, and shared memory is configured from its dual-ported
memory. This application has 20 tasks using shared-memory objects, and uses 12
message queues and 20 semaphores. The maximum size of the name database is
the default value (100), and only one user-defined memory partition is required.
On CPU 0, the shared-memory pool is configured to be on-board. This memory is
allocated from the processor’s system memory. On CPU 1 and CPU 2, the
shared-memory pool is configured to be off-board. Table 16-6 shows the
parameter values set for the INCLUDE_SM_OBJ and INCLUDE_SM_COMMON
750
16 Shared-Memory Objects: VxMP
16.5 Configuring VxWorks for Shared Memory Objects
components. Note that for the slave CPUs, the value of SM_OBJ_MEM_SIZE is not
actually used.
Table 16-6
Configuration Settings for Three CPU System
CPU
Configuration Parameter
Value
Master (CPU 0)
SM_OBJ_MAX_TASK
20
SM_OBJ_MAX_SEM
20
SM_OBJ_MAX_NAME
100
SM_OBJ_MAX_MSG_Q
12
SM_OBJ_MAX_MEM_PART
1
SM_OFF_BOARD
FALSE
SM_MEM_ADRS
NONE
SM_MEM_SIZE
0x10000
SM_OBJ_MEM_ADRS
NONE
SM_OBJ_MEM_SIZE
0x10000
SM_OBJ_MAX_TASK
20
SM_OBJ_MAX_SEM
20
SM_OBJ_MAX_NAME
100
SM_OBJ_MAX_MSG_Q
12
SM_OBJ_MAX_MEM_PART
1
SM_OFF_BOARD
TRUE
SM_ANCHOR_ADRS
(char *) 0xfb800000
SM_MEM_ADRS
SM_ANCHOR_ADRS
SM_MEM_SIZE
0x10000
SM_OBJ_MEM_ADRS
NONE
SM_OBJ_MEM_SIZE
0x10000
Slaves (CPU 1, CPU 2)
16
751
VxWorks
Kernel Programmer's Guide, 6.6
16.6 Displaying Information About Shared Memory Objects
The routine smObjShow( ) can be used to display the current number of used
shared-memory objects and other statistics, as follows:
-> smObjShow
value = 0 = 0x0
The smObjShow( ) routine is automatically included if VxWorks is configured
with the INCLUDE_SM_OBJ component.
The output of smObjShow( ) is sent to the standard output device, and looks like
the following:
Shared Mem Anchor Local Addr : 0x600
Shared Mem Hdr Local Addr
: 0x363ed0
Attached CPU
: 2
Max Tries to Take Lock
: 0
Shared Object Type
Current
Maximum
-----------------------------Tasks
1
40
Binary Semaphores
3
30
Counting Semaphores
0
30
Messages Queues
1
10
Memory Partitions
1
4
Names in Database
5
100
!
Available
--------39
27
27
9
3
95
CAUTION: If the master CPU is rebooted, it is necessary to reboot all the slaves. If
a slave CPU is to be rebooted, it must not have tasks pended on a shared-memory
object.
16.7 Troubleshooting
Problems with shared-memory objects can be due to a number of causes. This
section discusses the most common problems and a number of troubleshooting
tools. Often, you can locate the problem by rechecking your hardware and
software configurations.
752
16 Shared-Memory Objects: VxMP
16.7 Troubleshooting
16.7.1 Configuration Problems
Use the following list to confirm that your system is properly configured:
■
Be sure to verify that VxWorks is configured with the INCLUDE_SM_OBJ
component for each processor using VxMP.
■
Be sure the anchor address specified is the address seen by the CPU. This can
be defined statically (with the SM_ANCHOR_ADRS configuration parameter),
or at boot time (with the sm boot loader parameter) if the target is booted with
the shared-memory network.
■
If there is heavy bus traffic relating to shared-memory objects, bus errors can
occur. Avoid this problem by changing the bus arbitration mode or by
changing relative CPU priorities on the bus.
■
If memAddToPool( ), memPartSmCreate( ), or smMemAddToPool( ) fail,
check that any address you are passing to these routines is in fact a global
address.
■
If applications create more than the specified maximum number of objects, it
is possible to run out of memory. If this happens, the shared object creation
routine returns an error and errno is set to S_memLib_NOT_ENOUGH_MEM.
To solve this problem, first increase the maximum number of shared-memory
objects of corresponding type; see Table 16-5 for a list of the applicable
configuration parameters. This decreases the size of the shared-memory
system pool because the shared-memory pool uses the remainder of the
shared memory. If this is undesirable, increase both the number of the
corresponding shared-memory objects and the size of the overall
shared-memory region, SM_OBJ_MEM_SIZE. See 16.5 Configuring VxWorks for
Shared Memory Objects, p.744 for a discussion of configuration parameters.
■
Operating time for the spinlock cycle can vary greatly because it is affected by
the processor cache, access time to shared memory, and bus traffic. If the lock
is not obtained after the maximum number of tries specified by the
SM_OBJ_MAX_TRIES parameter), errno is set to
S_smObjLib_LOCK_TIMEOUT. If this error occurs, set the maximum number
of tries to a higher value. Note that any failure to take a spinlock prevents
proper functioning of shared-memory objects. In most cases, this is due to
problems with the shared-memory configuration (see above).
753
16
VxWorks
Kernel Programmer's Guide, 6.6
16.7.2 Troubleshooting Techniques
Use the following techniques to troubleshoot any problems you encounter:
■
The routine smObjTimeoutLogEnable( ) enables or disables the printing of an
error message indicating that the maximum number of attempts to take a
spinlock has been reached. By default, message printing is enabled.
■
The routine smObjShow( ) displays the status of the shared-memory objects
facility on the standard output device. It displays the maximum number of
tries a task took to get a spinlock on a particular CPU. A high value can
indicate that an application might run into problems due to contention for
shared-memory resources. For information about smObjShow( ), see
16.6 Displaying Information About Shared Memory Objects, p.752 and the API
reference for the routine.
■
The shared-memory heartbeat can be checked to verify that the master CPU
has initialized shared-memory objects. The shared-memory heartbeat is in the
first 4-byte word of the shared-memory object header. The offset to the header
is in the sixth 4-byte word in the shared-memory anchor. (See the Wind River
Network Stack for VxWorks 6 Programmer’s Guide.)
Thus, if the shared-memory anchor were located at 0x800000:
[VxWorks Boot]: d
800000: 8765 4321
800010: 0000 0000
800020: 0000 0000
0x800000
0000 0001 0000 0000 0000 002c *.eC!...........,*
0000 0170 0000 0000 0000 0000 *...p............*
0000 0000 0000 0000 0000 0000 *................*
The offset to the shared-memory object header is 0x170. To view the
shared-memory object header display 0x800170:
[VxWorks Boot]: d 0x800170
800170: 0000 0050 0000 0000 0000 0bfc 0000 0350 *...P...........P*
In the preceding example, the value of the shared-memory heartbeat is 0x50.
Display this location again to ensure that the heartbeat is alive; if its value has
changed, shared-memory objects are initialized.
754
17
Distributed Shared Memory:
DSHM
17.1 Introduction 755
17.2 Technology Overview 756
17.3 Configuring VxWorks for DSHM 763
17.4 Developing Custom Services 767
17.5 Developing a Hardware Interface 777
17.1 Introduction
The VxWorks distributed shared memory (DSHM) facility is a middleware
subsystem that allows multiple services to communicate over different types of
buses that support shared-memory communication. DSHM provides two main
features for the services that make use of distributed shared memory: messaging
over shared memory, and allocation of shared memory resources to services for
their use in writing custom data. DSHM currently provides optional support for
TIPC for communication over shared memory media.
Custom services can be developed for use with DSHM, and custom DSHM
hardware interfaces can be developed for hardware that is not supported currently
by Wind River. This chapter provides information about how to pursue both of
these development activities.
755
VxWorks
Kernel Programmer's Guide, 6.6
17.2 Technology Overview
On VxWorks the term shared memory refers to a variety of software and hardware
implementations, in which each node in a multi-node cluster has access to a shared
memory region in addition to its own private memory. (For the purposes of
discussion in this chapter, the term node refers to a single instance of an operating
system).
In contrast to DSHM, the legacy Wind River shared memory technology is
implemented with a master-slave model, in which the memory that is shared
between nodes is located exclusively on a master node or external dumb board.
That is, one centralized block of memory is shared, and one node is responsible for
its initialization and management. The master node is involved to a certain extent
in all transactions when the shared memory is local to its board, since remote nodes
must read from its local memory. (For information about legacy systems, see Wind
River TIPC for VxWorks 6 Programmer's Guide for information about TIPC over
shared memory, and 16. Shared-Memory Objects: VxMP for information about
VxMP shared memory objects.)
The architecture of DSHM ensures superior reliability, determinism, and
performance in comparison with the more conventional master-slave model. To
do so, DSHM makes use of memory that is truly distributed across all nodes
participating in a system. This distribution of memory can be accomplished in a
variety of ways, including the following:
■
Each node in a system has a pool of dedicated memory that is shared on the
common bus, and all nodes can write to each of their peers’ memory pools.
■
A dumb board in a VME chassis provides a dedicated memory pool for each
node.
■
A multi-core board with one physical RAM DIMM provides a virtual
distribution of memory; each OS instance allocates part of it as its own shared
memory pool.
DSHM provides both a messaging service and facilities for allocation of shared
memory resources to services for writing custom data messaging service. The
messaging service is also used for passing data itself, such as network packets.
Currently the messaging service is only supported over shared memory, although
the DSHM API allows for expansion to other types of messaging.
The messaging protocol provides the means for sending simple messages between
nodes, such as notification that a peer has come up or gone down, that a packet has
been sent, or that a remote node has thrown a synchronization event. A message
can be used like a semaphore. For example, Node 1 fills a buffer on Node 2, and
756
17 Distributed Shared Memory: DSHM
17.2 Technology Overview
then sends a message signalling the arrival of the data. Then it pends, waiting for
the buffer to be available again. When the buffer is available, Node 2 then sends a
message signalling this event.
17.2.1 Architecture
Figure 17-1 illustrates the architecture of the VxWorks distributed shared memory
facility. This architecture is designed for a system in which memory is distributed
across all nodes in the system.
Figure 17-1
Distributed Shared Memory Architecture
Applications
DSHM Services
DSHM API
DSHM MUX
17
DHSM
Adaptation
Layer
DSHM
Utilities
DSHM Hardware Interface
Hardware Buses
757
VxWorks
Kernel Programmer's Guide, 6.6
The DSHM elements illustrated in Figure 17-1 are as follows:
■
Services make use of DSHM facilities (messaging over shared memory and
allocation of shared memory resources for writing custom data) to support
applications. The DSHM management service performs node and resource
management activities.
■
The DSHM API layer is the interface to services.
■
The DSHM MUX (multiplexor) layer is the heart of the DSHM system. It
provides the send and receive APIs, service query and access routines, and
hardware query and access functionality. The MUX allows for multiple
hardware elements and services to register with DSHM. It routes outgoing
messages to the correct hardware bus on which they are transmitted, and
routes incoming messages to the correct service for processing.
■
The DSHM utilities are used internally by the DSHM implementation but are
available to service and hardware interface writers.
■
The DSHM adaptation layer provides macros and functions to OS-agnostic
code that needs to make OS-specific calls.
■
The hardware interface implements hardware-dependent features.
DSHM Management Service and Custom Services
Services interface directly with the DSHM facility. They are generally hidden from
the user. For example, the TIPC bearer accessed by way of the socket API
interacting with the TIPC stack.
The DSHM management service runs on each node in the system and handles the
messages that deal with nodes appearing and disappearing, management of
resources, and so on. One instance of the service runs on each node in the system
as service number zero. Each is registered as a service when a hardware interface
is ready to handle incoming messages.
Custom services can written based on the DSHM APIs. For information in this
regard, see 17.4 Developing Custom Services, p.767.
Hardware Interface
The hardware interface can implement hardware-dependent features, such as
hardware messaging functionality provided by modern hardware (for example,
RapidIO messaging). It currently implements shared memory messaging for cases
758
17 Distributed Shared Memory: DSHM
17.2 Technology Overview
in which hardware-assisted messaging is not available. For information about
developing a hardware interface, see 17.5 Developing a Hardware Interface, p.777.
17.2.2 Communication Model
Figure 17-2 illustrates the flow of messages and data using distributed shared
memory in a two-node system using a TIPC bearer. In this example, the DSHM
service knows that when it receives a message, it is intended for the TIPC bearer.
The DSHM MUX routes the message to that service. Then, if the message is a
DSHM_SVC_NET_SEND message, the TIPC bearer delivers the DSHM buffer to the
TIPC stack (the message could also be a management message for the DSHM
bearer, that does not go to TIPC). This type of message is associated with a buffer
in shared memory, with contents that are identified as a TIPC message. Only the
contents of the buffer are delivered to the TIPC stack. From that point on, DSHM
does not have anything to do with the buffer. TIPC takes care of delivering it to the
application, if it is application data (the contents could be a TIPC management
message).
By way of analogy, you can think of DSHM as replacing the ethernet portion, the
physical layer, of a network. DSHM performs a role similar to an ethernet device
when it receives data. An ethernet device receives data in ethernet frame format.
The software managing the device receives the physical data in a buffer in memory
that it provided to the device. It then takes ownership of that buffer and replaces
it in the device with another one (if available). Then, that buffer may be formatted,
and then given to whomever it was sent to. With VxWorks END devices,
networking MUX routes it to the correct protocol stack, which then routes it to the
correct socket, which delivers it to the correct application.
Broadcasting
There are two ways of accomplishing broadcasting of data packets with DSHM:
true broadcast and replicast.
True broadcast
The true broadcast implementation uses the broadcasting facility of the underlying
hardware interface implementation, coupled with the fact that the local shared
memory can also be read by remote nodes, on top of being written. The idea is to
have a certain memory pool that is written by the broadcasting node, where the
data packet is copied to. The node then sends a broadcast message of a certain type
that specifies that there is a broadcast data packet to be read from the broadcasting
759
17
VxWorks
Kernel Programmer's Guide, 6.6
node. This is sent using the broadcast address, in essence making use of the
hardware interface's capability to send broadcast messages efficiently. When a
remote node receives the message, it reads the data packet from the broadcasting
node's shared memory. Some kind of mutual exclusion between the nodes might
have to be used to prevent reusing (rewriting) the packet before all remote nodes
have read it.
This method of broadcasting should be used whenever possible since it puts less
burden on the broadcasting node, in effect sharing it on all receiving nodes.
However, it might be impossible to use in certain cases. The TIPC DSHM bearer,
when using shared memory messaging where a send operation can fail because of
a full message queue, is an example of this. If the message sending operation
cannot fail on a particular hardware interface implementation, it can be used.
Replicast
Replicasting is based on the concept of putting the burden of sending the broadcast
packet on the sender. The broadcasting node in effect has to obtain a buffer for
every node the broadcast is destined to, and copy the data into them. Then, a
unicast message has to be send to each one of them. It can be the same type as a
regular message signalling the arrival of a data packet since they are in essence the
exact same thing.
The TIPC bearer uses this type of broadcasting when running over the default
messaging implementation over shared memory, to be able to regulate the amount
of messages that are sent to each node and thus ensure that sending a message, in
a sane system, will always succeed. This allows for better flow control.
760
17 Distributed Shared Memory: DSHM
17.2 Technology Overview
Figure 17-2
Distributed Shared Memory Send Operation
1. Request buffer
Node 0
Node 1
2. Lend buffer
RAM
RAM
3. Write to buffer
Shared Memory
Shared Memory
4. Signal
Message
Data Flow
17
761
VxWorks
Kernel Programmer's Guide, 6.6
Send Operation
A send operations, as illustrated in Figure 17-2, takes place as follows:
1.
Node 0 requests a buffer used for writing from Node 1.
DSHM_BUILD(msg, DSHM_VNIC, 0, 1, DSHM_VNIC_REQUEST);
dshmMuxMsgSend (hwID, msg, 0, 0);
The DSHM_BUILD call fills the message header with the appropriate values
The arguments (defined in dshm.h) are as follows: message (an array of type
char; the size of which is defined in dshm.h), service number, source,
destination, and type of message.
Then Node 0 waits for buffer.
2.
Node 1 allocates a buffer and informs Node 0 that it is available.
DSHM_BUILD(msg, DSHM_VNIC, 1, 0, DSHM_VNIC_ALLOC);
DSHM_DAT32_SET(1, pBuffer);
dshmMuxMsgSend (hwID, msg, 0, 0);
3.
Node 0 writes to the buffer.
pBuffer = DSHM_DAT32_GET(msg, 1);
bcopy (pPacket, pBuffer, sizeof(pPacket);
4.
Node 0 sends message to Node 1 telling it that there telling it that there is data
in the buffer.
DSHM_BUILD(msg, DSHM_VNIC, 0, 1, DSHM_VNIC_SEND);
DSHM_DAT32_SET(1, pBuffer);
Then remote node reads the message and passes the packet to the stack.
For information about the C macro functions used in this example, see
17.4.2 DSHM Messaging Protocols and Macro Functions, p.769.
762
17 Distributed Shared Memory: DSHM
17.3 Configuring VxWorks for DSHM
Broadcast Operation
A broadcast operation involves the following steps (custom APIs required):
1.
Node 2 obtains a buffer from its stack.
2.
Node 2 writes to the buffer in local shared memory.
3.
Node 2 broadcasts a message to remote nodes.
4.
Node 0 and Node 1 receive the message, and obtain a broadcast tag, read the
remote share memory to get the data, and the buffer is passed up the stack.
A broadcast tag is part of the message. It is is a 32-bit value (DAT32), and is the
second 32-bit value in the message (at index 1, which explains the 1 in the call).
For information about the macros and APIs used in this example, see 17.4.2 DSHM
Messaging Protocols and Macro Functions, p.769 and 17.4.3 DSHM Service APIs,
p.770.
17.3 Configuring VxWorks for DSHM
VxWorks can be configured and built with DSHM using the standard Wind River
Workbench and vxprj facilities.
Workbench VxWorks Image Projects (VIPs) provide an asymmetric
multiprocessing (AMP) option for creating projects with DSHM. The vxprj
provides the -amp option for configuring and building VxWorks with DSHM from
the command line. For example:
vxprj create -amp hpcNet8641 diab
!
CAUTION: Boot loaders must not be built with the AMP build option—neither
with Workbench nor with vxprj. For more information about boot loaders, see
3. Boot Loader.
763
17
VxWorks
Kernel Programmer's Guide, 6.6
17.3.1 Components and Parameters
DSHM componentization is divided into multiple areas, some mandatory, some
removable, and some hardware dependant.
Core support
These are needed for correct functionality in a VxWorks system. They should be
pulled in automatically by any other component that depends on DSHM.
INCLUDE_DSHM
BSP support enabling. This component is only present in BSPs that support
DSHM. This is because DSHM needs some specific hardware to run, namely
the possibility of having shared memory across VxWorks instances.
MUX. Allows multiple services to use the messaging system over the same
medium, such as the shared memory between two cores on a multi-core chip.
It also allows the usage of the same DSHM API by multiple concurrent media
(for example, if there was a multi-core AMP system on a VME blade).
The INCLUDE_DSHM parameters are as follows:
■
Maximum instances of hw buses (DSHM_MAX_HW). Unless there are two
concurrent buses in your system, this should always be 1.
■
Maximum number of services per bus (DSHM_MAX_SERVICES). The
default is 2: one for the management service, and one for a user service,
which might be one provided by Wind River, such as the TIPC bearer. If
you intend to run more than one service concurrently, increase this value.
INCLUDE_DSHM_ADAPT
VxWorks adaptation layer. DSHM is meant to be portable to other operating
systems. This component is the adaptation layer for VxWorks.
Distributed Multi-Processing Messaging
This is the selection of the messaging type on a particular bus if more than one is
available. Currently, only shared memory messaging
(INCLUDE_DSHM_MSG_SM) is available. It should always be selected.
Peer-to-Peer Drivers
These are the drivers that implement the shared memory messaging and
housekeeping of the shared memory used for data passing. They are different for
each BSP that provides support for DSHM. However, some parts are shared
between implementations.
764
17 Distributed Shared Memory: DSHM
17.3 Configuring VxWorks for DSHM
Note that the driver implementation is a VxBus implementation. VxBus will
always get added to the VxWorks image if it is not already selected.
INCLUDE_DSHM_BUS
DSHM virtual bus. This should always be selected. It provides the framework
for drivers.
INCLUDE_DSHM_BUS_PLB
DSHM virtual bus on PLB. The current implementation all are for a
Processor-local-bus type bus, such as the one on a multi-core chip. Even the
VxWorks simulator implementation follows that model, as if the VxWorks
simulator instances would share a local bus. This must be selected. It is also
where most of the parameters live. The parameters are as follows:
DSHM_BUS_PLB_NODE
Address of local node. This is the unique address on the shared bus. The
current drivers are able to find their own address at runtime, using the
processor number assigned at boot time with the boot line. Use -1 for this,
or another number to force a specific one for a specific image. The address
must be less than the next parameter.
DSHM_BUS_PBL_MAXNODES
Maximum number of nodes. There can be no more nodes than this value
in the system. Note that all nodes must agree on that value so that it works
as intended.
DSHM_BUS_PLB_NENTRIES
Number of entries in the shared memory. Each message sent over DSHM
is sent asynchronously and takes up one entry in the message queue.
When the queue is full, if another message is sent, an error code is returned
to the sender. For implementation reasons, the real number of concurrent
messages is actually one less than this number.
DSHM_BUS_PLB_NRETRIES
Number of retries. When trying to send a message, the internals will
actually retry sending in the case where the queue is full. If you do not
want that to happen, you can set it to 0. If you would like more retries, pick
a higher number. It can help getting less sending errors. WARNING: This
is a ‘busy’ retry, in effect hogging the CPU.
DSHM_BUS_PLB_RMW
Read-modify-write routine. This is per-bus type. It should be left alone.
DSHM_BUS_PLB_POOLSIZE
Shared memory pool size. If you decide to share more or less memory on
this node, adjust this number accordingly.
765
17
VxWorks
Kernel Programmer's Guide, 6.6
DSHM_BUS_PLB_ENTRY_SIZE
Currently unsupported.
Virtual bus controller and peer drivers are BSP-specific, but have to be selected
for DSHM to work properly. They cannot be put in as defaults since each
different hardware implementation component has a different name. The
components are as follows:
■
wrSbc8561 and hpcNet8641: INCLUDE_DSHM_BUS_CTLR_8641 and
INCLUDE_DSHM_VX8641
■
Any VxWorks simulator: INCLUDE_DSHM_BUS_CTLR_SIM and
INCLUDE_DSHM_VXSIM
■
Any sibyte board (sb1250/sb1480): INCLUDE_DSHM_BUS_CTLR_SIBYTE
and INCLUDE_DSHM_VXSIBYTE
Interface Debug Aid
INCLUDE_DSHM_DEBUG
Provides a debugging aid that allows for multiple levels of debugging output,
selectable at runtime. You can choose an initial level (DSHM_INIT_DBG_LVL).
By default, it is OFF (no message is printed). See dshm/debug/dshmDebug.h
for more information on usage.
Services
INCLUDE_DSHM_SVC_MNG
Services provided by Wind River. The node manager must be present. The
TIPC bearer is a special case and lives under the TIPC component directory.
For information about the TIPC bearer, see the Wind River TIPC Programmer’s
Guide.
Utilities
These are utilities provided for service and hardware interface writers. They are
used internally in the TIPC bearer and the hardware interface implementations
provided by Wind River. They are pulled in when needed.
766
17 Distributed Shared Memory: DSHM
17.4 Developing Custom Services
17.4 Developing Custom Services
DSHM provides macros and APIs that can be used to develop custom services.
Custom services may, for example, be developed to provide the following sorts of
features:
■
Faster data transfers—without going through TIPC or IP—by using a ring of
data buffers in shared memory. When a number of buffers are ready for
reading by the peer, the local node sends a message with a payload describing
how many buffers should be read. The peer sends a message back when the
buffers are read and ready to be filled again, also with a message payload
specifying how many buffers (to keep the amount of messages sent to a
minimum). Each side keeps a local view of the state of the system. This
implementation would be less flexible than TIPC or IP, but would also have
considerably less overhead, and potentially faster rates of data transfer. A
simplified version of this type of feature is provided in 17.4.4 Service Code
Example, p.771.
■
Synchronization between tasks across nodes by sending a message to a remote
node when a resource is available. This can simply be implemented by having
local tasks pending on a local semaphore that is then given when the message
arrives. This is similar to VxMP semaphores, but more constrained, as this is
strictly peer-to-peer communication.
■
Control and command messages that do not need any data transfer, for a
specific application. The service could support a small payload, instant
feedback as to whether the message got through or not, and so on.
Services provide functionality over a specific bus type, and make calls to the
DSHM APIs to interface with the bus (to obtain a shared memory, to send
messages, and so on).
Services should also register callback routines for events such as a node joining or
departing from a system. When an instance of a DSHM hardware interface
discovers a node, it calls dshmSvcNodeJoin( ), which calls callbacks installed by
all services that need to be notified.
When discovering a node, the hardware interface propagates that information to
all services registered on that bus. This allows services to take actions such as
initialization and allocation of data structures used for that particular node. The
services have their join callback invoked at that point. The callbacks are described
17.5.2 Callbacks, p.778.
If a node is declared gone or dead, the hardware interface instance calls
dshmSvcNodeLeave( ), which similarly calls callbacks installed by all services that
767
17
VxWorks
Kernel Programmer's Guide, 6.6
need to be notified. If a service on the local node decides to quit, it can call
dshmMuxSvcWithdraw( ), which calls a callback registered by the service to do
cleanup.The service can then call dshmMuxSvcWithdrawComplete( ) when it is
satisfied that the cleanup is completed.
Callbacks can be used to take care of allocating shared memory pools, network
buffers, and so on. The custom service writer provides the desired functionality.
17.4.1 Service Numbers
All DSHM services are identified internally by unique service numbers. Wind
River reserves zero for the DSHM management service (for information about the
service, see DSHM Management Service and Custom Services, p.758).
For the greatest efficiency, use the smallest service numbers possible, since they are
used directly as indices into arrays. Service number should also be implemented
as configurable parameters, in the event that there is a conflict with another
software provider’s usage. The numbers should also be documented if the
software is provided to a third party.
The maximum number of services is defined with the DSHM_MAX_SERVICES
parameter of the INCLUDE_DSHM component. This parameter should be set to
one more than the number of services that will be supported (that is N-1), because
the service number zero is reserved for the DSHM management service. If that
number is 3, for example) each bus can have 2 (plus the management service)
services each, which can be totally different.
768
17 Distributed Shared Memory: DSHM
17.4 Developing Custom Services
17.4.2 DSHM Messaging Protocols and Macro Functions
Each custom service must provide its own message types for its own protocol.
DSHM provides a set of C macro functions to facilitate building messages. The
macros can, for example, be used to build the message header, access each one of
the per-message type parameters, and so on. The messaging macros are as follows:
■
■
■
■
■
■
■
■
■
■
■
DSHM_DAT8_SET( )
DSHM_DAT16_SET( )
DSHM_DAT32_SET( )
DSHM_DAT_GET( )
DSHM_DAT8_GET( )
DSHM_DAT16_GET( )
DSHM_DAT32_GET( )
DSHM_SVC_GET( )
DSHM_SRC_GET( )
DSHM_DST_GET( )
DSHM_TYP_GET( )
The macros are defined in installDir/vxworks-6.x/target/h/dshm/dshm.h.
The following macros builds a correctly formatted message in the msg parameter
using the four other parameters:
#define DSHM_BUILD(msg, svc, dest, src, type)
The next macros accesses the data of a specified size at a specified offset:
17
#define DSHM_DAT[8|16|32]_[GET|SET](msg, offset)
The offset units depends on the width of data to be set or retrieved. For example,
the following call retrieves the second byte in the message:
DSHM_DAT8_GET(msg, 1)
But the following macro retrieves the second word:
DSHM_DAT32_GET(msg, 1)
If a message body is comprised of one word, one byte, one byte, one half-word and
one word, the following would retrieve each, respectively:
DSHM_DAT32_GET(msg, 0)
DSHM_DAT8_GET(msg, 4)
DSHM_DAT8_GET(msg, 5)
DSHM_DAT16_GET(msg, 3)
DSHM_DAT32_GET(msg, 2)
769
VxWorks
Kernel Programmer's Guide, 6.6
The following macros used for obtaining a pointer to a message, and for casting a
pointer to a DSHM message pointer:
DSHM(variable_name);
DSHM_TYPE(ptr_name);
DSHM_CAST(ptr);
For an example of how the macros are used, see 17.4.4 Service Code Example, p.771.
17.4.3 DSHM Service APIs
The APIs described in Table 17-1 are provided by dshmMuxLib for use by custom
services.
Table 17-1
DSHM Service Routines
Routine
Description
dshmMuxHwGet( )
Obtain an hardware registration handle based
on name.
dshmMuxHwNodesNumGet( ) Obtain the maximum number of nodes on a
hardware bus.
dshmMuxHwRmwGet( )
Obtain the atomic Read-Modify-Write routine
on this bus.
dshmMuxHwOffToAddr( )
Translate a shared memory offset to a local
address.
dshmMuxHwAddrToOff( )
Translate a local address to a shared memory
offset.
dshmMuxHwLocalAddrGet( )
Obtain address of the local node.
dshmMuxSvcRegister( )
Register a service with the MUX.
dshmMuxSvcObjGet( )
Retrieve the reference to a service object.
dshmMuxSvcWithdraw( )
Remove service from MUX.
dshmMuxMsgSend( )
Transmit a message.
770
17 Distributed Shared Memory: DSHM
17.4 Developing Custom Services
Table 17-1
DSHM Service Routines
Routine
Description
dshmMuxMsgRecv( )
Receive a message.
dshmMuxMemAlloc( )
Allocate shared memory from a specific
hardware.
Once a service is up and running, the bulk of the API calls that are used would be
dshmMuxMsgSend( ) and dshmMuxMsgRecv( ), which are called on a
one-to-one basis with the number of messages directed to the local node, as well as
dshmMuxSvcObjGet( ) and dshmMuxSvcObjRelease( ) for obtaining the object
when sending or receiving. If data buffers are exchanged and are dynamic—as
with a network driver service—dshmMuxHwAddrToOff( ) and
dshmMuxHwOffToAddr( ) would also be used for converting buffer pointer to
shared memory offsets and back. The remainder of the APIs would be used
infrequently, primarily for housekeeping functions.
17.4.4 Service Code Example
The code provided below illustrates using messages for synchronization.
/* dshmTestDemoSync.c - Demo code: synchronization between nodes */
#include
#include
#include
#include
#include
#include
17
/*
DESCRIPTION
This is an example service, where two nodes each have one task that depends
on
the remote node having finished some operation. DSHM messages are used as
synchronization events across nodes.
*/
771
VxWorks
Kernel Programmer's Guide, 6.6
/* the service object */
typedef struct _svc_obj
{
DSHM_SVC_OBJ
obj;
/* base object */
struct
{
SEM_ID
sync;
/* binary sem: recv'd sync message */
uint16_t
addr;
/* remote node address */
uint16_t
pad;
} remote;
/* only support one remote peer */
} svc_obj;
/* the service type */
#define SVC_TYPE
0x02
/* service type */
/* message types */
#define MSG_SYNC
#define MSG_JOIN
0x11
0x22
/* normal message */
/* join */
/* seconds */
#define SECONDS(x)
(x * sysClkRateGet())
/* callback prototypes */
static STATUS rx
(
svc_obj * const pObj,
DSHM(msg)
);
static STATUS join
(
svc_obj * const pObj,
const uint_t addr
);
static void leave
(
svc_obj * const pObj,
const uint_t addr
);
/* the service object */
/* message received */
/* the service object */
/* address of node joining */
/* the service object */
/* address of node leaving */
/* service hooks structure */
static const DSHM_SVC_HOOKS svchooks =
{
(STATUS(*)(void * const, DSHM_TYPE()))
(STATUS(*)(void * const, const uint_t))
(void(*)(void * const, const uint_t))
(void(*)(void * const))
};
772
rx,
join,
leave,
NULL
17 Distributed Shared Memory: DSHM
17.4 Developing Custom Services
/* worker task */
static void worker (SEM_ID sync, uint16_t hw, uint16_t svc, uint16_t remote);
/* service init */
/****************************************************************************
**
*
* dshmTestMsgStart - start test service
*
*/
void dshmTestDemoSync
(
const char * const pHwName /* hw interface name */
)
{
svc_obj *pObj;
/* the service object */
int hw;
/* hw registration number */
DSHM(msg);
/* the messages */
if ((NULL == pHwName) || (-1 == (hw = dshmMuxHwGet (pHwName))))
{
printf ("FAILED: no such hw registered\n");
return;
}
pObj = malloc (sizeof (svc_obj));
if (!pObj)
{
printf ("FAILED: out-of-memory\n");
return;
}
17
memset ((char *)pObj, 0, sizeof (svc_obj));
pObj->obj.hw = hw;
pObj->obj.svc = SVC_TYPE;
pObj->remote.addr = DSHM_ADDR_INVALID;
pObj->remote.sync = NULL;
if (dshmMuxSvcRegister (hw, SVC_TYPE, pObj, &svchooks) == ERROR)
{
printf ("FAILED: cannot register service\n");
return;
}
/* from here on, cannot use pObj directly */
/* broadcast to remote nodes that we're ready */
DSHM_BUILD (msg, SVC_TYPE, 0, DSHM_ADDR_BCAST, MSG_JOIN);
dshmMuxMsgSend (hw, msg, 0, 0);
}
773
VxWorks
Kernel Programmer's Guide, 6.6
/* service callbacks */
/****************************************************************************
**
*
* rx - invoked when receiving a message
*
* This service handles the following types of messages:
* - remote has joined: perform needed initialization.
* - remote sync: signal worker task that it can resume its work.
*/
static STATUS rx
(
svc_obj * const pObj,
DSHM(msg)
)
{
uint16_t src;
/* the service object */
/* message received */
src = DSHM_SRC_GET (msg);
switch (DSHM_TYP_GET(msg))
{
case MSG_JOIN:
/* only acknowledge if not ourselves */
if (src != dshmMuxHwLocalAddrGet (pObj->obj.hw))
{
join (pObj, src);
}
break;
case MSG_SYNC:
/* signal worker task that remote finished its work */
semGive (pObj->remote.sync);
break;
default:
/* discard */
break;
}
dshmMuxSvcObjRelease (pObj->obj.hw, pObj->obj.svc);
return 0;
}
/****************************************************************************
**
*
* join - initialize service interaction with a remote node
*
* This service expects one remote node to interact with. If no node has
* previously joined the service, this routine will create the
synchronization
* semaphore, record the remote node's address and spawn the worker task that
* waits for synchronization events.
*/
774
17 Distributed Shared Memory: DSHM
17.4 Developing Custom Services
static STATUS join
(
svc_obj * const pObj,
/* the service object */
const uint_t addr
/* address of node joining */
)
{
DSHM(msg);
/* reply message */
int tid;
/* worker task id */
if (pObj->remote.addr != DSHM_ADDR_INVALID)
{
return ERROR;
/* remote node already registered */
}
pObj->remote.sync = semBCreate (SEM_Q_FIFO, SEM_EMPTY);
if (!pObj->remote.sync)
{
printf ("FAILED: out-of-memory\n");
return ERROR;
}
pObj->remote.addr = addr;
printf ("join called\n");
/* reply to remote node, telling it we're here */
DSHM_BUILD(msg, pObj->obj.svc, 0, addr, MSG_JOIN);
dshmMuxMsgSend (pObj->obj.hw, msg, 0, 0);
tid = taskSpawn ("worker", 100, 0, 0x1000, (FUNCPTR)worker,
(int)pObj->remote.sync, (int)pObj->obj.hw,
(int)pObj->obj.svc, (int)pObj->remote.addr,
0,0,0,0,0,0);
17
if (tid == ERROR)
{
logMsg ("Could not spawn worker task\n",
0,0,0,0,0,0);
pObj->remote.addr = DSHM_ADDR_INVALID;
semDelete (pObj->remote.sync);
return ERROR;
}
return OK;
}
/****************************************************************************
**
*
* leave - invoked when a remote node disappears
*
* This routine cleans up the node-specific service data when a remote node
* disappears.
*/
775
VxWorks
Kernel Programmer's Guide, 6.6
static void leave
(
svc_obj * const pObj,
/* the service object */
const uint_t addr
/* address of node leaving */
)
{
if (pObj->remote.addr == addr)
{
semDelete (pObj->remote.sync);
pObj->remote.addr = DSHM_ADDR_INVALID;
}
printf ("leave called\n");
}
/****************************************************************************
**
*
* worker - worker task that waits on sync events from remote node
*
* This routine simulates work that needs synchronization from a remote node.
* It pends on a semaphore that is given when the remote peer sends a sync
* event, signalling the local node that it a condition needed by it has been
* met so that it can resume work.
*/
static void worker
(
SEM_ID sync,
/* synchronization semaphore */
uint16_t hw,
/* hardware bus unique identifier */
uint16_t svc,
/* service unique identifier */
uint16_t remote
/* address of remote node on bus */
)
{
logMsg ("worker task: interacting on svc %d, hw %d with node %d\n",
svc, hw, remote, 0,0,0);
FOREVER
{
DSHM(msg);
/* placeholder: do real work */
taskDelay (SECONDS(2));
/* signal remote task we're done with our part */
DSHM_BUILD(msg, svc, 0, remote, MSG_SYNC);
dshmMuxMsgSend (hw, msg, 0, 0);
/* wait for sync event from remote node */
if (semTake (sync, WAIT_FOREVER) == ERROR)
{
logMsg ("Semaphore deleted, remote task must have quit\n",
0,0,0,0,0,0);
break;
}
776
17 Distributed Shared Memory: DSHM
17.5 Developing a Hardware Interface
logMsg ("Got sync event, resuming...\n",
0,0,0,0,0,0);
}
}
17.5 Developing a Hardware Interface
This section describes how to develop an interface for hardware that is not
currently supported by Wind River. A DSHM hardware interface must include
functionality for a bus controller device driver and a peer node device driver (for
remote node support). DSHM drivers must conform to the VxBus device driver
model (for information about VxBus, see the VxWorks Device Driver Developer’s
Guide).
!
CAUTION: The current implementations of hardware support and services require
that shared memory always be present. Systems in which memory can be swapped
out and cause exceptions when accessed are not currently supported.
A peer (remote) node is seen as a device by DSHM. Peer node device drivers
provide the means for accessing the distributed shared memory on a peer nodes,
and for signalling peer nodes with inter-processor interrupts.
Most of the functionality required by DSHM for the bus controller device driver
and peer node device driver can be provided in driver code, although some
functionality may require support in the BSP (for example, triggering interrupts on
the sb1250).
DSHM supports multiple concurrent buses that provide shared memory
functionality on the same node. As the DSHM MUX takes care of such
combinations, the BSP does not have to.
The interfaces provided by DSHM conform to the VxBus driver model. These
interfaces see the local node as a virtual controller and peer nodes as virtual
devices sitting on the virtual controller. The code can therefore be reused for
different BSPs that support the same devices. Furthermore, common base drivers
exist for similar implementations, such as multicore devices.
Currently, implementations of DSHM hardware interfaces are provided for the
following multicore devices: MIPS sb1250 and sb1480, and PowerPC hpcNet8641.
777
17
VxWorks
Kernel Programmer's Guide, 6.6
Most of the code is shared between them. These BSPs can be used as starting points
to implement support for similar BSPs. They should be a good starting point for
implementing support for different buses as well, not only for the local bus used
by multi-core devices, but particularly for bus controller drivers that are designed
to use shared memory as the messaging interface (including VMEbus).
Note that it is especially important that the local node should also be seen as a
remote node on the bus interface (to be able to participate in broadcasts for
example). If a service needs to treat the local node differently, the service should
be written so as to address this requirement—and not the bus controller driver.
17.5.1 Driver Initialization
Bus controller device drivers differ primarily in their xxxInit2( ) routine. The major
differences due to the location of the anchor for shared memory messaging and the
location of the pool of shared memory. In addition, some parameters (such as
shared memory) can be defined statically for some systems, but not for others; it
depends on the configuration of the physical shared memory.
The peer node device drivers also mostly differ in their the xxxInit2( ) routines,
and for mostly the same reasons as the bus controller device driver. An instance of
a peer driver is a peer device. There is a peer device for each node in the DSHM
system, on a bus. Peer devices must be able to find the remote shared memory data
structures in order to message the remote node that it is responsible for.
Also, peer drivers can differ in the means of interrupting the remote nodes, and so
on. For example, the hpcNet8641 uses the EPIC interrupt controller to send an
inter-processor interrupt to remote nodes, while the SB1250 uses a different
mechanism.
VxBus has three initialization phases. During the first phase, the kernel memory
allocator is not initialized yet, so malloc( ) cannot be used to get memory from the
kernel heap. If a device driver needs to be initialized during the first phase—and
it needs to dynamically allocate memory—VxBus has its own memory allocator
that can be used (which is more limited and manages a very small amount of
memory).
17.5.2 Callbacks
Each bus controller device driver must provide a set of callbacks that are invoked
by the DSHM MUX. The required types of callbacks are as follows:
778
17 Distributed Shared Memory: DSHM
17.5 Developing a Hardware Interface
Allocate
Shared memory allocation callback. A pool of shared memory on the local
target is managed by the bus controller driver. The memory can be reserved
for each service by passing a empty-size pool to the bus controller. If a
non-empty pool is passed, the interface can allocate from it. The dshmMem
library provides allocation and de-allocation routines that can be used by the
bus interface as its allocation and free callbacks. The interface writer is also free
to implement custom callbacks.
Free
Free shared memory callback. The counterpart to the allocation callback.
Transmit
Message transmission callback. This routine sends one message to the
destination provided.
Broadcast
Method for broadcasting a message to all nodes on the bus, including the local
node.
Test-and-Set
Test-and-set primitive provided by the bus controller driver. An atomic
operation that checks if a zero is stored at a memory location and replaces it
with a non-zero value, if so. Returns TRUE if so, FALSE if not.
Clear
The opposite of test-and-set. Can be set to NULL if not needed (such as for
multi-core devices). Some bus controllers need a special implementation
(some VMEbus bus controller chips need it).
Offset-To-Address
Callback that converts a shared memory offset from the start of the address
that the shared memory is visible, to a local pointer to that same location.
Address-To-Offset
Callback that converts a pointer to a local shared memory address into an
offset from the start of the address at which the shared memory is visible. This
allows passing values that can be converted to pointers on the remote nodes,
in case they see the shared memory at a different address.
The offset-to-address and address-to-offset routines should always be used
when passing addresses (of buffers) between nodes, unless the service is only
meant to be used in situations in which the shared memory address is the same
on all nodes.
779
17
VxWorks
Kernel Programmer's Guide, 6.6
Local Address
Obtain the address of the local node on the common bus. This is the
address used for messaging.
Virtual Memory Allocator
Used if the bus controller driver can allocate a virtual address range
dynamically. Unused in Wind River implementations. Can be left NULL.
Fast Copy
Used if the bus controller driver provides something better than bcopy( ) for
writing to shared memory. Can be left NULL.
780
17 Distributed Shared Memory: DSHM
17.5 Developing a Hardware Interface
Prototypes for Callbacks
void*(*alloc)
(
DSHM_PEER_ID,
int * const,
const int
);
STATUS(*free)
(
DSHM_PEER_ID,
void * const
);
STATUS(*tx)
(
DSHM_PEER_ID,
DSHM_TYPE()
);
STATUS(*bcast)
(
DSHM_PEER_ID,
DSHM_TYPE()
);
/* memory allocation */
/* memory deallocation */
/* regular transmit */
/* broadcast on hw bus */
DSHM_TAS tas;
/* test-and-set routine */
DSHM_TAS_CLEAR clear;
/* test-and-set clear routine */
void*(*offToAddr)
/*
(
DSHM_PEER_ID,
const uint32_t
);
uint32_t(*addrToOff)/*
(
DSHM_PEER_ID,
const void * const
);
uint16_t(*localAddr)/*
(
DSHM_PEER_ID
);
VIRT_ADDR(*vmAlloc) /*
(
DSHM_PEER_ID,
const int
);
void(*fastcopy)
/*
(
const char *,
char *,
int
);
sm offset to local address */
local address to sm offset */
17
obtain address of local node */
address range allocation (manager only) */
fast bcopy, dma, page swap, etc */
781
VxWorks
Kernel Programmer's Guide, 6.6
17.5.3 Registering with the DSHM MUX
When the bus controller device is ready to be accessed by services, it must
announce its presence. To do so, it calls dshmMuxHwRegister( ). This allows
services to register for that interface, and messages to be sent on it. At this point,
remote nodes are not visible to the DSHM hardware interface yet.
When a remote node subsequently is discovered, the bus controller device should
call dshmMuxSvcJoin( ) so that services running are notified of the nodes
appearance. Service may require notification so they can allocate resources for the
managing their view of the remote node. These resources may be either local or in
shared memory (for example, buffers for incoming data from the peer node). The
bus controller device should therefore also call dshmMuxSvcLeave( ) when a
remote node disappear, to let the service know of that status change as well.
However, a service can be started at any time, so when a service registers with the
DSHM MUX, it should broadcast its presence to the remote nodes, most likely by
sending a service-specific JOIN message. For more information about writing
services, see 17.4 Developing Custom Services, p.767.
17.5.4 Messaging Support
DSHM attempts to provide immediate feedback about whether a message has
reached the intended peer node recipient. In order to do so, the bus controller
driver must accomplish one of the following with its message transmission facility:
■
Feedback is immediately available from the messaging system if the message
cannot be delivered. This could be the result of a full queue, for example. The
shared memory messaging facility follows this model.
■
The messaging system is loss-less. This meaning that the send operation might
report that the message has been sent even if it has not got through to the
receiver yet, but there is some level of support in the messaging system to keep
retrying until the message goes through.
Note that a message getting through does not imply anything more than the fact
that it has arrived at its destination. It does not mean that there is any certainty of
it being processed by the service on the remote node that received it. Services
should, of course, be designed to handle all incoming messages. Failure to handle
a message should only occur if something is wrong on the remote node (the node
does not respond, it is rebooting, and so on).
DSHM expects messages to be delivered in the same order in which they were sent.
For example, if a JOIN message is sent before a REQUEST message, and the sender
782
17 Distributed Shared Memory: DSHM
17.5 Developing a Hardware Interface
expects the JOIN to be processed first, since if not, the REQUEST message will be
ignored. They must arrive in order.
Sending a message must also double as a memory barrier. If a buffer in shared
memory is filled, and then a message sent to indicate that the buffer is ready to
read, the sending of the message should not occur before the data is actually
written to memory. Services should not have to create the barrier themselves
before signaling remote nodes. DSHM provides the DSHM_MEM_BARRIER( )
macro, which is defined in installDir/vxworks-6.x/target/h/dshm/adapt/types.h.
Shared Memory Messaging Support
The messaging system provided with this release of DSHM is based on shared
memory. It does not require any special hardware support except for providing
shared memory between nodes, and a way of sending interrupts to peer (remote)
nodes. The shared memory itself must be fully coherent between nodes.
Coherency is required both for messaging and for the shared memory pool used
by services. It can be achieved by using a non-cached region of memory, by using
a snooping mechanism on a cached region, and so on.
Note that certain instructions used for inter-process synchronization may require
specific cache modes. For example, on PowerPC architectures, the ll/sc primitives
used to implement atomic operations require the cache to be in a certain mode;
otherwise they cause an exception. Consult your hardware architecture
documentation in this regard.
Each node using shared memory messaging must provide a small data structure
at a well-known address, accessible by all nodes in the system, called the anchor,
which is defined in the BSP. This data structure's size is determined as follow:
(12 + n * 4) bytes
where n is the max number of nodes allowed (for information about configuring
the maximum number of nodes, see 17.3 Configuring VxWorks for DSHM, p.763).
Most of the data in the anchor is for discovery and keep alive signal. It also
provides the location of the rest of the shared memory provided by the node.
As noted earlier, the action of sending a message must be protected by a memory
barrier as well (with the DSHM_MEM_BARRIER( ) macro). With the
shared-memory messaging implementation, this is achieved through a spinlock
mechanism that keeps messaging structures coherent when accessed concurrently
by multiple nodes.
783
17
VxWorks
Kernel Programmer's Guide, 6.6
Since this messaging mechanism relies on a portion of shared memory reserved in
advance (of a fixed size), the number of messages, that can be sent by the
messaging node and that have not yet been processed by the receiving node, is
limited. If the sender fills the message input on the remote node, it will try
re-sending a limited number of times (configurable) before giving up and
returning an error to the caller. Note that the sender is busy-waiting (spinning in a
loop) while repeatedly trying to send.
To avoid busy-waiting while sending a message, special consideration should be
given to the implementation services.
Services that are used over a bus controller device driver that uses shared memory
messaging should try to throttle their messages to prevent sending more than are
allowed. This should be combined with a configurable setting for the number of
concurrent messages to achieve the desired effect.
For example, consider a system with three nodes. Each node provides two buffers
for each of its peer to write to. Each of its peer does the same. So, the local node has
two buffers on each peer to write to, and two buffers of its own, per-peer, from
which it reads.
The service keeps a local view of the status of the remote buffers. In this service, a
message is sent every time a buffer is filled to let the remote node that it should
read it. A message is also sent when a local buffer is read as well, to signal the
remote node that it can use it again. In this case, the concurrent messages that can
have been sent by one node to the other are as follows:
■
Two messages that tell the remote node to read the buffers (the service should
stop sending when the buffers are full, so it won't send a next message until it
got a reply from the remote node telling it can reuse the first buffer)
■
Two messages that tell the remote node that it can reuse its two buffers (the
local node only sends these after the remote node has sent it messages telling
it to read the buffers, and we know it is limited to two of them at once).
So if this service is the only one running in this system, four message entries should
be sufficient.
The number of required messages entries must, however, be set to one more than
the number calculated. This is due to the implementation of the messaging system,
which uses a ring and must have one empty entry. In addition, for performance
considerations, the number of entries in the ring must be a power-of-two. If a
number that is not a power-of-two parameter is specified, it is automatically
rounded down to the nearest power-of-two.
784
17 Distributed Shared Memory: DSHM
17.5 Developing a Hardware Interface
Shared Memory Lock-Less Discovery
Bus controller device drivers using shared memory messaging must use a method
other than sending a join (JOIN) message to remote peers when a node comes up.
The reason behind this has to do with rebooting nodes and re-initialization of
shared memory data structures. This method is comprised of two types of
messages, reset (RST) and acknowledge (ACK). The following diagram illustrates
the state machine of the discovery mechanism, for each peer relative to the remote
node.
The first letter of the state is the state of the current node and the second is the state
of the remote peer, as viewed by the local node. The R is reset, U is unknown, and A
is received acknowledgment. The state transitions show the event received/action
taken. The AA designation is the fully-up connection state.
17.5.5 Management Service
There is a generic management service that exists in the VxWorks adaptation layer
component. It can be registered with a hardware interface by calling
dshmMngSvcInstall( ) (see dshmBusCtlrSim.c under target/src/dshm/drivers for
an example of usage). To be of any use, the driver must provide a
dshmBusCtlrMethodNodeReadyId method. Look in dshmBusCtlrPlb.c for an
example (that file is the base driver on top of which dshmBusCtlrSim.c is built).
dshmBusCtlrPlbNodeReady is registered as a method in the method table.
This service handles two types of messages currently: DSHM_TYP_MNG_JOIN and
DSHM_TYP_MNG_QUIT, although the second is a placeholder in the VxWorks
adaptation layer that simply acknowledge the message and logs a console
message. The implementation is in target/src/dshm/service/dshmSvcMng.c, and
can be used as a starting point for a more involved implementation if needed. The
service is very minimal at this point since all the current hardware interface use
another mechanism for detecting topology changes.
One of the main goals of this service is to propagate the information that a node
has either appeared or disappeared to the services, which are the modules that
provide the bulk of the functionality that is of interest to end users. Services that
register a join and a leave callback receive events those events when the hardware
interface sees a change in the nodes topology.
785
17
VxWorks
Kernel Programmer's Guide, 6.6
17.5.6 DSHM Hardware Interface APIs
The routines described in Table 17-1 are for use by a hardware interface.
Table 17-2
DSHM Hardware Interface Routines
Routine
Description
dshmMuxHwRegister( )
Register a hardware bus with the MUX.
dshmMuxSvcNodeJoin( )
Call join callbacks from registered services.
dshmMuxSvcNodeLeave( )
Call leave callbacks from registered services.
786
18
Message Channels
18.1 Introduction 787
18.2 Message Channel Facilities 789
18.3 Multi-Node Communication with TIPC 791
18.4 Single-Node Communication with COMP and DSI 791
18.5 Socket Name Service 795
18.6 Socket Application Libraries 799
18.7 Configuring VxWorks for Message Channels 803
18.8 Comparison of Message Channels and Message Queues 806
18.1 Introduction
Message channels are a socket-based facility that provides for inter-task
communication within a memory boundary, between memory boundaries (kernel
and processes), between nodes (processors) in a multi-node cluster, and between
between multiple clusters. In addition to providing a superior alternative to TCP
for multi-node intercommunication, message channels provide a useful
alternative to message queues for exchange data between two tasks on a single
node.
787
VxWorks
Kernel Programmer's Guide, 6.6
Message channels provide a connection-oriented messaging mechanism. Tasks
exchange information in the form of messages that can be of variable size and
format. They can be passed back and forth in full duplex mode once the connection
is established.
The implementation of message channels for multi-node communication is based
on TIPC, which provides faster throughput than TCP. The implementation of
message channels for single-node, inter-process communication is somewhat
slower than message queues, but provides notable advantages (including
portability of applications for multi-node message channel use; for more
information see 18.8 Comparison of Message Channels and Message Queues, p.806).
Message channel communication can take place between tasks running in the
kernel and tasks running in processes (RTPs) on a single node, as well as between
multiple nodes, regardless of the memory context in which the tasks are running.
For example, message channels can be used to communicate between:
■
a task in the kernel of one node and a task in a process on another node
■
a task in a process on one node and a task in a process on another node
■
a task in the kernel and a task in a process on a single node
■
a task in one process and a task in another process on a single node
and so on.
The scope of message channel communication can be configured to limit server
access to:
■
one memory space on a node (either the kernel or one process)
■
all memory spaces on a node (the kernel and all processes)
■
a cluster of nodes in a system (including all memory spaces in each node)
Message channels can also be used for multi-node communication between
multiple clusters.
788
18 Message Channels
18.2 Message Channel Facilities
18.2 Message Channel Facilities
The message channel technology consists of the following basic facilities:
■
Transparent Inter-Process Communication (TIPC) for multi-node
communication. See 18.3 Multi-Node Communication with TIPC, p.791.
■
Connection-Oriented Message Passing (COMP) protocol with the DSI socket
back end for single node communication. See 18.4 Single-Node Communication
with COMP and DSI, p.791. Note that COMP is also provides services for
multi-node communication.
■
Socket Name Service (SNS), which provides location and interface
transparency for message channel communication between tasks on a single
node, and maintains communication between nodes for multi-node message
channel communications. In addition, it controls the scope of message channel
communication (to two memory spaces, a node, a cluster of nodes, or a zone).
See 18.5 Socket Name Service, p.795.
■
Socket Application Libraries (SAL), which provide APIs for using message
channels in both client and server applications, as well as the mechanism for
registering the tasks that are using a message channel with a Socket Name
Service. See 18.6 Socket Application Libraries, p.799.
Client and server applications that make use of SAL can be developed as either
kernel applications or RTP applications (or both).
Figure 18-1 illustrates the architecture of the message channel facilities.
18
789
VxWorks
Kernel Programmer's Guide, 6.6
Figure 18-1
Message Channel Architecture
applications
Socket ApplicationLibraries
Socket Name Service
TIPC
Connection-Oriented
Message Passing
Protocol
Distributed Systems
Infrastructure
basic configuration
add for multi-node support
The basic configuration of VxWorks with support for message channels includes
SAL, SNS, COMP, and DSI—which provides support for single node use. TIPC
must be added to the basic configuration for multi-node use. TIPC can also be
added to provide connection-less socket types on a single node (for use outside of
message channels).
For detailed information about configuration, see 18.7 Configuring VxWorks for
Message Channels, p.803. For information about TIPC, see the Wind River TIPC
Programmer’s Guide.
790
18 Message Channels
18.3 Multi-Node Communication with TIPC
18.3 Multi-Node Communication with TIPC
The underlying transport mechanism for multi-node message channels is based on
the Transparent Inter-Process Communication (TIPC) protocol, which provides a
fast method for transferring messages across node boundaries in a cluster
environment, or across cluster boundaries in a zone environment. TIPC can also be
used within a single node.
TIPC is designed for use with the standard socket API. For connection-oriented
messaging, the socket type associated with message channels is the
SOCK_SEQPACKET. TIPC also supports SOCK_RDM, SOCK_DGRAM, and
SOCK_STREAM, but those socket types are not supported by the message channel
facility as only SOCK_SEQPACKET is supported by COMP. This simplifies
migration between single-node and multi-node applications of message channels.
The TIPC protocol is connection-based, like other stream-based protocols such as
TCP, but it carries variable-sized messages, like datagram-based protocols such as
UDP. In providing cluster and node based communications, TIPC sockets are
available in the AF_TIPC domain. TIPC provides several means of identifying end
points that are handled transparently through the SNS name server. For more
information about TIPC, see the Wind River TIPC for VxWorks 6 Programmer's
Guide.
18.4 Single-Node Communication with COMP and DSI
The Connection-Oriented Message Passing protocol (COMP) provides services for
multi-node as well as the protocol for single-node communication. The underlying
transport mechanism for single-node message channels is based on the COMP
protocol, which provides a fast method for transferring messages across memory
boundaries on a single node.
COMP, using the AF_LOCAL family, is designed for use with the standard socket
API. Because it provides connection-oriented messaging, the socket type
associated with message channels is the SOCK_SEQPACKET. The protocol is
connection-based, like other stream-based protocols such as TCP, but it carries
variable-sized messages, like datagram-based protocols such as UDP.
791
18
VxWorks
Kernel Programmer's Guide, 6.6
While COMP provides for standard socket support, it has no dependency on
TCP/IP networking facilities, which can be left out of a system if the facilities are
not otherwise needed.
In providing single-node local communications, COMP sockets are available as
part of the AF_LOCAL domain. Although this domain is traditionally related to the
UNIX file system, in VxWorks the addressing is completely independent of any
file system. Like UNIX sockets, COMP uses a string to define the address, and it
has a structure similar to a file path name, but this is the extent of the similarity in
this regard. The address is simply a logical representation of the end-point.
The transfer of data in message channels is based on an internal buffer
management implementation that allows for deterministic memory allocation,
which reduces the amount of copies needed to transfer the data whenever
possible. Only one copy is needed for the internal transfer; the data coming from
the user is directly moved into the receiver buffer space. Another copy is required
to submit and retrieve the data to and from the channel.
COMP supports the standard socket options, such as SO_SNDBUF or
SO_RECVBUF and SO_SNDTIMEO and SO_RCVTIME. For information about the
socket options, see the socket API references. For information about how COMP
uses them, see installDir/vxworks-6.x/target/src/dsi/backend/dsiSockLib.c.
Express Messaging
Express messaging is also available for sending and receiving a message. An
express message is placed on a special queue on the sending side and placed at the
front of the normal queue at the receiving end. This allows for urgent messages to
be sent and received with a higher priority than the normal messages. In order to
send an express message, the flags parameter of the standard send( ) routine must
have the MSG_EXP bit set. (Also see the socket send( ) API reference).
Show Routines
Because COMP is based on the standard socket API, traditional network show
routines can be used, such as netstat( ). In addition, information on local sockets
can be retrieved with the unstatShow( ) routine (for more information, see the
VxWorks API reference entry).
792
18 Message Channels
18.4 Single-Node Communication with COMP and DSI
18.4.1 COMP Socket Support with DSI
The COMP socket functional interface is provided by the Distributed Systems
Infrastructure (DSI) back end.
The DSI back end is a simplified version of the BSD socket back-end. It is designed
for optimized communications when both end points are in a single node. The DSI
back end provides the set of implementations of the standard socket functions for
the COMP protocol specific calls. The traditional network protocols in VxWorks,
such as TCP and UDP, use the BSD Internet Domain Socket back end and are
described in the Wind River Network Stack for VxWorks 6 Programmer’s Guide.
The DSI back end requires its own system and data memory pools, which are used
to handle the creation of sockets and the data transfers between two endpoints.
The pools are similar to those required for the network stack. In addition, the pools
are configured so as to enhance performance for the local transfers. The system
pool provides COMP with the memory it needs for its internal structures and data
types. The data pool provides COMP with the memory it needs for receiving data.
Because COMP is local, data transfer has been optimized so that data are put
directly in the receiver’s packet queue.
Both the DSI back end and DSI memory pools complement the BSD equivalent.
Therefore, both BSD and DSI sockets can coexist in the system. They do not depend
on each other, so that they can be added or removed, as needed.
COMP uses netBufLib to manage its internal system and data memory pools. For
detailed information on how buffers are configured, see the coverage of the similar
technology, netBufPool, in the Wind River Network Stack for VxWorks 6
Programmer’s Guide.
These pools are created automatically by the INCLUDE_DSI_POOL component.
The DSI parameters listed in Table 18-1 are used for memory pool configuration.
These parameters are used when usrNetDsiPoolConfig( ) routine is called, which
happens automatically when the system boots. The dsiSysPoolShow( ) and
dsiDataPoolShow( ) can be used to display related information (see the VxWorks
API reference for dsiSockLib).
793
18
VxWorks
Kernel Programmer's Guide, 6.6
Table 18-1
INCLUDE_DSI_POOL Component Parameters
Parameter
Default Value
DSI_NUM_SOCKETS
200
DSI_DATA_32
50
DSI_DATA_64
100
DSI_DATA_128
200
DSI_DATA_256
40
DSI_DATA_512
40
DSI_DATA_1K
10
DSI_DATA_2K
10
DSI_DATA_4K
10
DSI_DATA_8K
10
DSI_DATA_16K
4
DSI_DATA_32K
0
DSI_DATA_64K
0
The DSI pool is configured more strictly and more efficiently than the core network
pool since it is more contained, fewer scenarios are possible, and everything is
known in advance (as there is only the one node involved). The
DSI_NUM_SOCKETS parameter controls the size of the system pool. It controls the
number of clusters needed to fit a socket, for each family and each protocol
supported by the back end. Currently, only the AF_LOCAL address family is
supported by COMP.
The clusters allocated in the back end are of these sizes:
■
■
■
aligned sizeof (struct socket)
aligned sizeof (struct uncompcb)
aligned sizeof (struct sockaddr_un)
One cluster of size 328 and of size 36 are needed for each socket that is created since
currently, the COMP protocol is always linked to a DSI socket. Only one cluster of
sizeof (struct sockaddr_un) is required, therefore the size of the system pool is
794
18 Message Channels
18.5 Socket Name Service
basically determined by: (DSI_NUM_SOCKETS * (328 + 36) + 108)). Using these
sizes prevents any loss of space since they are the actual sizes needed.
All other parameters for the DSI pool are used to calculate the size of clusters in the
data pool, and at the same time, the size of the pool itself. The data pool is used as
packet holders during the transmissions between two sockets, between the time
the data is copied from the sender’s buffer to the receiver’s buffer. Each of them
represent a cluster size from 32 bytes to 64 kilobytes and the number of allocated
clusters of that specific size.
To set reasonable values for the parameters in this component, you must know
how much memory your deployed application will require. There is no simple
formula that you can use to anticipate memory usage. Your only real option is to
determine memory usage empirically. This means running your application under
control of the debugger, pausing the application at critical points in its execution,
and monitoring the state of the memory pool. You will need to perform these tests
under both stressed and unstressed conditions.
18.5 Socket Name Service
A Socket Name Service (SNS) allows a server application to associate a service
name with a collection of listening sockets, as well as to limit the visibility of the
service name to a restricted (but not arbitrary) set of clients.
Both Socket Application Library (SAL) client and server routines make use of an
SNS server to establish a connection to a specified service without the client having
to be aware of the address of the server's listening sockets, or the exact interface
type being utilized (see 18.6 Socket Application Libraries, p.799). This provides both
location transparency and interface transparency. Such transparency makes it
possible to design client and server applications that can operate efficiently
without requiring any knowledge of the system's topology.
An SNS server is a simple database that provides an easy mapping of service
names and their associated sockets. The service name has this URL format:
[SNS:]service_name[@scope]
The [SNS:] prefix is the only prefix accepted, and it can be omitted. The scope can
have the following values: private, node, cluster, or system. These values
designate an access scope for limiting access to the same single memory space (the
795
18
VxWorks
Kernel Programmer's Guide, 6.6
kernel or a process), the same node (the kernel and all processes on that node), a
set of nodes, or the entire system, respectively. A server can be accessed by clients
within the scope that is defined when the server is created with the salCreate( )
routine (see 18.6.1 SAL Server Library, p.800).
NOTE: The SNS server creates a COMP socket for its own use for local
communication. It has the socket address of 0x0405. All of the SAL routines send
messages to the SNS server at this socket address.
SNS provides a resource reclamation mechanism for servers created within
processes. If a process dies before salDelete( ) has been called on a SAL server, SNS
will be notified and will remove the entry from the database. Note, however, that
this mechanism is not available for tasks in the kernel. If a task in the kernel
terminates before salDelete( ) is called, the service name is not automatically
removed from SNS. In order to avoid stale entries that may prevent new services
with the same name from being created, the salRemove( ) routine should be used.
The SNS server can be configured to run in either kernel or user space (as a kernel
application or an RTP application, respectively). For more information, see
Running SNS as a Process, p.804).
A node should not be configured with more than one SNS server. The server starts
at boot time, and is named tSnsServer if it is running in the kernel, or iSnsServer
if it is running as a process. For a multi-node system, a monitoring task is
automatically spawned to maintain a list of all the SNS servers in the zone
(system). The monitoring task is named tDsalMonitor, and it runs in the kernel.
The snsShow( ) command allows a user to verify that SAL-based services are
correctly registered with the SNS server from the shell (see 18.5.2 snsShow( )
Example, p.797).
For more information, see theVxWorks API reference for snsLib.
18.5.1 Multi-Node Socket Name Service
For a multi-node system, each node in the system must be configured with the
Socket Name Service (SNS). Note that VxWorks SNS components for multi-node
use are different from those used on single node systems (see 18.7 Configuring
VxWorks for Message Channels, p.803).
When a distributed SNS server starts on a node at boot time, it uses a TIPC bind
operation to publish a TIPC port name. This is visible to all other nodes in the zone.
The other existing SNS servers then register the node in their tables of SNS servers.
796
18 Message Channels
18.5 Socket Name Service
A separate monitoring task (called tDsalMonitor) is started on each node at boot
time, which uses the TIPC subscription feature to detect topology-change events
such as a new SNS server coming online, or an existing SNS server leaving the zone
(system).
Note that if the TIPC networking layer does not start up properly at boot time, the
distributed SAL system will not initialize itself correctly with TIPC, and the SNS
server will work strictly in local mode. The SNS server does not check for a
working TIPC layer after the system boots, so that it will not detect the layer if it is
subsequently started manually, and the SNS server will continue to run in local
mode.
When a new node appears, each SNS server sends a command to that node
requesting a full listing of all sockets that are remotely accessible. The SNS server
on the new node sends a list of sockets that can be reached remotely.
Each time a new socket is created with salCreate( ) on a node that has a server
scope greater than node, this information is sent to all known SNS servers in the
cluster. All SNS servers are thereby kept up to date with relevant information.
Similarly, when a socket is deleted using the salRemove( ) function, this
information is sent to all known SNS servers in the cluster. The addition and
removal of sockets is an infrequent occurrence in most anticipated uses and should
be of minimal impact on network traffic and on the performance of the node.
When the tDsalMonitor task detects that an SNS server has been withdrawn from
the system, the local SNS server purges all entries related to the node that is no
longer a part of the distributed SNS system.
Note that only information on accessible sockets is transmitted to remote SNS
servers. While it is acceptable to create an AF_LOCAL socket with cluster scope,
this socket will use the COMP protocol which can only be accessed locally. SNS
servers on remote nodes will not be informed of the existence of this socket.
On a local node, if a socket name exists in the SNS database in both the AF_LOCAL
and AF_TIPC families, when a connection is made to that name using salOpen( ),
the AF_LOCAL socket will be used.
18.5.2 snsShow( ) Example
The snsShow( ) shell command provides information about all sockets that are
accessible from the local node, whether the sockets are local or remote. The
command is provided by the VxWorks INCLUDE_SNS_SHOW component.
797
18
VxWorks
Kernel Programmer's Guide, 6.6
The following examples illustrate snsShow( ) output from three different nodes in
a system.
From Node <1.1.22>
NAME
SCOPE FAMILY TYPE
-------------------------- ----- ------ ------astronaut_display
clust LOCAL SEQPKT
TIPC
SEQPKT
ground_control_timestamp
clust TIPC
SEQPKT
ground_control_weblog
systm TIPC
SEQPKT
heartbeat_private
priv LOCAL SEQPKT
TIPC
SEQPKT
local_temperature
node LOCAL SEQPKT
newsfeed
clust TIPC
SEQPKT
rocket_diagnostic_port
clust TIPC
SEQPKT
rocket_propellant_fuel_level_interface
---clust TIPC
SEQPKT
spacestation_docking_port clust TIPC
SEQPKT
PROTO
----0
0
0
*
0
*
0
0
0
0
*
0
0
0
ADDR
------------------/comp/socket/0x5
<1.1.22>,1086717967
<1.1.25>,1086717965
<1.1.25>,1086717961
/comp/socket/0x4
<1.1.22>,1086717966
/comp/socket/0x2
<1.1.50>,1086717962
<1.1.22>,1086717964
<1.1.22>,1086717960
* <1.1.55>,1086717963
From Node <1.1.25>
NAME
-------------------------astronaut_display
ground_control_timestamp
SCOPE
----clust
clust
FAMILY
-----TIPC
LOCAL
TIPC
ground_control_weblog
systm TIPC
local_billboard
node LOCAL
TIPC
newsfeed
clust TIPC
rocket_diagnostic_port
clust TIPC
rocket_propellant_fuel_level_interface
---clust TIPC
spacestation_docking_port clust TIPC
TYPE
------SEQPKT
SEQPKT
SEQPKT
SEQPKT
SEQPKT
SEQPKT
SEQPKT
SEQPKT
PROTO
----0
*
0
0
0
0
0
0
*
0
*
SEQPKT
SEQPKT
0
0
TYPE
------SEQPKT
SEQPKT
SEQPKT
SEQPKT
SEQPKT
PROTO
----0
*
0
*
0
*
0
*
0
*
SEQPKT
SEQPKT
SEQPKT
0
0
0
ADDR
------------------<1.1.22>,1086717967
/comp/socket/0x3
<1.1.25>,1086717965
<1.1.25>,1086717961
/comp/socket/0x2
<1.1.25>,1086717964
<1.1.50>,1086717962
<1.1.22>,1086717964
* <1.1.22>,1086717960
* <1.1.55>,1086717963
From Node <1.1.55>
NAME
SCOPE FAMILY
-------------------------- ----- -----astronaut_display
clust TIPC
ground_control_timestamp
clust TIPC
ground_control_weblog
systm TIPC
newsfeed
clust TIPC
rocket_diagnostic_port
clust TIPC
rocket_propellant_fuel_level_interface
---clust TIPC
spacestation_docking_port clust LOCAL
TIPC
ADDR
------------------<1.1.22>,1086717967
<1.1.25>,1086717965
<1.1.25>,1086717961
<1.1.50>,1086717962
<1.1.22>,1086717964
* <1.1.22>,1086717960
/comp/socket/0x2
<1.1.55>,1086717963
The output of the snsShow( ) command is fairly self-explanatory. The first field is
the name of the socket. If the name is longer than the space allocated in the output,
798
18 Message Channels
18.6 Socket Application Libraries
the entire name is printed and the other information is presented on the next line
with the name field containing several dashes.
The scope values are priv for private, node for node, clust for cluster, and systm
for system.
The family types can be TIPC for AF_TIPC or LOCAL for AF_LOCAL.
The socket type can be SEQPKT for SOCK_SEQPACKET, RDM.
The protocol field displays a numeric value and a location indicator. The numeric
value is reserved for future use, and currently only zero is displayed. The final
character in the field indicates whether the socket was created on a remote or local
node, with an asterisk (*) designating remote.
The address field indicates the address of the socket. All addresses of the form
/comp/socket belong to the AF_LOCAL family. All addresses of the form
,refID belong to the AF_TIPC family. The TIPC address gives the TIPC
portID which consists of the nodeID and the unique reference number.
18.6 Socket Application Libraries
The Socket Application Libraries (SAL) simplify creation of both server and client
applications by providing routines to facilitate use of the sockets API.
SAL also provides an infrastructure for the development of location-transparent
and interface-transparent applications. By allowing SAL to handle the basic
housekeeping associated with a socket-based application, developers can focus on
the application-specific portions of their designs. Developers are free to use the
complete range of SAL capabilities in their applications, or just the subset that suits
their needs; they can even bypass SAL entirely and develop a socket-based
application using nothing but custom software. The SAL client and server APIs
can be used in both kernel and user space.
Several VxWorks components are available to provide SAL support in different
memory spaces, for single or multi-node systems, and so on (see 18.7 Configuring
VxWorks for Message Channels, p.803).
SAL-based applications can also utilize the Socket Name Service (SNS), which
allows a client application to establish communication with a server application
799
18
VxWorks
Kernel Programmer's Guide, 6.6
without having to know the socket addresses used by the server (see 18.5 Socket
Name Service, p.795).
!
CAUTION: SAL applications should not use any of the following as part of a name:
* ? @ : # / < > % | [ ] { } , \\ \ ' & ; = + $
In addition, these should not be used when attempting to find a service:
@ : # / < > % | [ ] { } , \\ \ ' & ; = + $
18.6.1 SAL Server Library
The SAL server routines provide the infrastructure for implementing a
socket-based server application. The SAL server allows a server application to
provide service to any number of client applications. A server application
normally utilizes a single SAL server in its main task, but is free to spawn
additional tasks to handle the processing for individual clients if parallel
processing of client requests is required. The SAL server library is made of the
following routines:
salCreate( )
Creates a named socket-based server.
salDelete( )
Deletes a named socket-based server.
salServerRtnSet( )
Configures the processing routine with the SAL server.
salRun( )
Activates a socket-based server.
salRemove( )
Removes a service from the SNS by name.
A server application typically calls salCreate( ) to configure a SAL server with one
or more sockets that are then automatically registered with SNS under a specified
service identifier. The number of sockets created depends on which address
families, socket types, and socket protocols are specified by the server application.
AF_LOCAL and AF_TIPC sockets are supported.
If the address family specified is AF_UNSPEC, the system attempts to create sockets
in all of the supported address families (AF_LOCAL and AF_TIPC). The socket
addresses used for the server's sockets are selected automatically, and cannot be
specified by the server application with salCreate( ).
800
18 Message Channels
18.6 Socket Application Libraries
A server can be accessed by clients within the scope that is defined when the server
is created with the salCreate( ) routine.
The scope is identified as part of the first parameter, with one the following values:
private, node, cluster, or system. These values designate an access scope for
limiting access to the same task (kernel or process), the same node (the kernel and
all processes on that node), and a set of nodes, respectively. The second parameter
identifies the protocol (with 1 being COMP and 33 being TIPC; 0 is used for all
supported families). The third parameter identifies the socket type.
For example, the following call would create a socket named foo with cluster
scope, with the COMP protocol:
salCreate("foo@cluster",1,5)
!
CAUTION: A COMP (single node) socket can be created with cluster or system
scope, but this setting has no effect in a multi-node system. That is, in a multi-node
system, SNS will not transmit this information to other nodes because a COMP
socket is only available on the node on which it was created.
NOTE: It is possible to create both a COMP socket and a TIPC socket with the same
name. Only the TIPC socket information is sent to other nodes in a multi-node
system (assuming the scope is set appropriately).
Once created, a SAL server must be configured with one or more processing
routines before it is activated. These routines can be configured by calling
salServerRtnSet( ).
Once the server is ready, salRun( ) is called to start the server activities. The
salRun( ) routine never returns unless there is an error or one of the server
processing routines requests it. You must call salDelete( ) to delete the server and
its sockets regardless of whether or not the routine has terminated. This is
accomplished with salDelete( ). This routine can be called only by tasks in the
process (or the kernel) where the server was created. In order for tasks outside the
process to remove a service name from SNS, salRemove( ) must be used. The
salRemove( ) routine does not close sockets, nor does it delete the server. It only
deletes the SNS entry, and therefore access to any potential clients.
For more information, including sample service code, see the VxWorks API
reference for the salServer library.
801
18
VxWorks
Kernel Programmer's Guide, 6.6
18.6.2 SAL Client Library
The SAL client library provides a simple means for implementing a socket-based
client application. The data structures and routines provided by SAL allow the
application to easily communicate with socket-based server applications that are
registered with the Socket Name Service (see 18.5 Socket Name Service, p.795).
Additional routines can be used to communicate with server applications that are
not registered with the SNS. The SAL client library is made of the following
routines:
salOpen( )
Establishes communication with a named socket-based server.
salSocketFind( )
Finds sockets for a named socket-based server.
salNameFind( )
Finds services with the specified name.
salCall( )
Invokes a socket-based server.
A client application typically calls salOpen( ) to create a client socket and connect
it to the named server application. The client application can then communicate
with the server by passing the socket descriptor to standard socket API routines,
such as send( ) and recv( ).
As an alternative, the client application can perform a send( ) and recv( ) as a single
operation using salCall( ). When the client application no longer needs to
communicate with a server it calls the standard socket close( ) routine to close the
socket to the server.
A client socket can be shared between two or more tasks. In this case, however,
special care must be taken to ensure that a reply returned by the server application
is handled by the correct task.
The salNameFind( ) and salSocketFind( ) routines facilitate the search of the
server and provide more flexibility for the client application.
The salNameFind( ) routine provides a lookup mechanism for services based on
pattern matching, which can be used with (multiple) wild cards to locate similar
names. For example, if the names are foo, foo2, and foobar, then a search using
foo* would return them all. The scope of the search can also be specified. For
example, a client might want to find any server up to a given scope, or only within
a given scope. In the former case the upto_ prefix can be added to the scope
802
18 Message Channels
18.7 Configuring VxWorks for Message Channels
specification. For example, upto_node defines a search that look for services in all
processes and in the kernel in a node.
Once a service is found, the salSocketFind( ) routine can be used to return the
proper socket ID. This can be useful if the service has multiple sockets, and the
client requires use of a specific one. This routine can also be used with wild cards,
in which case the first matching server socket is returned.
For more information, including sample client code, see the VxWorks API
reference for the salClient library.
18.7 Configuring VxWorks for Message Channels
To provide the basic set of message channel facilities in a system, VxWorks must
be configured with COMP, DSI, SAL, and SNS components. Selection of SNS
components determines whether the system supports single-node or multi-node
communication.
COMP, DSI, and SAL Components
The required COMP, DSI, and SAL components are as follows:
■
■
■
■
■
INCLUDE_UN_COMP
INCLUDE_DSI_POOL
INCLUDE_DSI_SOCKET
INCLUDE_SAL_SERVER
INCLUDE_SAL_CLIENT
18
Note that INCLUDE_UN_COMP is required for both single and multi-node
systems, as it provides support for communication between SAL and SNS. While
COMP provides for standard socket support, it has no dependency on TCP/IP
networking facilities, which can be left out of a system if they are not otherwise
needed.
SNS Component Options
In addition to the COMP, DSI, and SAL components, one of the four following
components listed below is required for SNS support.
803
VxWorks
Kernel Programmer's Guide, 6.6
Single Node Options
■
INCLUDE_SNS to run SNS as a kernel daemon, for single node
communication.
■
INCLUDE_SNS_RTP to start SNS as a process automatically at boot time, for
single node communication.
Multi-Node Options
■
INCLUDE_SNS_MP to run SNS as a kernel daemon, supporting distributed
named sockets for multi-node communication.
■
INCLUDE_SNS_MP_RTP to start SNS as a process (RTP) automatically at boot
time, supporting distributed named sockets for multi-node communication.
Additional system configuration is required to run SNS as a process; for
information in this regard, see Running SNS as a Process, p.804.
Note that including a distributed SNS server automatically includes TIPC.
Running SNS as a Process
In order to run SNS as a process (RTP), the developer must also build the server,
add it to ROMFS, configure VxWorks with ROMFS support, and then rebuild the
entire system:
a.
Build installDir/vxworks-6.x/target/usr/apps/dsi/snsd/snsd.c (using the
makefile in the same directory) to create snsServer.vxe.
b.
Copy snsServer.vxe to the ROMFS directory (creating the directory first,
if necessary.
The INCLUDE_SNS_RTP and INCLUDE_SNS_MP_RTP components must
know the location of the server in order to start it at boot time. They expect
to find the server in the ROMFS directory. If you wish to store the server
somewhere else (in another file system to reduce the VxWorks image size,
for example) use the SNS_PATHNAME parameter to identify the location.
c.
Configure VxWorks with the ROMFS component.
d. Rebuild VxWorks.
These steps can also be performed with Wind River Workbench (see the Wind River
Workbench User’s Guide). For information about ROMFS, see 8.8 Read-Only Memory
File System: ROMFS, p.516.
804
18 Message Channels
18.7 Configuring VxWorks for Message Channels
SNS Configuration Parameters
The following SNS component parameters can usually be used without
modification:
SNS_LISTEN_BACKLOG
This parameter defines the number of outstanding service requests that the
SNS server can track on the socket that it uses to service SNS requests from
SAL routines. The default value is 5. The value may be increased if some SAL
requests are not processed on a busy system.
SNS_DISTRIBUTED_SERVER_TYPE and SNS_DISTRIBUTED_SERVER_INSTANCE
These parameters are used in the multi-node configuration of SNS servers to
define the TIPC port name that all SNS servers use. The default is type 51 and
instance 51 in the TIPC name tables. If this type and instance conflict with
other usages in the network, they can be changed to values that are unique for
the network. Note that it is recommended to use a type of 50 or above (types 0
through 7 are reserved by TIPC).
!
CAUTION: It is recommended that you do not change the default values of the
SNS_PRIORITY and SNS_STACK_SIZE parameters. The default for SNS_PRIORITY
is 50 and the default for SNS_STACK_SIZE is 20000.
Show Routines
The show routines related to COMP can be included by adding the
INCLUDE_UN_COMP_SHOW component. The snsShow( )routine is included with
the INCLUDE_SNS_SHOW component. In order to use netstat( ) the network show
routines must be included. Note that this will force the inclusion of networking
components.
For information about processes and applications, see VxWorks Application
Programmer’s Guide: Applications and Processes.
805
18
VxWorks
Kernel Programmer's Guide, 6.6
18.8 Comparison of Message Channels and Message Queues
In addition to providing a superior alternative to TCP for multi-node
intercommunication, message channels provide an alternative to message queues
for exchanging data between two tasks on a single node. Both message channels
and message queues allow multiple tasks to send and receive from the same
channel. The main differences between these facilities are:
■
Message channels can be used to communicate between nodes in a multi-node
cluster, but message queues cannot.
■
Implementation of message channels in a single-node system facilitates
porting an application to a multi-node system.
■
Message channels are slower than message queues because of the overhead of
the socket back end. While message channels are notably slower in a single
memory space (such as within the kernel), they are only slightly slower when
communicating across a memory barrier (such as between processes). Note
that message channels are much faster than TCP for inter-node
communication. Message channels are connection-oriented while message
queues are not. There is no way to establish a connection between two tasks
with message queues. In a connection-oriented communication, the two
end-points are aware of each other, and if one leaves the other eventually finds
out. By way of analogy, a connection-oriented communication is like a
telephone call, whereas a connection-less communication is like sending a
letter. Both models are valid, and the requirements of the application should
determine their use.
Each message queue is unidirectional. In order to establish a bidirectional
communication, two queues are needed, one for each end-point (see
Figure 4-13). Each message channel is bidirectional and data can be sent from
both end-points at any time. That is, each message channel provides
connection-oriented full-duplex communication.
■
The messages communicated by message channels can be of variable size,
whereas those communicated by message queues have a maximum size that
is defined when the queue is created. Message channels therefore allow for a
better utilization of system resources by using exactly what is needed for the
message, and nothing more.
■
Message queues have a fixed capacity. Only a pre-defined number of
messages can be in a queue at any one time. Message channels, on the other
hand, have a flexible capacity. There is no limit to the number of messages that
a message channel can handle.
806
18 Message Channels
18.8 Comparison of Message Channels and Message Queues
■
Message channels provide location transparency. An endpoint can be referred
to by a name, that is by a simple string of characters (but a specific address can
also be used). Message queues only provide location transparency for
inter-process communication when they are created as public objects.
■
Message channels provide a simple interface for implementing a client/server
paradigm. A location transparent connection can be established by using two
simple calls, one for the client and one for the server. Message queues do not
provide support for client/server applications.
■
Message channels use the standard socket interface and support the select( )
routine; message queues do not.
■
Message channels cannot be used with VxWorks events; message queues can.
■
Message queues can be used within an ISR, albeit only the msgQsend( )
routine. No message channel routines can be used within an ISR.
■
Message queues are based entirely on a proprietary API and are therefore
more difficult to port to a different operating systems than message channels,
which are based primarily on the standard socket API.
Message channels are better suited to applications that are based on a client/server
paradigm and for which location transparency is important.
18
807
VxWorks
Kernel Programmer's Guide, 6.6
808
Index
A
abort character (kernel shell) (CTRL+C) 590
changing default 589
abort character kernel shell) (CTRL+C) 589
access routines (POSIX) 265
ADDED_C++FLAGS 62
ADDED_CFLAGS
modifying run-time 62
affinity, CPU 691
interrupt 694
task 691
aio_cancel( ) 383
AIO_CLUST_MAX 382
aio_error( ) 385
testing completion 388
aio_fsync( ) 383
AIO_IO_PRIO_DFLT 383
AIO_IO_STACK_DFLT 383
AIO_IO_TASKS_DFLT 383
aio_read( ) 383
aio_return( ) 385
aio_suspend( ) 383
testing completion 388
AIO_TASK_PRIORITY 383
AIO_TASK_STACK_SIZE 383
aio_write( ) 383
aiocb, see control block (AIO)
aioPxLibInit( ) 383
aioShow( ) 383
aioSysDrv 382
aioSysInit( ) 382
ANSI C
function prototypes 53
header files 54
stdio package 378
application modules
linking 63
make variables 62
makefiles
include files, using 62
application modules, see object modules
applications
building kernel-based 62
configuring to run automatically 66
downloading kernel application modules 64
kernel component requirements 62
kernel-based 51
linking with VxWorks 64
starting automatically 66
structure for VxWorks-based applications 52
architecture,kernel 9
archive file attribute (dosFs) 494
ARCHIVE property (component object) 83
dummy component, creating a 74
using 70
809
VxWorks
Kernel Programmer's Guide, 6.6
asynchronous I/O (POSIX) 381
see also control block (AIO)
see online aioPxLib
cancelling operations 385
code examples 385
completion, determining 385
control block 384
driver, system 382
initializing 382
constants for 383
multiple requests, submitting 385
retrieving operation status 385
routines 382
atomic memory operations 690
attribute (POSIX)
prioceiling attribute 275
protocol attribute 274
attributes (POSIX) 265
specifying 266
AUTH_UNIX (RPC) 531
authentication, NFS 531
autosizing RAM 327
B
backplane network, see shared-memory networks
backspace character, see delete character
binary semaphores 201
BLK_DEV
creating a block device 462, 483
block devices 410–421
see also BLK_DEV; direct-access devices; disks;
SCSI devices; SEQ_DEV; sequential
devices
adding 429
code example 463, 486
creating 462, 483
defined 423
file systems, and 452–521
internal structure
drivers 426
naming 364
RAM disks 411
SCSI devices 412–421
810
board support packages (BSP)
make variables 62
boot loader 132
booting VxWorks 132, 136, 140
from host with TSFS 155
from network 152
from target file system 154
building 146
commands 132, 136
customizing 146
image types 133
parameters 132, 140
boot programs
TSFS, for 155
bootable applications
size of 65
bootImageLen 555
booting
rebooting VxWorks 145
bootrom 134
bootrom image
see also boot loader 134
bootrom_res 134
bootrom_uncmp 134
byte order
shared-memory objects (VxMP option)
720
C
C and C++ libraries, Dinkum 52
C++ development
C and C++, referencing symbols between
Run-Time Type Information (RTTI) 656
C++ support 647–658
see also iostreams (C++)
configuring 648
munching 650
static constructors 652
static destructors 652
cache
see also data cache
see online cacheLib
coherency 445
copyback mode 445
649
Index
writethrough mode 445
CACHE_DMA_FLUSH 448
CACHE_DMA_INVALIDATE 448
CACHE_DMA_PHYS_TO_VIRT 448
CACHE_DMA_VIRT_TO_PHYS 448
CACHE_FUNCS structure 448
cacheDmaMalloc( ) 448
cacheFlush( ) 446
cacheInvalidate( ) 446
cancelling threads (POSIX) 270
CD-ROM devices 510
cdromFs file systems 510
see online cdromFsLib
CFG_PARAMS property (component object) 83
character devices 423
see also drivers
adding 429
driver internal structure 426
naming 364
characters, control (CTRL+x)
tty 394
characters, control (CTRL+x)
kernel shell 586
checkStack( ) 243
CHILDREN property
folder object 93
selection object 95
CHILDREN property
using 73
CHILDREN property (component object) 84
client-server communications 217
CLOCK_MONOTONIC 260
CLOCK_REALTIME 260
clocks
see also system clock; clockLib(1)
monotonic 260
POSIX 259–262
real-time 260
system 182
close( )
example 441
using 371
closedir( ) 474, 493
clusters
cluster groups 496
disk space, allocating (dosFs) 496
absolutely contiguous 496
methods 496
nearly contiguous 496
single cluster 496
extents 496
code
interrupt service, see interrupt service routines
pure 189
shared 187
code example
device list 429
code examples
asynchronous I/O completion, determining
pipes, using 385
signals, using 388
data cache coherency 447
address translation driver 449
disk partitions
creating 490
formatting 490
dosFs file systems
block device, initializing 463, 486
file attributes, setting 495
maximum contiguous areas, finding 499
RAM disks, creating and formatting 491
drivers 424
makefiles
skeleton for application modules 63
message queues
attributes, examining (POSIX) 301, 302
POSIX 305
shared (VxMP option) 730
VxWorks 216
mutual exclusion 203
partitions
system (VxMP option) 735
user-created (VxMP option) 738
SCSI devices, configuring 417
select facility
driver code using 443
811
Index
VxWorks
Kernel Programmer's Guide, 6.6
semaphores
binary 203
named 296
recursive 208
shared (VxMP option) 726
unnamed (POSIX) 292
tasks
deleting safely 181
round-robin time slice (POSIX) 288
synchronization 203
threads
creating, with attributes 266–267
watchdog timers
creating and setting 240
COMP 791
component description files (CDF)
binding new CDFs to existing objects 81
conventions 80
paths, assigning 77
precedence 76
Component Description Language (CDL) 79
conventions 80
component object
contents 82
header files, specifying 70
naming 69
object code, specifying 69
parameters. declaring 72
source code, specifying 70
synopsis, providing a 69
syntax 82
components 67
archives, working with 74
custom kernel 67
dependencies
object module, analyzing 69
setting, explicitly 72
group membership, defining 73
initialization routine, specifying an 70
initialization sequence, setting
properties, using CDL object 71
modifying 74
parameters, defining 72
reference entries, linking 71
testing 78
812
VxWorks 17
CONFIGLETTES property (component object) 83
initialization routine, specifying an 70
using 70
configuration
C++ support 648
event 220
shared-memory objects (VxMP option) 744
signals 228
small VxWorks configuration profiles 24
configuration and build
components 5
tools 5
configuring
dosFs file systems 478, 480
HRFS file systems 459
kernel shell, with 581
SCSI devices 412–420
TSFS 520
contexts
task 160
CONTIG_MAX 498
control block (AIO) 384
fields 384
control characters (CTRL+x)
tty 394
control characters (CTRL+x)
kernel shell 586
conventions
device naming 363
file naming 363
task names 177
copyback mode, data cache 445
COUNT property (selection object) 95
counting semaphores 208, 290
cplusCtors( ) 653
cplusStratShow( ) 653
cplusXtorSet( ) 653
CPU
information (SMP) 694
management (SMP) 694
CPU affinity 691
interrupt 694
task 691
creat( ) 372
Index
CTRL+C (abort) 589
CTRL+C kernel shell abort) 394
CTRL+D (end-of-file) 394
CTRL+H
delete character
kernel shell 586
tty 394
CTRL+Q (resume)
kernel shell 586
tty 394
CTRL+S (suspend)
kernel shell 586
tty 394
CTRL+U (delete line)
kernel shell 586
tty 394
CTRL+X (reboot)
kernel shell 586
tty 394
custom
kernel components 67
kernel libraries 67
system calls 100
customizing VxWorks code 39
D
daemons
network tNet0 12
remote login tRlogind 13
RPC tJobTask 13
RPC tPortmapd 13
target agent tWdbTask 12
telnet tTelnetd 13
data cache
see also cache; cacheLib(1)
coherency 445
code examples 447
device drivers 445
copyback mode 445
flushing 446
invalidating 446
shared-memory objects (VxMP option) 744
writethrough mode 445
data structures, shared 196
dbgHelp command 585
debugging
error status values 184–187
kernel shell 588
SMP code 698
DEFAULT property (parameter object) 88
DEFAULTS property
folder object 93
selection object 95
_DEFAULTS property (component object) 84
delayed tasks 163
delayed-suspended tasks 163
delete character (CTRL+H)
kernel shell 586
tty 394
delete-line character (CTRL+U)
kernel shell 586
tty 394
dependency, component
object modules, analyzing 69
setting, explicitly 72
DEV_HDR 429
device descriptors 429
device header 429
device list 429
devices
see also block devices; character devices; directaccess devices; drivers and specific
device types
accessing 363
adding 429
block 410–421
flash memory 543
character 423
creating
NFS 399
non-NFS 401
pipes 397
default 364
dosFs 364
internal structure 429
naming 363
network 399
NFS 399
813
Index
VxWorks
Kernel Programmer's Guide, 6.6
non-NFS 400
pipes 396
pseudo-memory 398
RAM disk 411
SCSI 412–421
serial I/O (terminal and pseudo-terminal) 392
sockets 402
working with, in VxWorks 391–402
Dinkum C and C++ libraries 52
direct-access devices
initializing for rawFs 506
RAM disks 411
disks
changing
dosFs file systems 491
file systems, and 452–521
mounting volumes 508
organization (rawFs) 505
RAM 411
reformatting for dosFs 463, 484
synchronizing
dosFs file systems 492
displaying information
disk volume configuration, about 492
TrueFFS flash file systems, about 549
distributed shared memory
architecture 757
callbacks 778
communication model 759
configuration 763
custom services 767
driver initialization 778
hardware interface APIs 786
hardware interface development 777
macro functions 769
management service 785
messaging protocols 769
messaging support 782
MUX registration 782
service APIs 770
service code example 771
service numbers 768
technology overview 756
documentation 4
DOS_ATTR_ARCHIVE 494
814
DOS_ATTR_DIRECTORY 494
DOS_ATTR_HIDDEN 494
DOS_ATTR_RDONLY 494
DOS_ATTR_SYSTEM 494
DOS_ATTR_VOL_LABEL 494
DOS_O_CONTIG 498
dosFs file systems 478
see also block devices; CBIO interface; clusters;
FAT tables
see online dosFsLib
code examples
block devices, initializing 463, 486
file attributes, setting 495
maximum contiguous area on devices,
finding the 499
RAM disk, creating and formatting 491
configuring 478, 480
crash recovery 499
creating 461, 482
devices, naming 364
directories, reading 493
disk space, allocating 496
methods 496
disk volume
configuration data, displaying 492
disks, changing 491
FAT tables 484
file attributes 493
inconsistencies, data structure 499
initializing 482
ioctl( ) requests, supported 476, 500
MSFT Long Names 484
open( ), creating files with 370
partitions, creating and mounting 463, 483
reformatting disks 463, 484
short names format (8.3) 485
starting I/O 493
subdirectories
creating 492
removing 493
synchronizing volumes 492
TrueFFS flash file systems 548
volumes, formatting 463, 484
dosFsCacheCreate( ) 485
dosFsCacheDelete( ) 485
Index
dosFsChkDsk( ) 485
dosFsDrvNum global variable 483
dosFsFmtLib 478
dosFsLib 478
dosFsShow( ) 492
dosFsVolFormat( ) 484
downloading
kernel-based application modules 64
downloading, see loading
dpartDevCreate( ) 463, 483
driver number 427
driver table 427
drivers 363
see also devices and specific driver types
asynchronous I/O 382
code example 424
data cache coherency 445
file systems, and 452–521
hardware options, changing 396
installing 427
internal structure 426
interrupt service routine limitations 245
memory 398
NFS 399
non-NFS network 400
pipe 396
pty (pseudo-terminal) 392
RAM disk 411
SCSI 412–421
tty (terminal) 392
VxWorks, available in 391
DSHM
see distributed shared memory
DSI 793
E
ED&R, see error detection and reporting
edit mode (kernel shell) 586
encryption
login password 592
end-of-file character (CTRL+D) 394
__errno( ) 185
errno 184–187, 246
563
and task contexts 185
example 186
return values 186
error detection and reporting 563
APIs for application code 572
error records 566
fatal error handling options 569
persistent memory region 565
error handling options 569
error records 566
error status values 184–187
errors
memory error detection 331
run-time error checking (RTEC) 338
ESCAPE key (kernel shell) 586
eventClear( ) 224, 225
eventReceive( ) 224, 225
events 219
accessing event flags 223
and object deletion 222
and show routines 225
and task deletion 223
configuring 220
defined 219
and interrupt service routines 250
receiving 220
from message queues 221
from semaphores 220
from tasks and ISRs 220
routines 224
sending 221
task events register 224
eventSend( ) 224, 225
exception handling 187
and interrupts 248
signal handlers 187
task tExcTask 12
EXCLUDES property (component object)
using 72
excTask( )
abort facility 590
exit( ) 180
extended block device, see XBD
815
Index
VxWorks
Kernel Programmer's Guide, 6.6
F
FAT tables (dosFs)
supported formats 484
fclose( ) 380
fd table 435
fd, see file descriptors
FD_CLR 375
FD_ISSET 375
FD_SET 375
FD_ZERO 375
fdopen( ) 379
fdprintf( ) 380, 381
FIFO
message queues, VxWorks 215
file descriptor table 366
file descriptors (fd) 365
see also files
see online ioLib
device drivers, and 435
fd table 435
internal structure 435
pending on multiple (select facility) 374
redirection 367
standard input/output/error 366
file pointers (fp) 379
file system monitor 455
file systems
see also ROMFS file system;dosFs file systems;
TRFS file system;rawFs file systems;
tapeFs file systems; Target Server File
System (TSFS); TrueFFS flash file
systems
block devices, and 452–521
drivers, and 452–521
files
attributes (dosFs) 493
closing 371
example 441
contiguous (dosFs)
absolutely 496
nearly 496
creating 372
deleting 372
exporting to remote machines 399
816
hidden (dosFs) 494
I/O system, and 363
naming 363
opening 369
example 436
reading from 372
example 439
remote machines, on 399
read-write (dosFs) 494
system (dosFs) 494
truncating 373
write-only (dosFs) 494
writing to 372
-fimplicit-templates compiler option
FIOATTRIBSET 495
FIOBAUDRATE 396
FIOCANCEL 396
FIOCONTIG 500
FIODISKCHANGE 509
FIODISKFORMAT 507, 509
FIOFLUSH 476, 500, 509
pipes, using with 397
tty devices, using with 396
FIOFSTATGET 476, 500
FTP or RSH, using with 401
NFS client devices, using with
FIOGETNAME 476, 500
FTP or RSH, using with 401
NFS client devices, using with
pipes, using with 397
tty devices, using with 396
FIOGETOPTIONS 396
FIOLABELGET 500
FIOLABELSET 501
FIOMKDIR 492
FIOMOVE 476, 501
FIONCONTIG 501
FIONFREE 476, 501
FIONMSGS 397
FIONREAD 476, 501
FTP or RSH, using with 401
NFS client devices, using with
pipes, using with 397
tty devices, using with 396
FIONWRITE 396
655
400
400
400
Index
FIOREADDIR 476, 501
FTP or RSH, using with 401
NFS client devices, using with 400
FIORENAME 477, 501
FIORMDIR 474, 493
FIOSEEK 509
FTP or RSH, using with 401
memory drivers, using with 398
NFS client devices, using with 400
FIOSELECT 441
FIOSETOPTIONS
tty devices, using with 396
tty options, setting 392
FIOSYNC
FTP or RSH, using with 401
NFS client devices, using with 400
FIOTRUNC 496
FIOUNSELECT 441
FIOWHERE 477, 501
FTP or RSH, using with 401
memory drivers, using with 398
NFS client devices, using with 400
flash file systems, see TrueFFS flash file systems
flash memory 543
floating-point support
interrupt service routine limitations 245
task options 174
flow-control characters (CTRL+Q and S)
kernel shell 586
tty 394
-fno-implicit-templates compiler option 655
-fno-rtti compiler option (C++) 656
folder object 92
fopen( ) 379
formatArg argument 552
formatFlags 554
formatParams 553
fppArchLib 245
fprintf( ) 381
fread( ) 379
free( ) 245
fstat( ) 474, 493
FSTAT_DIR 492
FTL_FORMAT 554
FTL_FORMAT_IF_NEEDED 554
FTP (File Transfer Protocol)
ioctl functions, and 401
network devices for, creating
ftruncate( ) 373, 496
fwrite( ) 379
401
G
getc( ) 379
global variables 190
GROUP_EXPORTS 529
H
hardware
interrupts, see interrupt service routines
HDR_FILES property (component object) 83
header files
ANSI 54
function prototypes 53
hiding internal details 55
nested 55
private 55
searching for 54
VxWorks 53
heartbeat, shared-memory
troubleshooting, for 754
help command 585
HELP property (component object) 84
using 71
hidden files (dosFs) 494
Highly Reliable File System 459
transactional operations 469
Highly reliable File System
commit policies 469
hooks, task
routines callable by 184
host shell (WindSh)
kernel shell, differences from 578
hostAdd( )
remote file systems and 531
817
Index
VxWorks
Kernel Programmer's Guide, 6.6
HRFS
commit policies 469, 471
configuring 460
creating 461
transactional operations 469, 471
HRFS file systems
configuring 459
directories, reading 474
initializing 462
starting I/O 474
subdirectories
removing 474
TrueFFS flash file systems 548
HRFS, see Highly Reliable File System 459
htonl( )
shared-memory objects (VxMP option) 721
I
-I compiler option 54
I/O system
see also I/O, asynchronous I/O 381
XBD component 402
image types, VxWorks 15
include files
see also header files
INCLUDE_ATA
configuring dosFs file systems 460, 479
INCLUDE_CACHE_ENABLE 744
INCLUDE_CDROMFS 511
INCLUDE_CPLUS 648
INCLUDE_CPLUS_LANG 648
INCLUDE_CTORS_DTORS 648
INCLUDE_DISK_UTIL 479
INCLUDE_DOSFS 478
INCLUDE_DOSFS_CACHE
creating a disk cache 485
INCLUDE_DOSFS_CHKDSK 479
INCLUDE_DOSFS_DIR_FIXED 479
INCLUDE_DOSFS_DIR_VFAT 479
INCLUDE_DOSFS_FAT 479
INCLUDE_DOSFS_FMT 479
INCLUDE_DOSFS_MAIN 479
INCLUDE_MSG_Q_SHOW 730
818
INCLUDE_MTD_AMD 550
INCLUDE_MTD_CFIAMD 550
INCLUDE_MTD_CFISCS 550
INCLUDE_MTD_I28F008 550
INCLUDE_MTD_I28F016 550
INCLUDE_NFS 399
INCLUDE_PCMCIA 450
INCLUDE_POSIX_AIO 382
INCLUDE_POSIX_AIO_SYSDRV 382
INCLUDE_POSIX_FTRUNCATE 373
INCLUDE_POSIX_MEM 264
INCLUDE_POSIX_MQ 299
INCLUDE_POSIX_SCHED 287
INCLUDE_POSIX_SEM 289
INCLUDE_POSIX_SIGNALS 228
INCLUDE_RAWFS 505
INCLUDE_RLOGIN 591
INCLUDE_SCSI 413
booting dosFs file systems 503
INCLUDE_SCSI_BOOT 413
booting dosFs file systems using 503
ROM size, increasing 414
INCLUDE_SCSI_DMA 413
INCLUDE_SCSI2 413
INCLUDE_SECURITY 592
INCLUDE_SEM_SHOW 725
INCLUDE_SHELL 581
INCLUDE_SIGNALS 228
INCLUDE_SM_OBJ 744, 752
INCLUDE_TAR 479
INCLUDE_TELNET 591
INCLUDE_TFFS 549
INCLUDE_TFFS_BOOT_IMAGE 550
INCLUDE_TFFS_MOUNT 549
INCLUDE_TFFS_SHOW 549
INCLUDE_TL_FTL 551
INCLUDE_TL_SSFDC 551
INCLUDE_TSFS_BOOT_VIO_CONSOLE 155
INCLUDE_USR_MEMDRV 398
INCLUDE_VXEVENTS 220
INCLUDE_WDB_TSFS 520
INCLUDE_WHEN property (component object)
84
using 72
INCLUDE_XBD 479
Index
INCLUDE_XBD_PARTLIB 460, 480
creating disk partitions 463, 483
INCLUDE_XBD_RAMDISK 460, 480
INCLUDE_XBD_TRANS 480
INIT_AFTER property (component object)
using 71
INIT_BEFORE property (component object) 84
using 71
INIT_ORDER property (initGroup object) 89
_INIT_ORDER property (component object) 84
using 71
INIT_RTN property
component object 83
using 70
initGroup object 89
initGroups, see initialization group object
initialization group object 88
syntax 89
initialization routine
specifying 70
initialization sequence
see also booting
setting with CDL object properties 71
initializing
asynchronous I/O (POSIX) 382
dosFs file system 482
HRFS file system 462
rawFs file systems 506
SCSI interface 415
installing drivers 427
instantiation, template (C++) 655
intConnect( ) 241, 242
intCount( ) 241
interpreters, kernel shell
C and command 577
interrupt handling
application code, connecting to 242
callable routines 241
and exceptions 248
hardware, see interrupt service routines
pipes, using 397
stacks 242
interrupt latency 197
interrupt levels 248–249
interrupt masking 248
interrupt service routines (ISR) 241–250
see also interrupt handling; interrupts;
intArchLib(1); intLib(1)
and events 250
limitations 245–248
logging 245
see also logLib(1)
and message queues 249
and pipes 250
routines callable from 247
and semaphores 249
shared-memory objects (VxMP option),
working with 720
and signals 229, 250
interrupt stacks 242
interrupts
locking 197
shared-memory objects (VxMP option) 745
task-level code, communicating to 249
VMEbus 242
intertask communications 193–228
network 226
intLevelSet( ) 241
intLock( ) 241
intLockLevelSet( ) 248
intUnlock( ) 241
intVecBaseGet( ) 241
intVecBaseSet( ) 241
intVecGet( ) 241
intVecSet( ) 241
I/O system 360
asynchronous I/O 381
basic I/O (ioLib) 365
buffered I/O 378
control functions (ioctl( )) 373
differences between VxWorks and host
system 422
fd table 435
formatted I/O (fioLib) 380
internal structure 423
memory, accessing 398
message logging 381
PCI (Peripheral Component Interconnect) 450
PCMCIA 450
redirection 367
819
Index
VxWorks
Kernel Programmer's Guide, 6.6
serial devices 392
stdio package (ansiStdio) 378
ioctl( ) 373
dosFs file system support 476, 500
functions
FTP, using with 401
memory drivers, using with 398
NFS client devices, using with 400
pipes, using with 397
RSH, using with 401
tty devices, using with 396
non-NFS devices 401
raw file system support 509
tty options, setting 392
ioDefPathGet( ) 364
ioDefPathSet( ) 364
iosDevAdd( ) 429
iosDevFind( ) 429
iosDrvInstall( ) 427
dosFs, and 482
ioTaskStdSet( ) 367
ISR, see interrupt service routines
K
kernel
architecture 9
custom components 67
downloading kernel application modules 64
image types 15
kernel-based applications 51
libraries, custom 67
and multitasking 160
POSIX and VxWorks features, comparison of
252
message queues 300
power management 40
priority levels 166
Kernel shell
interpreters
commands and references 578
kernel shell 577
see online dbgLib; dbgPdLib; shellLib; usrLib;
usrPdLib
820
aborting (CTRL+C) 589, 590
changing default 589
tty 394
accessing from host 591
C interpreter
command interpreter
configuring VxWorks with 581
control characters (CTRL+x) 586
debugging 588
edit mode, specifying
toggle between input mode 586
help, getting 585
host shell, differences from 578
interpreters 577
switching between 578
line editing 586
loading
object modules 587
remote login 591
restarting 589
task tShell 12
kernelTimeSlice( ) 167, 170
keyboard shortcuts
kernel shell 586
tty characters 394
kill( ) 230
killing
kernel shell, see abort character
L
latency
interrupt locks 197
preemptive locks 197
line editor (kernel shell) 586
line mode (tty devices) 393
selecting 393
LINK_SYMS property (component object) 84
using 70
linking
application modules 63
lio_listio( ) 383
loader, target-resident 603–615
loading
Index
object modules 587
local objects 717
locking
interrupts 197
page (POSIX) 264
semaphores 289
spin-lock mechanism (VxMP option) 743
task preemptive locks 167, 197
logging facilities 381
and interrupt service routines 245
task tLogTask 11
login
password, encrypting 592
remote
daemon tRlogind 13
security 592
shell, accessing kernel 591
loginUserAdd( ) 592
longjmp( ) 187
M
makefiles
code examples
skeleton for application modules 63
include files
application modules, and 62
variables, include file
customizing run-time, for 62, 64
malloc( )
interrupt service routine limitations 245
MAX_AIO_SYS_TASKS 383
MAX_LIO_CALLS 382
MEM_BLOCK_CHECK 741
memory
barriers 687
driver (memDrv) 398
flash 543
writing boot image to 556
locking (POSIX) 264
see also mmanPxLib(1)
management, seememory management
paging (POSIX) 264
persistent memory region 565
pool 190
pseudo-I/O devices 398
shared-memory objects (VxMP) 717
swapping (POSIX) 264
system memory maps 317
memory management
component requirements 317
error detection 331
kernel heap and partition 329
RAM autosizing 327
reserved 328
shell commands 327
virtual memory 343
memory management unit, seeMMU 355
Memory Technology Driver (MTD) (TrueFFS)
component selection 546
JEDEC device ID 546
options 550
memPartOptionsSet( ) 741
memPartSmCreate( ) 738
message channels 226, 787
message logging, see logging facilities
message queues 213
see also msgQLib(1)
and VxWorks events 218
client-server example 217
displaying attributes 217, 303
and interrupt service routines 249
POSIX 299
see also mqPxLib(1)
attributes 300, 302
code examples
attributes, examining 301, 302
communicating by message queue
305, 307
notifying tasks 307
unlinking 304
VxWorks facilities, differences from 300
priority setting 216
queuing 217
shared (VxMP option) 728–733
code example 730
creating 728
local message queues, differences from
728
821
Index
VxWorks
Kernel Programmer's Guide, 6.6
VxWorks 215
code example 216
creating 215
deleting 215
queueing order 215
receiving messages 215
sending messages 215
timing out 215
waiting tasks 215
migrating
code to SMP 702
ml( ) 587
mlock( ) 264
mlockall( ) 264
mmanPxLib 264
MMU
processes without 355
shared-memory objects (VxMP option) 745
MODULES property (component object) 83
using 69
modules, see component modules; object modules
mount1Lib 537
mount3Lib 537
mounting file systems 530
mounting volumes
rawFs file systems 508
mq_close( ) 299, 303
mq_getattr( ) 299, 300
mq_notify( ) 299, 307
mq_open( ) 299, 303
mq_receive( ) 299, 303
mq_send( ) 299, 303
mq_setattr( ) 299, 300
mq_unlink( ) 299, 304
mqPxLib 299
mqPxLibInit( ) 299
MSFT Long Names format 484
msgQCreate( ) 215
msgQDelete( ) 215
msgQEvStart( ) 224
msgQEvStop( ) 224
msgQReceive( ) 215
msgQSend( ) 215
msgQSend( ) 225
msgQShow( ) 730
822
msgQSmCreate( ) 728
multitasking 160, 187
example 192
munching (C++) 650
munlock( ) 264
munlockall( ) 264
mutexes (POSIX) 273
mutual exclusion 196
see also semLib(1)
code example 203
counting semaphores 208
interrupt locks 197
preemptive locks 197
and reentrancy 190
VxWorks semaphores 205
binary 203
deletion safety 207
priority inheritance 206
priority inversion 205
recursive use 207
N
name database (VxMP option) 720–722
adding objects 721
displaying 722
NAME property
component object 82
folder object 93
initGroup object 89
parameter object 87
selection object 95
named semaphores (POSIX) 289
using 294
nanosleep( ) 181, 182
using 262
netDevCreate( ) 401
netDrv
compared with TSFS 520
netDrv driver 400
network devices
see also FTP; NFS; RSH
NFS 399
non-NFS 400
Index
Network File System 523
Network File System, see NFS
network task tNet0 12
networks
intertask communications 226
transparency 399
NFS 523
NFS (Network File System)
see online nfsDrv; nfsLib
authentication 531
authentication parameters 399
client, target as 530
devices 399
creating 399
naming 364
open( ), creating files with 370
group IDs, setting 531
ioctl functions, and 400
mounting file systems 530
network devices, creating 531
server
setting up 540
transparency 399
user IDs, setting 531
NFS server
reply cache 536
NFS_GROUP_ID 531
core NFS client 524
NFS server 537
NFS_MAXFILENAME 536
NFS_MAXPATH
NFS core client 524
NFS server 536
NFS_USER_ID 524, 536
NFS2_CLIENT_CACHE_DEFAULT_LINE_SIZE
525
NFS2_CLIENT_CACHE_DEFAULT_NUM_
LINES 525
NFS2_CLIENT_CACHE_DEFAULT_OPTIONS
526
nfs2dLib 537
nfs2Drv 525
nfs2Lib 525
NFS3_CLIENT_CACHE_DEFAULT_LINE_SIZE
527
NFS3_CLIENT_CACHE_DEFAULT_NUM_
LINES 527
NFS3_CLIENT_CACHE_DEFAULT_OPTIONS
528
nfs3dLib 537
nfs3Drv 526
nfs3Lib 526
nfs3StableWriteGet( ) 528
nfs3StableWriteSet( ) 528
nfsAuthUnixPrompt( ) 399, 531
nfsAuthUnixSet( ) 399, 531
nfsCommon 524
nfsDrv driver 399
nfsExport( ) 537, 540
nfsMount( ) 399, 531
nfsMountAll( ) 529
NO_FTL_FORMAT 554
non-block devices, see character devices
ntohl( )
shared-memory objects (VxMP option) 721
null devices 402
NUM_RAWFS_FILES 506
O
O_CREAT 492
O_NONBLOCK 300
O_CREAT 294
O_EXCL 294
O_NONBLOCK 303
object code, specifying 69
archive, from an 70
object ID (VxMP option) 718
object modules
loading dynamically 587
open( ) 369
access flags 369
example 436
files asynchronously, accessing
files with, creating 370
subdirectories, creating 492
opendir( ) 474, 493
operating system 264
OPT_7_BIT 393
823
Index
382
VxWorks
Kernel Programmer's Guide, 6.6
OPT_ABORT 393
OPT_CRMOD 393
OPT_ECHO 393
OPT_LINE 393
OPT_MON_TRAP 393
OPT_RAW 393
OPT_TANDEM 393
OPT_TERMINAL 393
optimizing performance (SMP) 699
optional components (TrueFFS)
options 549
P
page locking 264
see also mmanPxLib(1)
paging 264
parameter object 87
working with 72
partitions, disk
code examples
creating disk partitions 490
formatting disk partitions 490
passwd 531
password encryption
login 592
pause( ) 229
PCI (Peripheral Component Interconnect) 450
see online pciConfigLib; pciConfigShow;
pciInitLib
PCMCIA 450
see online pcmciaLib; pcmciaShow
pdHelp command 585
pended tasks 163
pended-suspended tasks 163
persistent memory region 565
pipeDevCreate( ) 219
pipes 218–219
see online pipeDrv
interrupt service routines 250
ioctl functions, and 397
ISRs, writing from 397
select( ), using with 219
polling
824
shared-memory objects (VxMP option) 745
POSIX
see also asynchronous I/O
and kernel 252
asynchronous I/O 381
clocks 259–262
see also clockLib(1)
file truncation 373
memory-locking interface 264
message queues 299
see also message queues; mqPxLib(1)
mutex attributes 273
prioceiling attribute 275
protocol attribute 274
page locking 264
see also mmanPxLib(1)
paging 264
priority limits, getting task 287
priority numbering 279
scheduling 286
see also scheduling; schedPxLib(1)
semaphores 289
see also semaphores; semPxLib(1)
signal functions 230
see also signals; sigLib(1)
routines 229
swapping 264
thread attributes 265–267
specifying 266
threads 264
timers 259–262
see also timerLib(1)
posixPriorityNumbering global variable 279
power management 40
precedence, component description file 76
preemptive locks 167, 197
preemptive priority scheduling 169
printErr( ) 380, 381
printErrno( ) 186
printf( ) 380
prioceiling attribute 275
priority
inheritance 206
inversion 205
message queues 216
Index
numbering 279
preemptive, scheduling 169
task, setting
VxWorks 166
prjConfig.c 89
processes
POSIX and 279
real-time 9
without MMU support 355
protocol attribute 274
pthread_attr_getdetachstate( ) 269
pthread_attr_getinheritsched( ) 269
pthread_attr_getschedparam( ) 269
pthread_attr_getscope( ) 269
pthread_attr_getstackaddr( ) 269
pthread_attr_getstacksize( ) 269
pthread_attr_setdetachstate( ) 269
pthread_attr_setinheritsched( ) 269
pthread_attr_setschedparam( ) 269
pthread_attr_setscope( ) 269
pthread_attr_setstackaddr( ) 269
pthread_attr_setstacksize( ) 269
pthread_attr_t 265
pthread_cancel( ) 272
pthread_cleanup_pop( ) 272
pthread_cleanup_push( ) 272
pthread_getspecific( ) 270
pthread_key_create( ) 270
pthread_key_delete( ) 270
pthread_mutex_getprioceiling( ) 275
pthread_mutex_setprioceiling( ) 275
pthread_mutexattr_getprioceiling( ) 275
pthread_mutexattr_getprotocol( ) 274
pthread_mutexattr_setprioceiling( ) 275
pthread_mutexattr_setprotocol( ) 274
pthread_mutexattr_t 273
PTHREAD_PRIO_INHERIT 274
PTHREAD_PRIO_NONE 274
PTHREAD_PRIO_PROTECT 274
pthread_setcancelstate( ) 272
pthread_setcanceltype( ) 272
pthread_setspecific( ) 270
pthread_testcancel( ) 272
pty devices 392
see online ptyDrv
public objects
tasks 177
publishing (VxMP option) 720
pure code 189
putc( ) 379
Q
queuedsignals 230
queues
see also message queues
ordering (FIFO vs. priority) 212
semaphore wait 212
queuing
message queues 217
R
-r linker option 64
-R option (TSFS) 521
raise( ) 229
RAM autosizing 327
RAM disks
see online ramDrv
code example 491
drivers 411
raw mode (tty devices) 393
rawFs file systems 505–509
see online rawFsLib
disk organization 505
disk volume, mounting 508
initializing 506
ioctl( ) requests, support for 509
starting I/O 509
rawFsDevInit( ) 507
rawFsDrvNum global variable 506
rawFsInit( ) 506
read( ) 372
example 439
read/write semaphores 209
readdir( ) 474, 493
ready tasks 163
825
Index
VxWorks
Kernel Programmer's Guide, 6.6
reboot character (CTRL+X)
kernel shell 586
tty 394
rebooting VxWorks 145
redirection 367
task-specific 367
reentrancy 188–192
reference entries
linking to a component 71
remote file system
mounting 531
remote login
daemon tRlogind 13
security 592
shell, accessing kernel 591
remove( ) 372
subdirectories, removing 474, 493
REQUIRES property (component object) 84
using 72
restart character (CTRL+C)
tty 394
restart character (kernel shell) (CTRL+C) 589, 590
changing default 589
resume character (CTRL+Q)
kernel shell 586
tty 394
rewinddir( ) 474, 493
RFC 1094
NFS v2 server 537
RFC 1813
NFS v3 server, mount protocol 538
supported client requests 538
ring buffers 246, 249
rlogin 591
rlogin (UNIX) 591
ROM
VxWorks in 16
ROM monitor trap (CTRL+X)
kernel shell 586
tty 394
ROM_SIZE
system images, creating 65
ROMFS file system 516
root task tUsrRoot 11
826
round-robin scheduling
defined 169
routines
scheduling, for 286
RPC (Remote Procedure Calls)
daemon tPortmapd 13
RSH (Remote Shell protocol)
ioctl functions, and 401
network devices for, creating 401
RTEC, seerun-time error checking 338
RTP, see processes
run-time error checking (RTEC) 338
Run-Time Type Information (RTTI) 656
-RW option (TSFS) 521
S
SAL 799
scalable configuration profiles 24
scanf( ) 381
sched_get_priority_max( ) 286
sched_get_priority_max( ) 287
sched_get_priority_min( ) 286
sched_get_priority_min( ) 287
sched_rr_get_interval( ) 286
sched_rr_get_interval( ) 287
schedPxLib 279, 286
schedulers 166
scheduling 166
POSIX 286
see also schedPxLib(1)
algorithms 279
priority limits 287
priority numbering 279
routines for 286
time slicing 287
VxWorks
preemptive locks 197
preemptive priority 169
round-robin 169
Wind
preemptive locks 167
Index
SCSI devices 412–421
see online scsiLib
booting from
ROM size, adjusting 414
bus failure 421
configuring 412–420
code examples 417
options 415
initializing support 415
libraries, supporting 414
SCSI bus ID
changing 420
configuring 413
SCSI-1 vs. SCSI-2 413
tagged command queuing 417
troubleshooting 420
VxWorks image size, affecting 413
wide data transfers 417
SCSI_AUTO_CONFIG 413
SCSI_OPTIONS structure 415
SCSI_TAG_HEAD_OF_QUEUE 417
SCSI_TAG_ORDERED 417
SCSI_TAG_SIMPLE 417
SCSI_TAG_UNTAGGED 417
scsi1Lib 414
scsi2Lib 414
scsiBlkDevCreate( ) 415
scsiCommonLib 414
scsiDirectLib 414
scsiLib 414
scsiPhysDevCreate( ) 415
scsiSeqLib 414
scsiTargetOptionsSet( ) 415
SCSI bus failure 421
security 592
TSFS 520
SEL_WAKEUP_LIST 441
SEL_WAKEUP_NODE 442
select facility 374
see online selectLib
code example
driver support of select( ) 443
macros 375
select( )
and pipes 219
select( ) 374
implementing 441
selection object 94
count, setting the 74
selNodeAdd( ) 442
selNodeDelete( ) 442
selWakeup( ) 442
selWakeupAll( ) 442
selWakeupListInit( ) 441
selWakeupType( ) 442
sem_close( ) 289, 295
SEM_DELETE_SAFE 207
sem_destroy( ) 289
sem_getvalue( ) 289
sem_init( ) 289, 291
SEM_INVERSION_SAFE 206
sem_open( ) 289, 294
sem_post( ) 289
sem_trywait( ) 289
sem_unlink( ) 289, 295
sem_wait( ) 289
semaphores 198
and VxWorks events 213
see also semLib(1)
counting 290
example 209
deleting 201, 290
giving and taking 201–202, 289
and interrupt service routines 249, 245
locking 289
POSIX 289
see also semPxLib(1)
named 289, 294
code example 296
unnamed 289, 290, 291–293
code example 292
posting 289
read/write 209
recursive 207
code example 208
shared (VxMP option) 722–727
code example 726
creating 723
displaying information about 725
local semaphores, differences from 724
827
Index
VxWorks
Kernel Programmer's Guide, 6.6
synchronization 198, 208
code example 203
unlocking 289
VxWorks 198
binary 201
code example 203
control 200
counting 208
mutual exclusion 203, 205
queuing 212
synchronization 203
timing out 212
waiting 289
semBCreate( ) 200
semBSmCreate( ) (VxMP option) 723
semCCreate( ) 200
semCSmCreate( ) (VxMP option) 723
semDelete( ) 200
shared semaphores (VxMP option) 724
semEvStart( ) 224
semEvStop( ) 224
semExchange( ) 200
semFlush( ) 200, 205
semGive( ) 200
semGive( ) 225
semInfo( ) 724
semMCreate( ) 200
semPxLib 289
semPxLibInit( ) 290
semRGive( ) 200
semRTake( ) 200
semRWCreate( ) 200
semShow( ) 724
semTake( ) 200
semWTake( ) 200
serial drivers 392
setjmp( ) 187
shared code 187
shared data structures 196
shared memory
distributed shared memory
shared memory objects 717
shared message queues (VxMP option) 728–733
code example 730
creating 728
828
displaying queue status 730
local message queues, differences from 728
shared semaphores (VxMP option) 722–727
code example 726
creating 723
displaying information about 725
local semaphores, differences from 724
shared-memory allocator (VxMP option) 733–741
shared-memory anchor
shared-memory objects, configuring (VxMP
option) 745–746
shared-memory networks
shared-memory objects, working with 743
shared-memory objects (VxMP option)
anchor, configuring shared-memory 745–746
cacheability 744, 745
configuring 744
constants 747
multiprocessor system 750
displaying number of used objects 752
heartbeat
troubleshooting, for 754
interrupt latency 743
interrupt service routines 720
interrupts
bus 745
mailbox 745
limitations 720–744
locking (spin-lock mechanism) 743
memory
allocating 733–741
running out of 720
memory layout 747
message queues, shared 728–733
see also shared message queues
code example 730
name database 720–722
object ID 718
partitions 733–741
routines 735
side effects 741
system 733–738
code example 735
user-created 734, 738–740
code example 738
Index
polling 745
publishing 720
semaphores, shared 722–727
see also shared semaphores (VxMP option)
code example 726
shared-memory networks, working with 743
shared-memory region 746
single- and multiprocessors, using with 719
system requirements 741
troubleshooting 752
types 722
shared-memory objects (VxMP) 717
shared-memory region (VxMP option) 746
shell
commands, for memory management 327
shell, see host shell; kernel shell
show( ) 217, 294
sigaction( ) 229
sigaddset( ) 229
sigdelset( ) 229
sigemptyset( ) 229
sigfillset( ) 229
sigInit( ) 228
sigismember( ) 229
signal handlers 229
signal( ) 229
signals 226, 226–228
see also sigLib(1)
configuring 228
and interrupt service routines 229, 250
POSIX 230
queued 230
routines 229
signal handlers 229
UNIX BSD 227
routines 229
sigpending( ) 229
sigprocmask( ) 229
sigqueue( ) 230
sigqueue( )
buffers to, allocating 228
sigqueueInit( ) 228
sigsuspend( ) 229
sigtimedwait( ) 230
sigvec( ) 229
sigwaitinfo( ) 230
SIO_HW_OPTS_SET 396
SM_ANCHOR_ADRS 746
SM_INT_BUS 745
SM_INT_MAILBOX 745
SM_INT_NONE 745
SM_INT_TYPE 745
SM_OBJ_MAX_MEM_PART 747
SM_OBJ_MAX_MSG_Q 747
SM_OBJ_MAX_NAME 747
SM_OBJ_MAX_SEM 747
SM_OBJ_MAX_TASK 747
SM_OBJ_MAX_TRIES 753
SM_TAS_HARD 742
SM_TAS_TYPE 742
small computer system interface, see SCSI devices
small VxWorks configurations 24
smCpuInfoGet( ) (VxMP option) 745
smMemAddToPool( ) (VxMP option) 735
smMemCalloc( ) (VxMP option) 735
smMemFindMax( ) (VxMP option) 735
smMemFree( ) (VxMP option) 735
smMemMalloc( ) (VxMP option) 735
smMemOptionsSet( ) (VxMP option) 735, 741
smMemRealloc( ) (VxMP option) 735
smMemShow( ) (VxMP option) 735
smNameAdd( ) (VxMP option) 721
smNameFind( ) (VxMP option) 721
smNameFindByValue( ) (VxMP option) 721
smNameRemove( ) (VxMP option) 721
smNameShow( ) (VxMP option) 721
smObjShow( ) (VxMP option) 752
troubleshooting, for 754
smObjTimeoutLogEnable( ) (VxMP option) 754
SMP
atomic memory operations 690
booting 676
comparison with AMP 671
configuration and build 674
CPU affinity 691
interrupt 694
task 691
CPU information 694
CPU management 694
CPU mutual exclusion, interrupts 685
829
Index
VxWorks
Kernel Programmer's Guide, 6.6
CPU-specific mutual exclusion 685
interrupts 685
tasks 686
debugging code 698
hardware 669
memory barriers 687
migrating code 702
optimizing performance 699
programming 676
sample programs 702
spinlocks 679
caveats 683
debug 681
ISR-callable 682
routine restrictions 683
task-only 682
types 680
spinlocks as memory barriers 680
technology overview 666
terminology 667
VxWorks operating system features 668
SNS 795
socket component drivers (TrueFFS)
translation layer 547
socket( ) 402
sockets
I/O devices, as 402
TSFS 519
source code (VxWorks)
customizing 39
source-scalable configuration profiles 24
spawning tasks 172–173, 192
spin-lock mechanism (VxMP option) 743
interrupt latency 743
spinlocks 679
caveats 683
debug 681
ISR-callable 682
memory barriers 680
routine restrictions 683
task-only 682
types 680
sprintf( ) 380
sscanf( ) 380
830
stacks
interrupt 242
no fill 174
standard input/output/error
basic I/O 366
buffered I/O (ansiStdio) 380
stat( ) 474, 493
stdio package
ANSI C support 378
omitting 381
printf( ) 380
sprintf( ) 380
sscanf( ) 380
stopped tasks 163
subdirectories (dosFs)
creating 492
file attribute 494
suspended tasks 163
swapping 264
symmetric multiprocessing, see SMP
synchronization (task) 198
code example 203
counting semaphores, using 208
semaphores 203
synchronizing media
dosFs file systems 492
SYNOPSIS property
component object 82
folder object 93
initGroup object 89
selection object 95
SYS_SCSI_CONFIG 503
sysIntDisable( ) 242
sysIntEnable( ) 242
sysPhysMemDesc[ ]
shared-memory objects (VxMP option)
sysScsiConfig( ) 413
sysScsiInit( ) 415
system
image types 15
system calls
custom 100
system clock 182
system files (dosFs) 494
745
Index
system images
boot loader
compressed 134
ROM-resident 134
uncompressed 134
boot ROM
uncompressed 134
system startup 13
system tasks 11
sysTffs.c 547
sysTffsFormat( ) 554
sysTffsInit( ) 552
T
T_SM_BLOCK 722
T_SM_MSG_Q 722
T_SM_PART_ID 722
T_SM_SEM_B 722
T_SM_SEM_C 722
tape devices
SCSI, supporting 413
tapeFs file systems
SCSI drivers, and 413
target
name (tn) (boot parameter) 143
target agent
task (tWdbTask) 12
target agent, see WDB 626
Target Server File System (TSFS) 518
boot program for, configuring 155
configuring 520
error handling 520
file access permissions 520
sockets, working with 519
task control blocks (TCB) 160, 179, 182, 245
task variables
__thread storage class 190
taskActivate( ) 173
taskCreate( ) 173
taskCreateHookAdd( ) 183
taskCreateHookDelete( ) 183
taskDelay( ) 181
taskDelete( ) 180
taskDeleteHookAdd( ) 183
taskDeleteHookDelete( ) 183
taskIdListGet( ) 179
taskIdSelf( ) 178
taskIdVerify( ) 178
taskInfoGet( ) 179
taskIsPended( ) 179
taskIsReady( ) 179
taskIsSuspended( ) 179
taskLock( ) 167
taskName( ) 178
taskNameToId( ) 178
taskOptionsGet( ) 175
taskOptionsSet( ) 175
taskPriorityGet( ) 179
taskPrioritySet( ) 167
taskRegsGet( ) 179
taskRegsSet( ) 179
taskRestart( ) 181
taskResume( ) 181
taskRotate( ) 167
tasks
__thread task variables 190
blocked 167
contexts 160
control blocks 160, 179, 182, 245
creating 172–173
delayed 163
delayed-suspended 163
delaying 161, 163, 181, 239–240
deleting safely 180–181
code example 181
semaphores, using 207
displaying information about 179
error status values 184–187
see also errnoLib(1)
exception handling 187
see also signals; sigLib(1); excLib(1)
tExcTask 12
executing 181
hooks
see also taskHookLib(1)
extending with 182–184
troubleshooting 183
IDs 177
831
Index
VxWorks
Kernel Programmer's Guide, 6.6
interrupt level, communicating at 249
pipes 397
kernel shell (tShell) 12
logging (tLogTask) 11
names 177
automatic 178
private 177
public 177
network (tNet0) 12
option parameters 173
pended 163
pended-suspended 163
priority inversion safe (tJobTask) 13
priority, setting
application tasks 166
driver support tasks 166
VxWorks 166
public 177
ready 163
remote login (tRlogind, tRlogInTask,
tRlogOutTask) 13
root (tUsrRoot) 11
RPC server (tPortmapd) 13
scheduling
POSIX 286
preemptive locks 167, 197
preemptive priority 169
priority limits, getting 287
round-robin 169
time slicing 287
VxWorks 166
shared code 187
and signals 187, 226–228
spawning 172–173, 192
stack allocation 175
stack protection 175
states 161–163
stopped 163
suspended 163
suspending and resuming 181
synchronization 198
code example 203
counting semaphores, using 208
system 11
832
target agent (tWdbTask) 12
task events register 224
API 225
telnet (tTelnetd, tTelnetInTask,
tTelnetOutTask) 13
variables 190
taskSafe( ) 180
taskSpawn( ) 172
taskStatusString( ) 179
taskSuspend( ) 181
taskSwitchHookAdd( ) 183
taskSwitchHookDelete( ) 183
taskTcb( ) 179
taskUnlock( ) 167
taskUnsafe( ) 180, 181
telnet 591
daemon tTelnetd 13
terminal characters, see control characters
TFFS_STD_FORMAT_PARAMS 553
tffsBootImagePut( ) 550
tffsDevCreate( ) 557
tffsDevFormat( ) 552
tffsDevFormatParams 552
tffsDriveNo argument 552
tffsRawio( ) 556
tffsShow( ) 549
tffsShowAll( ) 549
__ 190
thread-local variable 190
threads (POSIX) 264
attributes 265–267
specifying 266
keys 270
private data, accessing 270
terminating 270
time slicing 170
determining interval length 287
timeout
message queues 215
semaphores 212
timeouts
semaphores 212
timers
see also timerLib(1)
message queues, for (VxWorks) 215
Index
POSIX 259–262
semaphores, for (VxWorks) 212
watchdog 239–240
code examples 240
TIPC
and WDB target agent 638
tools
configuration and build 5
tools, target-based development 576–646
Transaction-Based Reliable File System, see TRFS
405
translation layers (TrueFFS)
options 551
TRFS file system 405
troubleshooting
SCSI devices 420
shared-memory objects (VxMP option) 752
TrueFFS flash file systems 543
and boot image region 554
boot image region
creating 554
writing to 556
building
device formatting 552
drive mounting 556
Memory Technology Driver (MTD) 546
overview 545
socket driver 547
displaying information about 549
drives
attaching to dosFs 556
formatting 552
mounting 556
numbering 552
Memory Technology Driver (MTD)
component selection 546
JEDEC device ID 546
truncation of files 373
tty devices 392
see online tyLib
control characters (CTRL+x) 394
ioctl( ) functions, and 396
line mode 393
selecting 393
options 392
all, setting 393
none, setting 393
raw mode 393
X-on/X-off 393
tyAbortSet( ) 589
tyBackspaceSet( ) 395
tyDeleteLineSet( ) 395
tyEOFSet( ) 395
tyMonitorTrapSet( ) 395
TYPE property (parameter object) 87
U
unnamed semaphores (POSIX) 289, 290, 291–293
usrAppInit( ) 66
usrFdiskPartCreate( ) 463, 483
usrScsiConfig( ) 415
usrTffsConfig( ) 556
V
variables
__thread task variables 190
global 190
static data 190
task 190
virtual memory, see memory management, virtual
memory
Virtual Root File System 457
VM, seememory management, virtual memory
VMEbus interrupt handling 242
volume labels (dosFs)
file attribute 494
VX_ALTIVEC_TASK 174
VX_DSP_TASK 174
VX_FP_TASK 174, 649
VX_FP_TASK option 174
VX_GLOBAL_NO_STACK_FILL 175
VX_NO_STACK_FILL 174
VX_PRIVATE_ENV 174
VX_UNBREAKABLE 174
vxencrypt 592
833
Index
VxWorks
Kernel Programmer's Guide, 6.6
VxMP shared-memory objects 717
VxMP, see shared-memory objects (VxMP option)
vxsize command 65
VxWorks
components 5, 17
components, and application requirements 62
configuration 14
configuration and build 5
configuring applications to run automatically
66
customizing code 39
header files 53
image types 15
linking applications with 64
message queues 215
rebooting 145
VxWorks events, see events
VxWorks facilities
POSIX, differences from
message queues 300
scheduling 166
VxWorks kernel, see kernel
VxWorks SMP, see SMP
vxWorks.h 54
vxWorks_ROM 134
vxWorks_romCompress 134
vxWorks_romCompress image
see also boot loader 134
W
WAIT_FOREVER 212
watchdog timers 239–240
code examples
creating a timer 240
WDB
target agent proxy 638
WDB target agent 626
and exceptions 640
scaling 639
starting before kernel 640
wdCancel( ) 240
wdCreate( ) 240
wdDelete( ) 240
834
wdStart( ) 240
workQPanic 248
write( ) 372
pipes and ISRs 397
writethrough mode, cache 445
X
XBD I/O component 402
Source Exif Data:
File Type : PDF File Type Extension : pdf MIME Type : application/pdf PDF Version : 1.4 Linearized : Yes Page Mode : UseOutlines XMP Toolkit : 3.1-701 Producer : Acrobat Distiller 7.0.5 (Windows) Creator Tool : FrameMaker 7.2 Modify Date : 2007:11:14 12:34:34Z Create Date : 2007:11:14 12:34:27Z Format : application/pdf Title : VxWorks Kernel Programmer's Guide, 6.6 Creator : Wind River Description : VxWorks 6.6 Document ID : uuid:2a36f8d6-0ac9-4e7a-8159-fa02f3f7c2ae Instance ID : uuid:9a98f187-6d8e-4596-bb19-7626c3f2cb5b Page Count : 866 Subject : VxWorks 6.6 Author : Wind RiverEXIF Metadata provided by EXIF.tools