MPLAB XC8 C Compiler User's Guide User
User Manual: Pdf
Open the PDF directly: View PDF .
Page Count: 580
Download | |
Open PDF In Browser | View PDF |
MPLAB® XC8 C Compiler User’s Guide 2012-2013 Microchip Technology Inc. DS50002053D Note the following details of the code protection feature on Microchip devices: • Microchip products meet the specification contained in their particular Microchip Data Sheet. • Microchip believes that its family of products is one of the most secure families of its kind on the market today, when used in the intended manner and under normal conditions. • There are dishonest and possibly illegal methods used to breach the code protection feature. All of these methods, to our knowledge, require using the Microchip products in a manner outside the operating specifications contained in Microchip’s Data Sheets. Most likely, the person doing so is engaged in theft of intellectual property. • Microchip is willing to work with the customer who is concerned about the integrity of their code. • Neither Microchip nor any other semiconductor manufacturer can guarantee the security of their code. Code protection does not mean that we are guaranteeing the product as “unbreakable.” Code protection is constantly evolving. We at Microchip are committed to continuously improving the code protection features of our products. Attempts to break Microchip’s code protection feature can be a violation of the Digital Millennium Copyright Act. If such acts allow unauthorized access to your software or other copyrighted work, you can have a right to sue for relief under that Act. Information contained in this publication regarding device applications and the like is provided only for your convenience and can be superseded by updates. It is your responsibility to ensure that your application meets with your specifications. MICROCHIP MAKES NO REPRESENTATIONS OR WARRANTIES OF ANY KIND WHETHER EXPRESS OR IMPLIED, WRITTEN OR ORAL, STATUTORY OR OTHERWISE, RELATED TO THE INFORMATION, INCLUDING BUT NOT LIMITED TO ITS CONDITION, QUALITY, PERFORMANCE, MERCHANTABILITY OR FITNESS FOR PURPOSE. Microchip disclaims all liability arising from this information and its use. Use of Microchip devices in life support and/or safety applications is entirely at the buyer’s risk, and the buyer agrees to defend, indemnify and hold harmless Microchip from any and all damages, claims, suits, or expenses resulting from such use. No licenses are conveyed, implicitly or otherwise, under any Microchip intellectual property rights. Trademarks The Microchip name and logo, the Microchip logo, dsPIC, FlashFlex, KEELOQ, KEELOQ logo, MPLAB, PIC, PICmicro, PICSTART, PIC32 logo, rfPIC, SST, SST Logo, SuperFlash and UNI/O are registered trademarks of Microchip Technology Incorporated in the U.S.A. and other countries. FilterLab, Hampshire, HI-TECH C, Linear Active Thermistor, MTP, SEEVAL and The Embedded Control Solutions Company are registered trademarks of Microchip Technology Incorporated in the U.S.A. Silicon Storage Technology is a registered trademark of Microchip Technology Inc. in other countries. Analog-for-the-Digital Age, Application Maestro, BodyCom, chipKIT, chipKIT logo, CodeGuard, dsPICDEM, dsPICDEM.net, dsPICworks, dsSPEAK, ECAN, ECONOMONITOR, FanSense, HI-TIDE, In-Circuit Serial Programming, ICSP, Mindi, MiWi, MPASM, MPF, MPLAB Certified logo, MPLIB, MPLINK, mTouch, Omniscient Code Generation, PICC, PICC-18, PICDEM, PICDEM.net, PICkit, PICtail, REAL ICE, rfLAB, Select Mode, SQI, Serial Quad I/O, Total Endurance, TSHARC, UniWinDriver, WiperLock, ZENA and Z-Scale are trademarks of Microchip Technology Incorporated in the U.S.A. and other countries. SQTP is a service mark of Microchip Technology Incorporated in the U.S.A. GestIC and ULPP are registered trademarks of Microchip Technology Germany II GmbH & Co. KG, a subsidiary of Microchip Technology Inc., in other countries. All other trademarks mentioned herein are property of their respective companies. © 2012-2013, Microchip Technology Incorporated, Printed in the U.S.A., All Rights Reserved. Printed on recycled paper. ISBN: 978-1-62077-723-7 QUALITY MANAGEMENT SYSTEM CERTIFIED BY DNV == ISO/TS 16949 == DS50002053D-page 2 Microchip received ISO/TS-16949:2009 certification for its worldwide headquarters, design and wafer fabrication facilities in Chandler and Tempe, Arizona; Gresham, Oregon and design centers in California and India. The Company’s quality system processes and procedures are for its PIC® MCUs and dsPIC® DSCs, KEELOQ® code hopping devices, Serial EEPROMs, microperipherals, nonvolatile memory and analog products. In addition, Microchip’s quality system for the design and manufacture of development systems is ISO 9001:2000 certified. 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Table of Contents Preface ........................................................................................................................... 7 Chapter 1. Compiler Overview 1.1 Introduction ................................................................................................... 13 1.2 Compiler Description and Documentation .................................................... 13 1.3 Device Description ....................................................................................... 14 Chapter 2. Common C Interface 2.1 Introduction ................................................................................................... 15 2.2 Background – The Desire for Portable Code ............................................... 15 2.3 Using the CCI ............................................................................................... 18 2.4 ANSI Standard Refinement .......................................................................... 19 2.5 ANSI Standard Extensions ........................................................................... 27 2.6 Compiler Features ........................................................................................ 41 Chapter 3. How To’s 3.1 Introduction ................................................................................................... 43 3.2 Installing and Activating the Compiler .......................................................... 43 3.3 Invoking the Compiler ................................................................................... 45 3.4 Writing Source Code .................................................................................... 48 3.5 Getting My Application to Do What I Want ................................................... 59 3.6 Understanding the Compilation Process ...................................................... 64 3.7 Fixing Code That Does Not Work ................................................................. 72 Chapter 4. XC8 Command-line Driver 4.1 Introduction ................................................................................................... 77 4.2 Invoking the Compiler ................................................................................... 78 4.3 The Compilation Sequence .......................................................................... 81 4.4 Runtime Files ............................................................................................... 87 4.5 Compiler Output ........................................................................................... 89 4.6 Compiler Messages ...................................................................................... 91 4.7 MPLAB XC8 Driver Options ......................................................................... 96 4.8 Option Descriptions ...................................................................................... 97 4.9 MPLAB X Option Equivalents ..................................................................... 127 Chapter 5. C Language Features 5.1 Introduction ................................................................................................. 135 5.2 ANSI C Standard Issues ............................................................................ 135 5.3 Device-Related Features ............................................................................ 137 5.4 Supported Data Types and Variables ........................................................ 147 5.5 Memory Allocation and Access .................................................................. 168 5.6 Operators and Statements ......................................................................... 186 2012-2013 Microchip Technology Inc. DS50002053D-page 3 MPLAB® XC8 C Compiler User’s Guide 5.7 Register Usage ........................................................................................... 188 5.8 Functions .................................................................................................... 189 5.9 Interrupts .................................................................................................... 198 5.10 Main, Runtime Startup and Reset ............................................................ 204 5.11 Library Routines ....................................................................................... 208 5.12 Mixing C and Assembly Code .................................................................. 210 5.13 Optimizations ............................................................................................ 219 5.14 Preprocessing .......................................................................................... 222 5.15 Linking Programs ..................................................................................... 234 Chapter 6. Macro Assembler 6.1 Introduction ................................................................................................. 255 6.2 Assembler Usage ....................................................................................... 255 6.3 Options ....................................................................................................... 256 6.4 MPLAB XC8 Assembly Language .............................................................. 260 6.5 Assembly-Level Optimizations ................................................................... 286 6.6 Assembly List Files ..................................................................................... 287 Chapter 7. Linker 7.1 Introduction ................................................................................................. 297 7.2 Operation .................................................................................................... 297 7.3 Relocation and Psects ................................................................................ 306 7.4 Map Files .................................................................................................... 307 Chapter 8. Utilities 8.1 Introduction ................................................................................................. 313 8.2 Librarian ..................................................................................................... 314 8.3 HEXMATE .................................................................................................. 317 Appendix A. Library Functions A.1 Introduction ................................................................................................ 327 Appendix B. Embedded Compiler Compatibility Mode B.1 Introduction ................................................................................................ 421 B.2 Compiling in Compatibility Mode ................................................................ 421 B.3 Syntax Compatibility .................................................................................. 422 B.4 Data Type .................................................................................................. 423 B.5 Operator ..................................................................................................... 423 B.6 Extended Keywords ................................................................................... 424 B.7 Intrinsic Functions ...................................................................................... 425 B.8 Pragmas ..................................................................................................... 426 Appendix C. Error and Warning Messages C.1 Introduction ................................................................................................ 427 Appendix D. Implementation-Defined Behavior D.1 Introduction ................................................................................................ 541 D.2 Translation (G.3.1) ..................................................................................... 541 D.3 Environment (G.3.2) .................................................................................. 541 D.4 Identifiers (G.3.3) ....................................................................................... 542 DS50002053D-page 4 2012-2013 Microchip Technology Inc. D.5 Characters (G.3.4) ..................................................................................... 542 D.6 Integers (G.3.5) .......................................................................................... 543 D.7 Floating-Point (G.3.6) ................................................................................ 544 D.8 Arrays and Pointers (G.3.7) ....................................................................... 544 D.9 Registers (G.3.8) ....................................................................................... 544 D.10 Structures, Unions, Enumerations, and Bit-Fields (G.3.9) ....................... 545 D.11 Qualifiers (G.3.10) ................................................................................... 545 D.12 Declarators (G.3.11) ................................................................................ 545 D.13 Statements (G.3.12) ................................................................................ 545 D.14 Preprocessing Directives (G.3.13) ........................................................... 546 D.15 Library Functions (G.3.14) ....................................................................... 547 Glossary ..................................................................................................................... 549 Index ........................................................................................................................... 569 Worldwide Sales and Service .................................................................................. 580 2012-2013 Microchip Technology Inc. DS50002053D-page 5 MPLAB® XC8 C Compiler User’s Guide NOTES: DS50002053D-page 6 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Preface NOTICE TO CUSTOMERS All documentation becomes dated, and this manual is no exception. Microchip tools and documentation are constantly evolving to meet customer needs, so some actual dialogs and/or tool descriptions can differ from those in this document. Please refer to our web site (www.microchip.com) to obtain the latest documentation available. Documents are identified with a “DS” number. This number is located on the bottom of each page, in front of the page number. The numbering convention for the DS number is “DSXXXXXA”, where “XXXXX” is the document number and “A” is the revision level of the document. For the most up-to-date information on development tools, see the MPLAB® IDE online help. Select the Help menu, and then Topics to open a list of available online help files. INTRODUCTION This chapter contains general information that will be useful to know before using the MPLAB® XC8 C Compiler User’s Guide. Items discussed in this chapter include: • • • • • • • • Document Layout Conventions Used in this Guide Recommended Reading Recommended Reading The Microchip Web Site Development Systems Customer Change Notification Service Customer Support Document Revision History DOCUMENT LAYOUT The MPLAB XC8 C Compiler User’s Guide is organized as follows: • • • • • • • • • • • • • • Chapter 1. Compiler Overview Chapter 2. Common C Interface Chapter 3. How To’s Chapter 4. XC8 Command-line Driver Chapter 5. C Language Features Chapter 6. Macro Assembler Chapter 7. Linker Chapter 8. Utilities Appendix A. Library Functions Appendix B. Embedded Compiler Compatibility Mode Appendix C. Error and Warning Messages Appendix D. Implementation-Defined Behavior Glossary Index 2012-2013 Microchip Technology Inc. DS50002053D-page 7 MPLAB® XC8 C Compiler User’s Guide CONVENTIONS USED IN THIS GUIDE This manual uses the following documentation conventions: DOCUMENTATION CONVENTIONS Description Arial font: Italic characters Initial caps Quotes Underlined, italic text with right angle bracket Bold characters N‘Rnnnn Text in angle brackets < > Courier New font: Plain Courier New Represents Examples Referenced books Emphasized text A window A dialog A menu selection A field name in a window or dialog A menu path MPLAB® IDE User’s Guide ...is the only compiler... the Output window the Settings dialog select Enable Programmer “Save project before build” A dialog button A tab A number in verilog format, where N is the total number of digits, R is the radix and n is a digit. A key on the keyboard Click OK Click the Power tab 4‘b0010, 2‘hF1 Italic Courier New Sample source code Filenames File paths Keywords Command-line options Bit values Constants A variable argument Square brackets [ ] Optional arguments Curly brackets and pipe character: { | } Ellipses... Choice of mutually exclusive arguments; an OR selection Replaces repeated text Represents code supplied by user DS50002053D-page 8 File>Save Press, #define START autoexec.bat c:\mcc18\h _asm, _endasm, static -Opa+, -Opa0, 1 0xFF, ‘A’ file.o, where file can be any valid filename mcc18 [options] file [options] errorlevel {0|1} var_name [, var_name...] void main (void) { ... } 2012-2013 Microchip Technology Inc. Preface RECOMMENDED READING This user’s guide describes how to use MPLAB XC8 C Compiler. Other useful documents are listed below. The following Microchip documents are available and recommended as supplemental reference resources. Readme for MPLAB XC8 C Compiler For the latest information on using MPLAB XC8 C Compiler, read MPLAB® XC8 C Compiler Release Notes (a PDF file) in the Docs subdirectory of the compiler’s installation directory. The release notes contain update information and known issues that cannot be included in this user’s guide. Readme Files For the latest information on using other tools, read the tool-specific Readme files in the Readmes subdirectory of the MPLAB IDE installation directory. The Readme files contain update information and known issues that cannot be included in this user’s guide. THE MICROCHIP WEB SITE Microchip provides online support via our web site at www.microchip.com. This web site is used as a means to make files and information easily available to customers. Accessible by using your favorite Internet browser, the web site contains the following information: • Product Support – Data sheets and errata, application notes and sample programs, design resources, user’s guides and hardware support documents, latest software releases and archived software • General Technical Support – Frequently Asked Questions (FAQs), technical support requests, online discussion groups, Microchip consultant program member listing • Business of Microchip – Product selector and ordering guides, latest Microchip press releases, listing of seminars and events, listings of Microchip sales offices, distributors and factory representatives 2012-2013 Microchip Technology Inc. DS50002053D-page 9 MPLAB® XC8 C Compiler User’s Guide DEVELOPMENT SYSTEMS CUSTOMER CHANGE NOTIFICATION SERVICE Microchip’s customer notification service helps keep customers current on Microchip products. Subscribers will receive e-mail notification whenever there are changes, updates, revisions or errata that are related to a specified product family or development tool of interest. To register, access the Microchip web site at www.microchip.com, click on Customer Change Notification and follow the registration instructions. The Development Systems product group categories are: • Compilers – The latest information on Microchip C compilers, assemblers, linkers and other language tools. These include all MPLAB C compilers; all MPLAB assemblers (including MPASM™ assembler); all MPLAB linkers (including MPLINK™ object linker); and all MPLAB librarians (including MPLIB™ object librarian). • Emulators – The latest information on Microchip in-circuit emulators.This includes the MPLAB REAL ICE™ and MPLAB ICE 2000 in-circuit emulators. • In-Circuit Debuggers – The latest information on the Microchip in-circuit debuggers. This includes MPLAB ICD 3 in-circuit debuggers and PICkit™ 3 debug express. • MPLAB® IDE – The latest information on Microchip MPLAB IDE, the Windows® Integrated Development Environment for development systems tools. This list is focused on the MPLAB IDE, MPLAB IDE Project Manager, MPLAB Editor and MPLAB SIM simulator, as well as general editing and debugging features. • Programmers – The latest information on Microchip programmers. These include production programmers such as MPLAB REAL ICE in-circuit emulator, MPLAB ICD 3 in-circuit debugger and MPLAB PM3 device programmers. Also included are nonproduction development programmers such as PICSTART® Plus and PICkit 2 and 3. CUSTOMER SUPPORT Users of Microchip products can receive assistance through several channels: • • • • Distributor or Representative Local Sales Office Field Application Engineer (FAE) Technical Support Customers should contact their distributor, representative or field application engineer (FAE) for support. Local sales offices are also available to help customers. A listing of sales offices and locations is included in the back of this document. Technical support is available through the web site at: http://www.microchip.com/support DS50002053D-page 10 2012-2013 Microchip Technology Inc. Preface DOCUMENT REVISION HISTORY Revision D (Dec 2013) • • • • • • • • • • Added new information relating to the software stack and function reentrancy. Added information relating to code profiling features offered by the compiler. Removed information pertaining to MPLAB IDE v8. Added new “How To’s” Removed sections on OBJTOHEX and CROMWELL. Added additional information relating to assembly code formats and operators. Corrected Fletcher algorithms used by HEXMATE. Added new driver options and updated existing option descriptions. Added and updated macros, built-ins and functions in Library Function chapter. Updated compiler warning and error messages. Revision C (May 2013) • • • • • • • • • • • • • Added Embedded Compiler Compatibility Mode chapter. Added information relating to new ELF/DWARF debugging files. Added new driver options and updated existing option descriptions. Updated MPLAB X IDE option dialog descriptions relating to compiler options. Expanded information relating to the available optimizations. Added code to illustrate checksum algorithms used by HEXMATE. Updated compiler warning and error messages. Updated information relating to list and map file contents. Added information about multiplication routines. Expanded information about eeprom variables and bit objects. Expanded information relating to the configuration pragma. Added information and examples using the __section() specifier. Expanded and extended information relating to assembly code deviations and assembler directives. Revision B (July 2012) • • • • • • • • • • • • • • Added How To’s chapter. Expanded section relating to PIC18 erratas. Updated the section relating to compiler optimization settings. Updated MPLAB v8 and MPLAB X IDE project option dialogs. Added sections describing PIC18 far qualifier and in-line function qualifier. Expanded section describing the operation of the main() function Expanded information about equivalent assembly symbols for Baseline parts. Updated the table of predefined macro symbols. Added section on #pragma addrqual Added sections to do with in-lining functions Updated diagrams and text associated with call graphs in the list file Updated library function section to be consistent with packaged libraries Added new compiler warnings and errors. Added new chapter describing the Common C Interface Standard (CCI) Revision A (February 2012) Initial release of this document. 2012-2013 Microchip Technology Inc. DS50002053D-page 11 MPLAB® XC8 C Compiler User’s Guide NOTES: DS50002053D-page 12 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Chapter 1. Compiler Overview 1.1 INTRODUCTION This chapter is an overview of the MPLAB® XC8 C Compiler, including these topics. • Compiler Description and Documentation • Device Description 1.2 COMPILER DESCRIPTION AND DOCUMENTATION The MPLAB XC8 C Compiler is a free-standing, optimizing ISO C90 (popularly known as ANSI C) compiler. It supports all 8-bit PIC® microcontrollers: PIC10, PIC12, PIC16 and PIC18 series devices, as well as the PIC14000 device. The compiler is available for several popular operating systems, including 32- and 64-bit Windows® (excluding Windows Server), Linux® and Mac OS® X. The compiler is available in three operating modes: Free, Standard or PRO. The Standard and PRO operating modes are licensed modes and require a serial number to enable them. Free mode is available for unlicensed customers. The basic compiler operation, supported devices and available memory are identical across all modes. The modes only differ in the level of optimization employed by the compiler. 1.2.1 Conventions Throughout this manual, the term “compiler” is used. It can refer to all, or a subset of, the collection of applications that comprise the MPLAB XC8 C Compiler. When it is not important to identify which application performed an action, it will be attributed to “the compiler”. Likewise, “compiler” is often used to refer to the command-line driver. Although specifically, the driver for the MPLAB XC8 C Compiler package is called xc8. The driver and its options are discussed in Section 4.7 “MPLAB XC8 Driver Options”. Accordingly, “compiler options” commonly relates to command-line driver options. In a similar fashion, “compilation” refers to all or a selection of steps involved in generating source code into an executable binary image. 2012-2013 Microchip Technology Inc. DS50002053D-page 13 MPLAB® XC8 C Compiler User’s Guide 1.3 DEVICE DESCRIPTION This compiler supports 8-bit Microchip PIC devices with baseline, mid-range, Enhanced mid-range, and PIC18 cores. The following descriptions indicate the distinctions within those device cores: The baseline core uses a 12-bit-wide instruction set and is available in PIC10, PIC12 and PIC16 part numbers. The enhanced baseline core also use a 12-bit instruction set, but this set includes additional instructions. Some of the enhanced baseline chips support interrupts and the additional instructions used by interrupts. These devices are available in PIC12 and PIC16 part numbers. The mid-range core uses a 14-bit-wide instruction set that includes more instructions than the baseline core. It has larger data memory banks and program memory pages, as well. It is available in PIC12, PIC14 and PIC16 part numbers. The Enhanced mid-range core also uses a 14-bit-wide instruction set but incorporates additional instructions and features. There are both PIC12 and PIC16 part numbers that are based on the Enhanced mid-range core. The PIC18 core instruction set is 16 bits wide and features additional instructions and an expanded register set. PIC18 core devices have part numbers that begin with PIC18. The compiler takes advantage of the target device’s instruction set, addressing modes, memory, and registers whenever possible. See Section 4.8.19 “--CHIPINFO: Display List of Supported Devices” for information on finding the full list of devices that are supported by the compiler. DS50002053D-page 14 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Chapter 2. Common C Interface 2.1 INTRODUCTION The Common C Interface (CCI) is available with all MPLAB XC C compilers and is designed to enhance code portability between these compilers. For example, CCI-conforming code would make it easier to port from a PIC18 MCU using the MPLAB XC8 C compiler to a PIC24 MCU using the MPLAB XC16 C compiler. The CCI assumes that your source code already conforms to the ANSI Standard. If you intend to use the CCI, it is your responsibility to write code that conforms. Legacy projects will need to be migrated to achieve conformance. A compiler option must also be set to ensure that the operation of the compiler is consistent with the interface when the project is built. The following topics are examined in this chapter of the MPLAB XC8 C Compiler User’s Guide: • • • • • 2.2 Background – The Desire for Portable Code Using the CCI ANSI Standard Refinement ANSI Standard Extensions Compiler Features BACKGROUND – THE DESIRE FOR PORTABLE CODE All programmers want to write portable source code. Portability means that the same source code can be compiled and run in a different execution environment than that for which it was written. Rarely can code be one hundred percent portable, but the more tolerant it is to change, the less time and effort it takes to have it running in a new environment. Embedded engineers typically think of code portability as being across target devices, but this is only part of the situation. The same code could be compiled for the same target but with a different compiler. Differences between those compilers might lead to the code failing at compile time or runtime, so this must be considered as well. You can only write code for one target device and only use one brand of compiler, but if there is no regulation of the compiler’s operation, simply updating your compiler version can change your code’s behavior. Code must be portable across targets, tools, and time to be truly flexible. Clearly, this portability cannot be achieved by the programmer alone, since the compiler vendors can base their products on different technologies, implement different features and code syntax, or improve the way their product works. Many a great compiler optimization has broken many an unsuspecting project. Standards for the C language have been developed to ensure that change is managed and code is more portable. The American National Standards Institute (ANSI) publishes standards for many disciplines, including programming languages. The ANSI C Standard is a universally adopted standard for the C programming language. 2012-2013 Microchip Technology Inc. DS50002053D-page 15 MPLAB® XC8 C Compiler User’s Guide 2.2.1 The ANSI Standard The ANSI C Standard has to reconcile two opposing goals: freedom for compilers vendors to target new devices and improve code generation, with the known functional operation of source code for programmers. If both goals can be met, source code can be made portable. The standard is implemented as a set of rules which detail not only the syntax that a conforming C program must follow, but the semantic rules by which that program will be interpreted. Thus, for a compiler to conform to the standard, it must ensure that a conforming C program functions as described by the standard. The standard describes implementation, the set of tools, and the runtime environment on which the code will run. If any of these change, e.g., you build for, and run on, a different target device, or if you update the version of the compiler you use to build, then you are using a different implementation. The standard uses the term behavior to mean the external appearance or action of the program. It has nothing to do with how a program is encoded. Since the standard is trying to achieve goals that could be construed as conflicting, some specifications appear somewhat vague. For example, the standard states that an int type must be able to hold at least a 16-bit value, but it does not go as far as saying what the size of an int actually is; and the action of right-shifting a signed integer can produce different results on different implementations; yet, these different results are still ANSI C compliant. If the standard is too strict, device architectures cannot allow the compiler to conform.1 But, if it is too weak, programmers would see wildly differing results within different compilers and architectures, and the standard would lose its effectiveness. The standard organizes source code whose behavior is not fully defined into groups that include the following behaviors: Implementation-defined behavior This is unspecified behavior in which each implementation documents how the choice is made. Unspecified behavior The standard provides two or more possibilities and imposes no further requirements on which possibility is chosen in any particular instance. Undefined behavior This is behavior for which the standard imposes no requirements. Code that strictly conforms to the standard does not produce output that is dependent on any unspecified, undefined, or implementation-defined behavior. The size of an int, which was used as an example earlier, falls into the category of behavior that is defined by implementation. That is to say, the size of an int is defined by which compiler is being used, how that compiler is being used, and the device that is being targeted. All the MPLAB XC compilers conform to the ANS X3.159-1989 Standard for programming languages (with the exception of the MPLAB XC8 compiler’s inability to allow recursion, as mentioned in the footnote). This is commonly called the C89 Standard. Some features from the later standard, C99, are also supported. 1. For example, the mid-range PIC® microcontrollers do not have a data stack. Because a compiler targeting this device cannot implement recursion, it (strictly speaking) cannot conform to the ANSI C Standard. This example illustrates a situation in which the standard is too strict for mid-range devices and tools. DS50002053D-page 16 2012-2013 Microchip Technology Inc. Common C Interface For freestanding implementations (or for what we typically call embedded applications), the standard allows non-standard extensions to the language, but obviously does not enforce how they are specified or how they work. When working so closely to the device hardware, a programmer needs a means of specifying device setup and interrupts, as well as utilizing the often complex world of small-device memory architectures. This cannot be offered by the standard in a consistent way. While the ANSI C Standard provides a mutual understanding for programmers and compiler vendors, programmers need to consider the implementation-defined behavior of their tools and the probability that they may need to use extensions to the C language that are non-standard. Both of these circumstances can have an impact on code portability. 2.2.2 The Common C Interface The Common C Interface (CCI) supplements the ANSI C Standard and makes it easier for programmers to achieve consistent outcomes on all Microchip devices when using any of the MPLAB XC C compilers. It delivers the following improvements, all designed with portability in mind. Refinement of the ANSI C Standard The CCI documents specific behavior for some code in which actions are implementation-defined behavior under the ANSI C Standard. For example, the result of right-shifting a signed integer is fully defined by the CCI. Note that many implementation-defined items that closely couple with device characteristics, such as the size of an int, are not defined by the CCI. Consistent syntax for non-standard extensions The CCI non-standard extensions are mostly implemented using keywords with a uniform syntax. They replace keywords, macros and attributes that are the native compiler implementation. The interpretation of the keyword can differ across each compiler, and any arguments to the keywords can be device specific. Coding guidelines The CCI can indicate advice on how code should be written so that it can be ported to other devices or compilers. While you may choose not to follow the advice, it will not conform to the CCI. 2012-2013 Microchip Technology Inc. DS50002053D-page 17 MPLAB® XC8 C Compiler User’s Guide 2.3 USING THE CCI The CCI allows enhanced portability by refining implementation-defined behavior and standardizing the syntax for extensions to the language. The CCI is something you choose to follow and put into effect, thus it is relevant for new projects, although you can choose to modify existing projects so they conform. For your project to conform to the CCI, you must do the following things. Enable the CCI Select the MPLAB IDE widget Use CCI Syntax in your project, or use the command-line option that is equivalent. Include in every module Some CCI features are only enabled if this header is seen by the compiler. Ensure ANSI compliance Code that does not conform to the ANSI C Standard does not confirm to the CCI. Observe refinements to ANSI by the CCI Some ANSI implementation-defined behavior is defined explicitly by the CCI. Use the CCI extensions to the language Use the CCI extensions rather than the native language extensions The next sections detail specific items associated with the CCI. These items are segregated into those that refine the standard, those that deal with the ANSI C Standard extensions, and other miscellaneous compiler options and usage. Guidelines are indicated with these items. If any implementation-defined behavior or any non-standard extension is not discussed in this document, then it is not part of the CCI. For example, GCC case ranges, label addresses, and 24-bit short long types are not part of the CCI. Programs which use these features do not conform to the CCI. The compiler may issue a warning or error to indicate a non-CCI feature has been used and the CCI is enabled. DS50002053D-page 18 2012-2013 Microchip Technology Inc. Common C Interface 2.4 ANSI STANDARD REFINEMENT The following topics describe how the CCI refines the implementation-defined behaviors outlined in the ANSI C Standard. 2.4.1 Source File Encoding Under the CCI, a source file must be written using characters from the 7-bit ASCII set. Lines can be terminated using a line feed ('\n') or carriage return ('\r') that is immediately followed by a line feed. Escaped characters can be used in character constants or string literals to represent extended characters that are not in the basic character set. 2.4.1.1 EXAMPLE The following shows a string constant being defined that uses escaped characters. const char myName[] = "Bj\370rk\n"; 2.4.1.2 DIFFERENCES All compilers have used this character set. 2.4.1.3 MIGRATION TO THE CCI No action required. 2.4.2 The Prototype for main The prototype for the main() function is: int main(void); 2.4.2.1 EXAMPLE The following shows an example of how main() might be defined: int main(void) { while(1) process(); } 2.4.2.2 DIFFERENCES The 8-bit compilers used a void return type for this function. 2.4.2.3 MIGRATION TO THE CCI Each program has one definition for the main() function. Confirm the return type for main() in all projects previously compiled for 8-bit targets. 2.4.3 Header File Specification Header file specifications that use directory separators do not conform to the CCI. 2.4.3.1 EXAMPLE The following example shows two conforming include directives. #include #include "global.h" 2012-2013 Microchip Technology Inc. DS50002053D-page 19 MPLAB® XC8 C Compiler User’s Guide 2.4.3.2 DIFFERENCES Header file specifications that use directory separators have been allowed in previous versions of all compilers. Compatibility problems arose when Windows-style separators “\” were used and the code was compiled under other host operating systems. Under the CCI, no directory separators should be used. 2.4.3.3 MIGRATION TO THE CCI Any #include directives that use directory separators in the header file specifications should be changed. Remove all but the header file name in the directive. Add the directory path to the compiler’s include search path or MPLAB IDE equivalent. This will force the compiler to search the directories specified with this option. For example, the following code: #include should be changed to: #include and the path to the inc directory added to the compiler’s header search path in your MPLAB IDE project properties, or on the command-line as follows: -Ilcd 2.4.4 Include Search Paths When you include a header file under the CCI, the file should be discoverable in the paths searched by the compiler that are detailed below. Header files specified in angle bracket delimiters < > should be discoverable in the search paths that are specified by -I options (or the equivalent MPLAB IDE option), or in the standard compiler include directories. The -I options are searched in the order in which they are specified. Header files specified in quote characters " " should be discoverable in the current working directory or in the same directories that are searched when the header files are specified in angle bracket delimiters (as above). In the case of an MPLAB X project, the current working directory is the directory in which the C source file is located. If unsuccessful, the search paths should be to the same directories searched when the header file is specified in angle bracket delimiters. Any other options to specify search paths for header files do not conform to the CCI. 2.4.4.1 EXAMPLE If including a header file, as in the following directive: #include "myGlobals.h" The header file should be locatable in the current working directory, or the paths specified by any -I options, or the standard compiler directories. A header file being located elsewhere does not conform to the CCI. 2.4.4.2 DIFFERENCES The compiler operation under the CCI is not changed. This is purely a coding guideline. 2.4.4.3 MIGRATION TO THE CCI Remove any option that specifies header file search paths other than the -I option (or the equivalent MPLAB IDE option), and use the -I option in place of this. Ensure the header file can be found in the directories specified in this section. DS50002053D-page 20 2012-2013 Microchip Technology Inc. Common C Interface 2.4.5 The Number of Significant Initial Characters in an Identifier At least the first 255 characters in an identifier (internal and external) are significant. This extends upon the requirement of the ANSI C Standard that states a lower number of significant characters are used to identify an object. 2.4.5.1 EXAMPLE The following example shows two poorly named variables, but names which are considered unique under the CCI. int stateOfPortBWhenTheOperatorHasSelectedAutomaticModeAndMotorIsRunningFast; int stateOfPortBWhenTheOperatorHasSelectedAutomaticModeAndMotorIsRunningSlow; 2.4.5.2 DIFFERENCES Former 8-bit compilers used 31 significant characters by default, but an option allowed this to be extended. The 16- and 32-bit compilers did not impose a limit on the number of significant characters. 2.4.5.3 MIGRATION TO THE CCI No action required. You can take advantage of the less restrictive naming scheme. 2.4.6 Sizes of Types The sizes of the basic C types, e.g., char, int and long, are not fully defined by the CCI. These types, by design, reflect the size of registers and other architectural features in the target device. They allow the device to efficiently access objects of this type. The ANSI C Standard does, however, indicate minimum requirements for these types, as specified in . If you need fixed-size types in your project, use the types defined in , e.g., uint8_t or int16_t. These types are consistently defined across all XC compilers, even outside of the CCI. Essentially, the C language offers a choice of two groups of types: those that offer sizes and formats that are tailored to the device you are using, or those that have a fixed size, regardless of the target. 2.4.6.1 EXAMPLE The following example shows the definition of a variable, native, whose size will allow efficient access on the target device; and a variable, fixed, whose size is clearly indicated and remains fixed, even though it may not allow efficient access on every device. int native; int16_t fixed; 2.4.6.2 DIFFERENCES This is consistent with previous types implemented by the compiler. 2.4.6.3 MIGRATION TO THE CCI If you require a C type that has a fixed size, regardless of the target device, use one of the types defined by . 2012-2013 Microchip Technology Inc. DS50002053D-page 21 MPLAB® XC8 C Compiler User’s Guide 2.4.7 Plain char Types The type of a plain char is unsigned char. It is generally recommended that all definitions for the char type explicitly state the signedness of the object. 2.4.7.1 EXAMPLE The following example char foobar; defines an unsigned char object called foobar. 2.4.7.2 DIFFERENCES The 8-bit compilers have always treated plain char as an unsigned type. The 16- and 32-bit compilers used signed char as the default plain char type. The -funsigned-char option on those compilers changed the default type to be unsigned char. 2.4.7.3 MIGRATION TO THE CCI Any definition of an object defined as a plain char and using the 16- or 32-bit compilers needs review. Any plain char that was intended to be a signed quantity should be replaced with an explicit definition, for example. signed char foobar; You can use the -funsigned-char option on MPLAB XC16 and XC32 to change the type of plain char, but since this option is not supported on MPLAB XC8, the code is not strictly conforming. 2.4.8 Signed Integer Representation The value of a signed integer is determined by taking the two’s complement of the integer. 2.4.8.1 EXAMPLE The following shows a variable, test, that is assigned the value -28 decimal. signed char test = 0xE4; 2.4.8.2 DIFFERENCES All compilers have represented signed integers in the way described in this section. 2.4.8.3 MIGRATION TO THE CCI No action required. 2.4.9 Integer Conversion When converting an integer type to a signed integer of insufficient size, the original value is truncated from the most-significant bit to accommodate the target size. 2.4.9.1 EXAMPLE The following shows an assignment of a value that is truncated. signed char destination; unsigned int source = 0x12FE; destination = source; Under the CCI, the value of destination after the alignment is -2 (i.e., the bit pattern 0xFE). DS50002053D-page 22 2012-2013 Microchip Technology Inc. Common C Interface 2.4.9.2 DIFFERENCES All compilers have performed integer conversion in an identical fashion to that described in this section. 2.4.9.3 MIGRATION TO THE CCI No action required. 2.4.10 Bitwise Operations on Signed Values Bitwise operations on signed values act on the two’s complement representation, including the sign bit. See also Section 2.4.11 “Right-shifting Signed Values”. 2.4.10.1 EXAMPLE The following shows an example of a negative quantity involved in a bitwise AND operation. signed char output, input = -13; output = input & 0x7E; Under the CCI, the value of output after the assignment is 0x72. 2.4.10.2 DIFFERENCES All compilers have performed bitwise operations in an identical fashion to that described in this section. 2.4.10.3 MIGRATION TO THE CCI No action required. 2.4.11 Right-shifting Signed Values Right-shifting a signed value will involve sign extension. This will preserve the sign of the original value. 2.4.11.1 EXAMPLE The following shows an example of a negative quantity involved in a bitwise AND operation. signed char input, output = -13; output = input >> 3; Under the CCI, the value of output after the assignment is -2 (i.e., the bit pattern 0xFE). 2.4.11.2 DIFFERENCES All compilers have performed right-shifting as described in this section. 2.4.11.3 MIGRATION TO THE CCI No action required. 2012-2013 Microchip Technology Inc. DS50002053D-page 23 MPLAB® XC8 C Compiler User’s Guide 2.4.12 Conversion of Union Member Accessed Using Member With Different Type If a union defines several members of different types and you use one member identifier to try to access the contents of another (whether any conversion is applied to the result) is implementation-defined behavior in the standard. In the CCI, no conversion is applied and the bytes of the union object are interpreted as an object of the type of the member being accessed, without regard for alignment or other possible invalid conditions. 2.4.12.1 EXAMPLE The following shows an example of a union defining several members. union { signed char code; unsigned int data; float offset; } foobar; Code that attempts to extract offset by reading data is not guaranteed to read the correct value. float result; result = foobbar.data; 2.4.12.2 DIFFERENCES All compilers have not converted union members accessed via other members. 2.4.12.3 MIGRATION TO THE CCI No action required. 2.4.13 Default Bit-field int Type The type of a bit-field specified as a plain int is identical to that of one defined using unsigned int. This is quite different from other objects where the types int, signed and signed int are synonymous. It is recommended that the signedness of the bit-field be explicitly stated in all bit-field definitions. 2.4.13.1 EXAMPLE The following shows an example of a structure tag containing bit-fields that are unsigned integers and with the size specified. struct int int int }; 2.4.13.2 OUTPUTS { direction :1; parity :3; value :4; DIFFERENCES The 8-bit compilers have previously issued a warning if type int was used for bit-fields, but would implement the bit-field with an unsigned int type. The 16- and 32-bit compilers have implemented bit-fields defined using int as having a signed int type, unless the option -funsigned-bitfields was specified. DS50002053D-page 24 2012-2013 Microchip Technology Inc. Common C Interface 2.4.13.3 MIGRATION TO THE CCI Any code that defines a bit-field with the plain int type should be reviewed. If the intention was for these to be signed quantities, then the type of these should be changed to signed int. For example, in the following example: struct WAYPT { int log int direction }; :3; :4; the bit-field type should be changed to signed int, as in: struct WAYPT { signed int log :3; signed int direction :4; }; 2.4.14 Bit-fields Straddling a Storage Unit Boundary The standard indicates that implementations can determine whether bit-fields cross a storage unit boundary. In the CCI, bit-fields do not straddle a storage unit boundary; a new storage unit is allocated to the structure, and padding bits fill the gap. Note that the size of a storage unit differs with each compiler, as this is based on the size of the base data type (e.g., int) from which the bit-field type is derived. On 8-bit compilers this unit is 8-bits in size; for 16-bit compilers, it is 16 bits; and for 32-bit compilers, it is 32 bits in size. 2.4.14.1 EXAMPLE The following shows a structure containing bit-fields being defined. struct { unsigned first : 6; unsigned second :6; } order; Under the CCI and using MPLAB XC8, the storage allocation unit is byte sized. The bit-field, second, is allocated a new storage unit since there are only 2 bits remaining in the first storage unit in which first is allocated. The size of this structure, order, is 2 bytes. 2.4.14.2 DIFFERENCES This allocation is identical with that used by all previous compilers. 2.4.14.3 MIGRATION TO THE CCI No action required. 2.4.15 The Allocation Order of Bit-fields The memory ordering of bit-fields into their storage unit is not specified by the ANSI C Standard. In the CCI, the first bit defined is the least significant bit of the storage unit in which it is allocated. 2012-2013 Microchip Technology Inc. DS50002053D-page 25 MPLAB® XC8 C Compiler User’s Guide 2.4.15.1 EXAMPLE The following shows a structure containing bit-fields being defined. struct { unsigned lo : 1; unsigned mid :6; unsigned hi : 1; } foo; The bit-field lo is assigned the least significant bit of the storage unit assigned to the structure foo. The bit-field mid is assigned the next 6 least significant bits; and hi, the most significant bit of that same storage unit byte. 2.4.15.2 DIFFERENCES This is identical with the previous operation of all compilers. 2.4.15.3 MIGRATION TO THE CCI No action required. 2.4.16 The NULL Macro The NULL macro is defined in ; however, its definition is implementation-defined behavior. Under the CCI, the definition of NULL is the expression (0). 2.4.16.1 EXAMPLE The following shows a pointer being assigned a null pointer constant via the NULL macro. int * ip = NULL; The value of NULL, (0), is implicitly converted to the destination type. 2.4.16.2 DIFFERENCES The 32-bit compilers previously assigned NULL the expression ((void *)0). 2.4.16.3 MIGRATION TO THE CCI No action required. 2.4.17 Floating-point Sizes Under the CCI, floating-point types must not be smaller than 32 bits in size. 2.4.17.1 EXAMPLE The following shows the definition for outY, which is at least 32-bit in size. float outY; 2.4.17.2 DIFFERENCES The 8-bit compilers have allowed the use of 24-bit float and double types. 2.4.17.3 MIGRATION TO THE CCI When using 8-bit compilers, the float and double type will automatically be made 32 bits in size once the CCI mode is enabled. Review any source code that may have assumed a float or double type and may have been 24 bits in size. No migration is required for other compilers. DS50002053D-page 26 2012-2013 Microchip Technology Inc. Common C Interface 2.5 ANSI STANDARD EXTENSIONS The following topics describe how the CCI provides device-specific extensions to the standard. 2.5.1 Generic Header File A single header file must be used to declare all compiler- and device-specific types and SFRs. You must include this file into every module to conform with the CCI. Some CCI definitions depend on this header being seen. 2.5.1.1 EXAMPLE The following shows this header file being included, thus allowing conformance with the CCI, as well as allowing access to SFRs. #include 2.5.1.2 DIFFERENCES Some 8-bit compilers used as the equivalent header. Previous versions of the 16- and 32-bit compilers used a variety of headers to do the same job. 2.5.1.3 MIGRATION TO THE CCI Change: #include previously used in 8-bit compiler code, or family-specific header files, e.g., from: #include #include #include #include #include "p30f6014.h" to: #include 2.5.2 Absolute Addressing Variables and functions can be placed at an absolute address by using the __at() construct. Stack-based (auto and parameter) variables cannot use the __at() specifier. 2.5.2.1 EXAMPLE The following shows two variables and a function being made absolute. int scanMode __at(0x200); const char keys[] __at(123) = { ’r’, ’s’, ’u’, ’d’}; int modify(int x) __at(0x1000) { return x * 2 + 3; } 2.5.2.2 DIFFERENCES The 8-bit compilers have used an @ symbol to specify an absolute address. The 16- and 32-bit compilers have used the address attribute to specify an object’s address. 2.5.2.3 MIGRATION TO THE CCI Avoid making objects and functions absolute if possible. 2012-2013 Microchip Technology Inc. DS50002053D-page 27 MPLAB® XC8 C Compiler User’s Guide In MPLAB XC8, change absolute object definitions, e.g., from: int scanMode @ 0x200; to: int scanMode __at(0x200); In MPLAB XC16 and XC32, change code, e.g., from: int scanMode __attribute__(address(0x200))); to: int scanMode __at(0x200); 2.5.2.4 CAVEATS If the __at() and __section() specifiers are both applied to an object when using MPLAB XC8, the __section() specifier is currently ignored. 2.5.3 Far Objects and Functions The __far qualifier can be used to indicate that variables or functions are located in ‘far memory’. Exactly what constitutes far memory is dependent on the target device, but it is typically memory that requires more complex code to access. Expressions involving far-qualified objects usually generate slower and larger code. Use the native keywords discussed in the Differences section to look up information on the semantics of this qualifier. Some devices may not have such memory implemented; in which case, use of this qualifier is ignored. Stack-based (auto and parameter) variables cannot use the __far specifier. 2.5.3.1 EXAMPLE The following shows a variable and function qualified using __far. __far int serialNo; __far int ext_getCond(int selector); 2.5.3.2 DIFFERENCES The 8-bit compilers have used the qualifier far to indicate this meaning. Functions could not be qualified as far. The 16-bit compilers have used the far attribute with both variables and functions. The 32-bit compilers have used the far attribute with functions, only. DS50002053D-page 28 2012-2013 Microchip Technology Inc. Common C Interface 2.5.3.3 MIGRATION TO THE CCI For 8-bit compilers, change any occurrence of the far qualifier, e.g., from: far char template[20]; to: __far, i.e., __far char template[20]; In the 16- and 32-bit compilers, change any occurrence of the far attribute, e.g., from: void bar(void) __attribute__ ((far)); int tblIdx __attribute__ ((far)); to: void __far bar(void); int __far tblIdx; 2.5.3.4 CAVEATS None. 2.5.4 Near Objects The __near qualifier can be used to indicate that variables or functions are located in ‘near memory’. Exactly what constitutes near memory is dependent on the target device, but it is typically memory that can be accessed with less complex code. Expressions involving near-qualified objects generally are faster and result in smaller code. Use the native keywords discussed in the Differences section to look up information on the semantics of this qualifier. Some devices may not have such memory implemented; in which case, use of this qualifier is ignored. Stack-based (auto and parameter) variables cannot use the __near specifier. 2.5.4.1 EXAMPLE The following shows a variable and function qualified using __near. __near int serialNo; __near int ext_getCond(int selector); 2.5.4.2 DIFFERENCES The 8-bit compilers have used the qualifier near to indicate this meaning. Functions could not be qualified as near. The 16-bit compilers have used the near attribute with both variables and functions. The 32-bit compilers have used the near attribute for functions, only. 2012-2013 Microchip Technology Inc. DS50002053D-page 29 MPLAB® XC8 C Compiler User’s Guide 2.5.4.3 MIGRATION TO THE CCI For 8-bit compilers, change any occurrence of the near qualifier, e.g., from: near char template[20]; to: __near, i.e., __near char template[20]; In 16- and 32-bit compilers, change any occurrence of the near attribute, e.g., from: void bar(void) __attribute__ ((near)); int tblIdx __attribute__ ((near)); to: void __near bar(void); int __near tblIdx; 2.5.4.4 CAVEATS None. 2.5.5 Persistent Objects The __persistent qualifier can be used to indicate that variables should not be cleared by the runtime startup code. Use the native keywords discussed in the Differences section to look up information on the semantics of this qualifier. 2.5.5.1 EXAMPLE The following shows a variable qualified using __persistent. __persistent int serialNo; 2.5.5.2 DIFFERENCES The 8-bit compilers have used the qualifier, persistent, to indicate this meaning. The 16- and 32-bit compilers have used the persistent attribute with variables to indicate they were not to be cleared. 2.5.5.3 MIGRATION TO THE CCI With 8-bit compilers, change any occurrence of the persistent qualifier, e.g., from: persistent char template[20]; to: __persistent, i.e., __persistent char template[20]; For the 16- and 32-bit compilers, change any occurrence of the persistent attribute, e.g., from: int tblIdx __attribute__ ((persistent)); to: int __persistent tblIdx; 2.5.5.4 CAVEATS None. DS50002053D-page 30 2012-2013 Microchip Technology Inc. Common C Interface 2.5.6 X and Y Data Objects The __xdata and __ydata qualifiers can be used to indicate that variables are located in special memory regions. Exactly what constitutes X and Y memory is dependent on the target device, but it is typically memory that can be accessed independently on separate buses. Such memory is often required for some DSP instructions. Use the native keywords discussed in the Differences section to look up information on the semantics of these qualifiers. Some devices may not have such memory implemented; in which case, use of these qualifiers is ignored. 2.5.6.1 EXAMPLE The following shows a variable qualified using __xdata, as well as another variable qualified with __ydata. __xdata char data[16]; __ydata char coeffs[4]; 2.5.6.2 DIFFERENCES The 16-bit compilers have used the xmemory and ymemory space attribute with variables. Equivalent specifiers have never been defined for any other compiler. 2.5.6.3 MIGRATION TO THE CCI For 16-bit compilers, change any occurrence of the space attributes xmemory or ymemory, e.g., from: char __attribute__((space(xmemory)))template[20]; to: __xdata, or __ydata, i.e., __xdata char template[20]; 2.5.6.4 CAVEATS None. 2.5.7 Banked Data Objects The __bank(num) qualifier can be used to indicate that variables are located in a particular data memory bank. The number, num, represents the bank number. Exactly what constitutes banked memory is dependent on the target device, but it is typically a subdivision of data memory to allow for assembly instructions with a limited address width field. Use the native keywords discussed in the Differences section to look up information on the semantics of these qualifiers. Some devices may not have banked data memory implemented; in which case, use of this qualifier is ignored. The number of data banks implemented will vary from one device to another. 2.5.7.1 EXAMPLE The following shows a variable qualified using __bank(). __bank(0) char start; __bank(5) char stop; 2012-2013 Microchip Technology Inc. DS50002053D-page 31 MPLAB® XC8 C Compiler User’s Guide 2.5.7.2 DIFFERENCES The 8-bit compilers have used the four qualifiers bank0, bank1, bank2 and bank3 to indicate the same, albeit more limited, memory placement. Equivalent specifiers have never been defined for any other compiler. 2.5.7.3 MIGRATION TO THE CCI For 8-bit compilers, change any occurrence of the bankx qualifiers, e.g., from: bank2 int logEntry; to: __bank(), i.e., __bank(2) int logEntry; 2.5.7.4 CAVEATS This feature is not yet implemented in MPLAB XC8. 2.5.8 Alignment of Objects The __align(alignment) specifier can be used to indicate that variables must be aligned on a memory address that is a multiple of the alignment specified. The alignment term must be a power of 2. Positive values request that the object’s start address be aligned; negative values imply the object’s end address be aligned. Use the native keywords discussed in the Differences section to look up information on the semantics of this specifier. 2.5.8.1 EXAMPLE The following shows variables qualified using __align() to ensure they end on an address that is a multiple of 8, and start on an address that is a multiple of 2, respectively. __align(-8) int spacer; __align(2) char coeffs[6]; 2.5.8.2 DIFFERENCES An alignment feature has never been implemented on 8-bit compilers. The 16- and 32-bit compilers used the aligned attribute with variables. 2.5.8.3 MIGRATION TO THE CCI For 16- and 32-bit compilers, change any occurrence of the aligned attribute, e.g., from: char __attribute__((aligned(4)))mode; to: __align, i.e., __align(4) char mode; 2.5.8.4 CAVEATS This feature is not yet implemented on MPLAB XC8. DS50002053D-page 32 2012-2013 Microchip Technology Inc. Common C Interface 2.5.9 EEPROM Objects The __eeprom qualifier can be used to indicate that variables should be positioned in EEPROM. Use the native keywords discussed in the Differences section to look up information on the semantics of this qualifier. Some devices may not implement EEPROM. Use of this qualifier for such devices generates a warning. Stack-based (auto and parameter) variables cannot use the __eeprom specifier. 2.5.9.1 EXAMPLE The following shows a variable qualified using __eeprom. __eeprom int serialNos[4]; 2.5.9.2 DIFFERENCES The 8-bit compilers have used the qualifier, eeprom, to indicate this meaning for some devices. The 16-bit compilers have used the space attribute to allocate variables to the memory space used for EEPROM. 2.5.9.3 MIGRATION TO THE CCI For 8-bit compilers, change any occurrence of the eeprom qualifier, e.g., from: eeprom char title[20]; to: __eeprom, i.e., __eeprom char title[20]; For 16-bit compilers, change any occurrence of the eedata space attribute, e.g., from: int mainSw __attribute__ ((space(eedata))); to: int __eeprom mainSw; 2.5.9.4 CAVEATS MPLAB XC8 does not implement the __eeprom qualifiers for any PIC18 devices; this qualifier works as expected for other 8-bit devices. 2012-2013 Microchip Technology Inc. DS50002053D-page 33 MPLAB® XC8 C Compiler User’s Guide 2.5.10 Interrupt Functions The __interrupt(type) specifier can be used to indicate that a function is to act as an interrupt service routine. The type is a comma-separated list of keywords that indicate information about the interrupt function. The current interrupt types are: Implement the default interrupt function low_priority The interrupt function corresponds to the low priority interrupt source (MPLAB XC8 – PIC18 only) high_priority The interrupt function corresponds to the high priority interrupt source (MPLAB XC8) save(symbol-list) Save on entry and restore on exit the listed symbols (XC16) irq(irqid) Specify the interrupt vector associated with this interrupt (XC16) altirq(altirqid) Specify the alternate interrupt vector associated with this interrupt (XC16) preprologue(asm) Specify assembly code to be executed before any compiler-generated interrupt code (XC16) shadow Allow the ISR to utilize the shadow registers for context switching (XC16) auto_psv The ISR will set the PSVPAG register and restore it on exit (XC16) no_auto_psv The ISR will not set the PSVPAG register (XC16) Use the native keywords discussed in the Differences section to look up information on the semantics of this specifier. Some devices may not implement interrupts. Use of this qualifier for such devices generates a warning. If the argument to the __interrupt specifier does not make sense for the target device, a warning or error is issued by the compiler. 2.5.10.1 EXAMPLE The following shows a function qualified using __interrupt. __interrupt(low_priority) void getData(void) { if (TMR0IE && TMR0IF) { TMR0IF=0; ++tick_count; } } DS50002053D-page 34 2012-2013 Microchip Technology Inc. Common C Interface 2.5.10.2 DIFFERENCES The 8-bit compilers have used the interrupt and low_priority qualifiers to indicate this meaning for some devices. Interrupt routines were, by default, high priority. The 16- and 32-bit compilers have used the interrupt attribute to define interrupt functions. 2.5.10.3 MIGRATION TO THE CCI For 8-bit compilers, change any occurrence of the interrupt qualifier, e.g., from: void interrupt myIsr(void) void interrupt low_priority myLoIsr(void) to the following, respectively: void __interrupt(high_priority) myIsr(void) void __interrupt(low_priority) myLoIsr(void) For 16-bit compilers, change any occurrence of the interrupt attribute, e.g., from: void __attribute__((interrupt,auto_psv,(irq(52)))) myIsr(void); to: void __interrupt(auto_psv,(irq(52)))) myIsr(void); For 32-bit compilers, the __interrupt() keyword takes two parameters, the vector number and the (optional) IPL value. Change code that uses the interrupt attribute, similar to these examples: void __attribute__((vector(0), interrupt(IPL7AUTO), nomips16)) myisr0_7A(void) {} void __attribute__((vector(1), interrupt(IPL6SRS), nomips16)) myisr1_6SRS(void) {} /* Determine IPL and context-saving mode at runtime */ void __attribute__((vector(2), interrupt(), nomips16)) myisr2_RUNTIME(void) {} to: void __interrupt(0,IPL7AUTO) myisr0_7A(void) {} void __interrupt(1,IPL6SRS) myisr1_6SRS(void) {} /* Determine IPL and context-saving mode at runtime */ void __interrupt(2) myisr2_RUNTIME(void) {} 2.5.10.4 CAVEATS None. 2012-2013 Microchip Technology Inc. DS50002053D-page 35 MPLAB® XC8 C Compiler User’s Guide 2.5.11 Packing Objects The __pack specifier can be used to indicate that structures should not use memory gaps to align structure members, or that individual structure members should not be aligned. Use the native keywords discussed in the Differences section to look up information on the semantics of this specifier. Some compilers cannot pad structures with alignment gaps for some devices, and use of this specifier for such devices is ignored. 2.5.11.1 EXAMPLE The following shows a structure qualified using __pack, as well as a structure where one member has been explicitly packed. __pack struct DATAPOINT { unsigned char type; int value; } x-point; struct LINETYPE { unsigned char type; __pack int start; long total; } line; 2.5.11.2 DIFFERENCES The __pack specifier is a new CCI specifier that is available with MPLAB XC8. This specifier has no apparent effect since the device memory is byte addressable for all data objects. The 16- and 32-bit compilers have used the packed attribute to indicate that a structure member was not aligned with a memory gap. 2.5.11.3 MIGRATION TO THE CCI No migration is required for MPLAB XC8. For 16- and 32-bit compilers, change any occurrence of the packed attribute, e.g., from: struct DOT { char a; int x[2] __attribute__ ((packed)); }; to: struct DOT { char a; __pack int x[2]; }; Alternatively, you can pack the entire structure, if required. 2.5.11.4 CAVEATS None. DS50002053D-page 36 2012-2013 Microchip Technology Inc. Common C Interface 2.5.12 Indicating Antiquated Objects The __deprecate specifier can be used to indicate that an object has limited longevity and should not be used in new designs. It is commonly used by the compiler vendor to indicate that compiler extensions or features can become obsolete, or that better features have been developed and should be used in preference. Use the native keywords discussed in the Differences section to look up information on the semantics of this specifier. 2.5.12.1 EXAMPLE The following shows a function that uses the __deprecate keyword. void __deprecate getValue(int mode) { //... } 2.5.12.2 DIFFERENCES No deprecate feature was implemented on 8-bit compilers. The 16- and 32-bit compilers have used the deprecated attribute (note the different spelling) to indicate that objects should be avoided, if possible. 2.5.12.3 MIGRATION TO THE CCI For 16- and 32-bit compilers, change any occurrence of the deprecated attribute, e.g., from: int __attribute__(deprecated) intMask; to: int __deprecate intMask; 2.5.12.4 CAVEATS None. 2.5.13 Assigning Objects to Sections The __section() specifier can be used to indicate that an object should be located in the named section (or psect, using the MPLAB XC8 terminology). This is typically used when the object has special and unique linking requirements that cannot be addressed by existing compiler features. Use the native keywords discussed in the Differences section to look up information on the semantics of this specifier. 2.5.13.1 EXAMPLE The following shows a variable which uses the __section keyword. int __section("comSec") commonFlag; 2012-2013 Microchip Technology Inc. DS50002053D-page 37 MPLAB® XC8 C Compiler User’s Guide 2.5.13.2 DIFFERENCES The 8-bit compilers have used the #pragma psect directive to redirect objects to a new section, or psect. The operation of the __section() specifier differs from this pragma in several ways, as described below. Unlike with the pragma, the new psect created with __section() does not inherit the flags of the psect in which the object would normally have been allocated. This means that the new psect can be linked in any memory area, including any data bank. The compiler also makes no assumptions about the location of the object in the new section. Objects redirected to new psects using the pragma must always be linked in the same memory area, albeit at any address in that area. The __section() specifier allows objects that are initialized to be placed in a different psect. Initialization of the object is still performed, even in the new psect. This requires the automatic allocation of an additional psect (its name is the same as the new psect, prefixed with the letter i), that will contain the initial values. The pragma cannot be used with objects that are initialized. Objects allocated a different psect with __section()are cleared by the runtime startup code, unlike objects that use the pragma. You must reserve memory, and locate via a linker option, for any new psect created with a __section() specifier in the current MPLAB XC8 compiler implementation. The 16- and 32-bit compilers have used the section attribute to indicate a different destination section name. The __section() specifier works in a similar way to the attribute. 2.5.13.3 MIGRATION TO THE CCI For MPLAB XC8, change any occurrence of the #pragma psect directive, such as: #pragma psect text%%u=myText int getMode(int target) { //... } to the __section() specifier, as in: int __section ("myText") getMode(int target) { //... } For 16- and 32-bit compilers, change any occurrence of the section attribute, e.g., from: int __attribute__((section("myVars"))) intMask; to: int __section("myVars") intMask; 2.5.13.4 CAVEATS None. 2.5.14 Specifying Configuration Bits The #pragma config directive can be used to program the Configuration bits for a device. The pragma has the form: #pragma config setting = state|value where setting is a configuration setting descriptor (e.g., WDT), state is a descriptive value (e.g., ON) and value is a numerical value. Use the native keywords discussed in the Differences section to look up information on the semantics of this directive. DS50002053D-page 38 2012-2013 Microchip Technology Inc. Common C Interface 2.5.14.1 EXAMPLE The following shows Configuration bits being specified using this pragma. #pragma config WDT=ON, WDTPS = 0x1A 2.5.14.2 DIFFERENCES The 8-bit compilers have used the __CONFIG() macro for some targets that did not already have support for the #pragma config. The 16-bit compilers have used a number of macros to specify the configuration settings. The 32-bit compilers supported the use of #pragma config. 2.5.14.3 MIGRATION TO THE CCI For the 8-bit compilers, change any occurrence of the __CONFIG() macro, e.g., __CONFIG(WDTEN & XT & DPROT) to the #pragma config directive, e.g., #pragma config WDTE=ON, FOSC=XT, CPD=ON No migration is required if the #pragma config was already used. For the 16-bit compilers, change any occurrence of the _FOSC() or _FBORPOR() macros attribute, e.g., from: _FOSC(CSW_FSCM_ON & EC_PLL16); to: #pragma config FCKSMEM = CSW_ON_FSCM_ON, FPR = ECIO_PLL16 No migration is required for 32-bit code. 2.5.14.4 CAVEATS None. 2.5.15 Manifest Macros The CCI defines the general form for macros that manifest the compiler and target device characteristics. These macros can be used to conditionally compile alternate source code based on the compiler or the target device. The macros and macro families are details in Table 2-1. TABLE 2-1: MANIFEST MACROS DEFINED BY THE CCI Name Meaning if defined Example __XC__ Compiled with an MPLAB XC compiler __CCI__ Compiler is CCI compliant and CCI enforcement is enabled __CCI__ __XC##__ The specific XC compiler used (## can be 8, 16 or 32) __XC8__ __DEVICEFAMILY__ __DEVICENAME__ 2012-2013 Microchip Technology Inc. The family of the selected target device The selected target device name __XC__ __dsPIC30F__ __18F452__ DS50002053D-page 39 MPLAB® XC8 C Compiler User’s Guide 2.5.15.1 EXAMPLE The following shows code that is conditionally compiled dependent on the device having EEPROM memory. #ifdef __XC16__ void __interrupt(__auto_psv__) myIsr(void) #else void __interrupt(low_priority) myIsr(void) #endif 2.5.15.2 DIFFERENCES Some of these CCI macros are new (for example __CCI__), and others have different names to previous symbols with identical meaning (e.g., __18F452 is now __18F452__). 2.5.15.3 MIGRATION TO THE CCI Any code that uses compiler-defined macros needs review. Old macros will continue to work as expected, but they are not compliant with the CCI. 2.5.15.4 CAVEATS None. 2.5.16 In-line Assembly The asm() statement can be used to insert assembly code in-line with C code. The argument is a C string literal that represents a single assembly instruction. Obviously, the instructions contained in the argument are device specific. Use the native keywords discussed in the Differences section to look up information on the semantics of this statement. 2.5.16.1 EXAMPLE The following shows a MOVLW instruction being inserted in-line. asm("MOVLW _foobar"); 2.5.16.2 DIFFERENCES The 8-bit compilers have used either the asm() or #asm ... #endasm constructs to insert in-line assembly code. This is the same syntax used by the 16- and 32-bit compilers. 2.5.16.3 MIGRATION TO THE CCI For 8-bit compilers, change any instance of #asm ... #endasm, so that each instruction in the #asm block is placed in its own asm() statement, e.g., from: #asm MOVLW 20 MOVWF _i CLRF Ii+1 #endasm to: asm("MOVLW20"); asm("MOVWF _i"); asm("CLRFIi+1"); No migration is required for the 16- or 32-bit compilers. DS50002053D-page 40 2012-2013 Microchip Technology Inc. Common C Interface 2.5.16.4 CAVEATS None. 2.6 COMPILER FEATURES The following item details the compiler options used to control the CCI. 2.6.1 Enabling the CCI It is assumed that you are using the MPLAB X IDE to build projects that use the CCI. The widget in the MPLAB X IDE Project Properties to enable CCI conformance is Use CCI Syntax in the Compiler category. If you are not using this IDE, then the command-line options are --EXT=cci for MPLAB XC8 or -mcci for MPLAB XC16 and XC32. 2.6.1.1 DIFFERENCES This option has never been implemented previously. 2.6.1.2 MIGRATION TO THE CCI Enable the option. 2.6.1.3 CAVEATS None. 2012-2013 Microchip Technology Inc. DS50002053D-page 41 MPLAB® XC8 C Compiler User’s Guide NOTES: DS50002053D-page 42 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Chapter 3. How To’s 3.1 INTRODUCTION This section contains help and references for situations that are frequently encountered when building projects for Microchip 8-bit devices. Click the links at the beginning of each section to assist in finding the topic relevant to your question. Some topics are indexed in multiple sections. Start here: • • • • • • 3.2 Installing and Activating the Compiler Invoking the Compiler Writing Source Code Getting My Application to Do What I Want Understanding the Compilation Process Fixing Code That Does Not Work INSTALLING AND ACTIVATING THE COMPILER This section details questions that might arise when installing or activating the compiler. • How Do I Install and Activate My Compiler? • How Can I Tell if the Compiler has Activated Successfully? • Can I Install More Than One Version of the Same Compiler? 3.2.1 How Do I Install and Activate My Compiler? Installation of the compiler and activation of the license are performed simultaneously by the XC compiler installer. The guide Installing and Licensing MPLAB XC C Compilers (DS52059) is available on www.microchip.com. It provides details on single-user and network licenses, as well as how to activate a compiler for evaluation purposes. 3.2.2 How Can I Tell if the Compiler has Activated Successfully? If you think the compiler cannot have installed correctly or is not working, it is best to verify its operation outside of MPLAB IDE to isolate possible problems. Try running the compiler from the command line to check for correct operation. You do not have to actually compile code. From your terminal or DOS-prompt, run the compiler driver xc8 (see Section 4.2 “Invoking the Compiler”) with the option --VER. This option instructs the compiler to print version information and exit. Under Windows, for example, type the following line (replacing the path information with a path that is relevant to your installation). "C:\Program Files\Microchip\xc8\v1.00\bin\xc8" --ver The compiler should run, print an informative banner and quit. The operating mode is printed by the compiler each time you build. Note that if it is not activated properly, the compiler will continue to operate, but only in the Free mode. If an error is displayed, or the compiler indicates Free mode, your activation was not successful. 2012-2013 Microchip Technology Inc. DS50002053D-page 43 MPLAB® XC8 C Compiler User’s Guide 3.2.3 Can I Install More Than One Version of the Same Compiler? Yes, the compilers and installation process has been designed to allow you to have more than one version of the same compiler installed, and you can easily move between the versions by changing options in MPLAB IDE; see Section 3.3.4 “How Can I Select Which Compiler I Want to Build With?”. Compilers should be installed into a directory whose name is related to the compiler version. This is reflected in the default directory specified by the installer. For example, the 1.00 and 1.10 MPLAB XC8 compilers would typically be placed in separate directories. C:\Program Files\Microchip\xc8\v1.00\ C:\Program Files\Microchip\xc8\v1.10\ DS50002053D-page 44 2012-2013 Microchip Technology Inc. How To’s 3.3 INVOKING THE COMPILER This section discusses how the compiler is run, on the command-line or from the MPLAB IDE. It includes information about how to get the compiler to do what you want it to do, in terms of options and the build process itself. • • • • • • • • • How Do I Compile From Within MPLAB X IDE? How Do I Compile on the Command-line? How Do I Compile Using a Make Utility? How Can I Select Which Compiler I Want to Build With? How Can I Change the Compiler's Operating Mode? How Do I Build Libraries? How Do I Know What Compiler Options Are Available and What They Do? How Do I Know What the Build Options in MPLAB X IDE Do? What is Different About an MPLAB X IDE Debug Build? See, also, the following linked information in other sections. • • • • What Do I Need to Do When Compiling to Use a Debugger? How Do I Use Library Files in My Project? How Do I Stop the Compiler From Using Certain Memory Locations? What Optimizations Are Employed by the Compiler? 3.3.1 How Do I Compile From Within MPLAB X IDE? MPLAB X IDE user’s guide and online help provide directions for setting up a project in the MPLAB X integrated development environment. If you have one or more MPLAB XC8 compilers installed, you select the compiler you wish to use in the Configuration category in the Project Properties dialog. The options for that compiler are then shown in the XC8 Compiler and XC8 Linker categories. Note that each of these compiler categories have several Option categories. 3.3.2 How Do I Compile on the Command-line? The compiler driver is called xc8 for all 8-bit PIC devices; e.g., in Windows, it is named xc8.exe. This application should be invoked for all aspects of compilation. It is located in the bin directory of the compiler distribution. Avoid running the individual compiler applications (such as the assembler or linker) explicitly. You can compile and link in the one command, even if your project is spread among multiple source files. The driver is introduced in Section 4.2 “Invoking the Compiler”. See Section 3.3.4 “How Can I Select Which Compiler I Want to Build With?”, to ensure you are running the correct driver if you have more than one installed. The command-line options to the driver are detailed in Section 4.7 “MPLAB XC8 Driver Options”. The files that can be passed to the driver are listed and described in Section 4.2.3 “Input File Types”. 3.3.3 How Do I Compile Using a Make Utility? When compiling using a make utility (such as make), the compilation is usually performed as a two-step process: first generating the intermediate files, then the final compilation and link step to produce one binary output. This is described in Section 4.3.3 “Multi-Step Compilation”. The MPLAB XC8 compiler uses a unique technology called OCG that uses an intermediate file format that is different than traditional compilers (including XC16 and XC32). The intermediate file format used by XC8 is a p-code file (.p1 extension), not an object file. Generating object files as an intermediate file for multi-step compilation defeats many of the advantages of this technology. 2012-2013 Microchip Technology Inc. DS50002053D-page 45 MPLAB® XC8 C Compiler User’s Guide 3.3.4 How Can I Select Which Compiler I Want to Build With? The compilation and installation process has been designed to allow you to have more than one compiler installed at the same time. You can create a project in MPLAB X IDE and then build this project with different compilers by simply changing a setting in the project properties. To select which compiler is actually used when building a project under MPLAB X IDE, go to the Project Properties dialog. Select the Configuration category in the Project Properties dialog (Conf: [default]). A list of MPLAB XC8 compilers is shown in the Compiler Toolchain, on the far right. Select the compiler that you require. Once selected, the controls for that compiler are then shown by selecting the MPLAB XC8 global options, MPLAB XC8 Compiler and MPLAB XC8 Linker categories. These reveal a pane of options on the right. Note that each category has several panes which can be selected from a pull-down menu that is near the top of the pane. 3.3.5 How Can I Change the Compiler's Operating Mode? The compiler’s operating mode (Free, Standard or PRO, see Section 1.2 “Compiler Description and Documentation”) can be specified as a command line option when building on the command line; see Section 4.8.37 “--MODE: Choose Compiler Operating Mode”. If you are building under MPLAB X IDE, there is a Project Properties selector in the XC8 Compiler category, under the Optimizations option selector; see Section 4.9.1 “Global Category”. You can only select modes that your license entitles you to use. The Free mode is always available; Standard or PRO can be selected if you have purchased a license for those modes. 3.3.6 How Do I Build Libraries? Note that XC8 uses a different code generation framework (OCG) that uses additional library files to those used by traditional compilers (including XC16 and XC32). See Section 4.3.1 “The Compiler Applications”, for general information on the library types available and how they fit into the compilation process. When you have functions and data that are commonly used in applications, you can either make all the C source and header files available so that other developers can copy these into their projects. Alternatively you can bundle these source files up into a library which, along with the accompanying header files, can be linked into a project. Libraries are more convenient because there are fewer files to deal with. Compiling code from a library can also be fractionally faster. However, libraries do need to be maintained. XC8 must use LPP libraries for library routines written in C; the old-style LIB libraries are used for library routines written in assembly source. It is recommended that even these libraries be rebuilt if your project is moving to a new compiler version. Using the compiler driver, libraries can be built by listing all of the files that are to be included into the library on the command line. None of these files should contain a main() function, nor settings for Configuration bits or any other such data. Use the --OUTPUT=lpp option; see Section 4.8.44 “--OUTPUT= type: Specify Output File Type”, to indicate that a library file is required. For example: XC8 --chip=16f877a --output=lpp lcd.c utils.c io.c creates a library file called lcd.lpp. You can specify another name using the -O option; see Section 4.8.9 “-O: Specify Output File”, or just rename the file. DS50002053D-page 46 2012-2013 Microchip Technology Inc. How To’s To build a library in MPLAB X IDE, create a regular project.1 Add your source files in the usual way. Add in the option --OUTPUT=lpp to the Additional Options field in the MPLAB XC8 Linker category. Click Build. The IDE will issue a warning about the HEX file being missing, but this can be ignored. The library output can be found in the dist/default/production folder of the project directory. 3.3.7 How Do I Know What Compiler Options Are Available and What They Do? A list of all compiler options can be obtained by using the --HELP option on the command line; see Section 4.8.33 “--HELP: Display Help”. If you give the --HELP option an argument, being an option name, it will give specific information on that option, for example --HELP=runtime. Alternatively, all options are all listed in Section 4.8 “Option Descriptions” in this user’s guide. If you are compiling in MPLAB X IDE, see Section 4.9 “MPLAB X Option Equivalents”. 3.3.8 How Do I Know What the Build Options in MPLAB X IDE Do? Each of the widgets and controls, in the MPLAB X IDE Project Properties, map directly to one command-line driver option or suboption, in most instances. Section 4.8 “Option Descriptions” in this user’s guide lists all command-line driver options and includes cross references, where appropriate, to corresponding sections that relate to accessing those options from the IDE. (see Section 4.9 “MPLAB X Option Equivalents”). 3.3.9 What is Different About an MPLAB X IDE Debug Build? In MPLAB X, there are distinct build buttons and menu items to build (production) a project and to debug a project. While there are many differences between the builds in the IDE – in the XC8 compiler, there is very little that is different between the two types of build. The main difference is the setting of a preprocessor macro called __DEBUG, which is assigned 1 when a performing a debug build. This macro is not defined for production builds. You can make code in your source conditional on this macro using #ifdef directives, etc., (see Section 5.14.2 “Preprocessor Directives”); so that you can have your program behave differently when you are still in a development cycle. Some compiler errors are easier to track down after performing a debug build. In MPLAB X IDE, memory is reserved for your debugger (if selected) only when you perform a debug build. See Section 3.5.3 “What Do I Need to Do When Compiling to Use a Debugger?” for more information. 1. At present, the IDE library projects are incompatible with MPLAB XC8. 2012-2013 Microchip Technology Inc. DS50002053D-page 47 MPLAB® XC8 C Compiler User’s Guide 3.4 WRITING SOURCE CODE This section presents issues that pertain to the source code you write. It has been subdivided into the sections listed below. • • • • • • • C Language Specifics Device-Specific Features Memory Allocation Variables Functions Interrupts Assembly Code 3.4.1 C Language Specifics This section discusses source code issues that directly relate to the C language itself, but are commonly asked. • • • • When Should I Cast Expressions? Can Implicit Type Conversions Change the Expected Results of My Expressions? How Do I Enter Non-English Characters Into My Program? How Can I Use a Variable Defined in Another Source File? 3.4.1.1 WHEN SHOULD I CAST EXPRESSIONS? Expressions can be explicitly case using the cast operator -- a type in round brackets, e.g., (int). In all cases, conversion of one type to another must be done with caution and only when absolutely necessary. Consider the example: unsigned long l; unsigned int i; i = l; Here, a long type is being assigned to an int type, and the assignment will truncate the value in l. The compiler will automatically perform a type conversion from the type of the expression on the right of the assignment operator (long) to the type of the lvalue on the left of the operator (int).This is called an implicit type conversion. The compiler typically produces a warning concerning the potential loss of data by the truncation. A cast to type int is not required and should not be used in the above example if a long to int conversion was intended. The compiler knows the types of both operands and performs the conversion accordingly. If you did use a cast, there is the potential for mistakes if the code is later changed. For example, if you had: i = (int)l; the code works the same way; but if, in future, the type of i is changed to a long, for example, then you must remember to adjust the cast, or remove it, otherwise the contents of l will continue to be truncated by the assignment, which cannot be correct. Most importantly, the warning issued by the compiler will not be produced if the cast is in place. DS50002053D-page 48 2012-2013 Microchip Technology Inc. How To’s Only use a cast in situations where the types used by the compiler are not the types that you require. For example, consider the result of a division assigned to a floating point variable: int i, j; float fl; fl = i/j; In this case, integer division is performed, then the rounded integer result is converted to a float format. So, if i contained 7 and j contained 2, the division yields 3 and this is implicitly converted to a float type (3.0) and then assigned to fl. If you wanted the division to be performed in a float format, then a cast is necessary: fl = (float)i/j; (Casting either i or j forces the compiler to encode a floating-point division.) The result assigned to fl now is 3.5. An explicit cast can suppress warnings that might otherwise have been produced. This can also be the source of many problems. The more warnings the compiler produces, the better chance you have of finding potential bugs in your code. 3.4.1.2 CAN IMPLICIT TYPE CONVERSIONS CHANGE THE EXPECTED RESULTS OF MY EXPRESSIONS? Yes! The compiler will always use integral promotion and there is no way to disable this; see Section 5.6.1 “Integral Promotion”. In addition, the types of operands to binary operators are usually changed so that they have a common type, as specified by the C Standard. Changing the type of an operand can change the value of the final expression, so it is very important that you understand the type C Standard conversion rules that apply when dealing with binary operators. You can manually change the type of an operand by casting; see Section 3.4.1.1 “When Should I Cast Expressions?”. 3.4.1.3 HOW DO I ENTER NON-ENGLISH CHARACTERS INTO MY PROGRAM? The ANSI standard and HI-TECH C do not support extended characters set in character and string literals in the source character set. See Section 5.4.6 “Constant Types and Formats”, to see how these characters can be entered using escape sequences. 3.4.1.4 HOW CAN I USE A VARIABLE DEFINED IN ANOTHER SOURCE FILE? Provided the variable defined in the other source file is not static (see Section 5.5.2.1.1 “Static Variables”) or auto (see Section 5.5.2.2 “Auto Variable Allocation and access”), then adding a declaration for that variable into the current file will allow you to access it. A declaration consists of the keyword extern in addition to the type and the name of the variable, as specified in its definition, e.g., extern int systemStatus; This is part of the C language. Your favorite C textbook will give you more information. The position of the declaration in the current file determines the scope of the variable. That is, if you place the declaration inside a function, it will limit the scope of the variable to that function. If you place it outside of a function, it allows access to the variable in all functions for the remainder of the current file. Often, declarations are placed in header files and then they are #included into the C source code; see Section 5.14.2 “Preprocessor Directives”. 2012-2013 Microchip Technology Inc. DS50002053D-page 49 MPLAB® XC8 C Compiler User’s Guide 3.4.2 Device-Specific Features This section discusses the code that needs to be written to set up or control a feature that is specific to Microchip PIC devices. • • • • • How Do I Set the Configuration Bits? How Do I Use the PIC Device’s ID Locations? How Do I Determine the Cause of Reset on Mid-range Parts? How Do I Access SFRs? How Do I Stop the Compiler From Using Certain Memory Locations? See, also, the following linked information in other sections. • What Do I Need to Do When Compiling to Use a Debugger? 3.4.2.1 HOW DO I SET THE CONFIGURATION BITS? These should be set in your code using either a macro or a pragma. Earlier versions of MPLAB IDE allowed you to set these bits in a dialog, but MPLAB X IDE requires that they be specified in your source code. See Section 5.3.5 “Configuration Bit Access”, for details about how these are set. 3.4.2.2 HOW DO I USE THE PIC DEVICE’S ID LOCATIONS? There is a supplied macro or pragma that allows these values to be programmed; see Section 5.3.7 “ID Locations”. 3.4.2.3 HOW DO I DETERMINE THE CAUSE OF RESET ON MID-RANGE PARTS? The TO and PD bits in the STATUS register allow you to determine the cause of a Reset. However, these bits are quickly overwritten by the runtime startup code that is executed before main is executed; see Section 5.10.1 “Runtime Startup Code”. You can have the STATUS register saved into a location that is later accessible from C code, so that the cause of Reset can be determined by the application after it is running again; see Section “If the software stack is being used by the program, the stack pointer (FSR1) is also initialized by the runtime startup code. See Section 5.5.2.2.2 “Software Stack Operation”.”. 3.4.2.4 HOW DO I ACCESS SFRS? The compiler ships with header files; see Section 5.3.3 “Device Header Files”, that define the variables that are mapped over the top of memory-mapped SFRs. Since these are C variables, they can be used like any other C variables and no new syntax is required to access these registers. Bits within SFRs can also be accessed. Individual bit-wide variables are defined that are mapped over the bits in the SFR. Bit-fields are also available in structures that map over the SFR as a whole. You can use either in your code; see Section 5.3.6 “Using SFRs From C Code”. The name assigned to the variable is usually the same as the name specified in the device data sheet. See Section 3.4.2.5 “How Do I Find The Names Used to Represent SFRs and Bits?”, if these names are not recognized. DS50002053D-page 50 2012-2013 Microchip Technology Inc. How To’s 3.4.2.5 HOW DO I FIND THE NAMES USED TO REPRESENT SFRS AND BITS? Special function registers and the bits within them are accessed via special variables that map over the register; see Section 3.4.2.4 “How Do I Access SFRs?”. However, the names of these variables sometimes differ from those indicated in the data sheet for the device you are using. If required, you can examine the header file to find the device-specific header file that is relevant for your device. This file will define the variables that allow access to these special variables. However, an easier way to find these variable names is to look in any of the preprocessed files left behind from a previous compilation. Provided the corresponding source file included , the preprocessed file will show the definitions for all the SFR variables and bits for your target device. If you are compiling under MPLAB X IDE, the preprocessed file(s) are left under the build/default/production directory of your project for regular builds, or under build/default/debug for debug builds. They are typically left in the source file directory if you are compiling on the command line. These files have a .pre extension. 3.4.3 Memory Allocation Here are questions relating to how your source code affects memory allocation. • • • • How Do I Position Variables at an Address I Nominate? How Do I Position Functions at an Address I Nominate? How Do I Place Variables in Program Memory? How Do I Stop the Compiler From Using Certain Memory Locations? See, also, the following linked information in other sections. • Why Are Some Objects Positioned Into Memory That I Reserved? 3.4.3.1 HOW DO I POSITION VARIABLES AT AN ADDRESS I NOMINATE? The easiest way to do this is to make the variable absolute by using the @ address construct, see Section 5.5.4 “Absolute Variables”. This means that the address you specify is used in preference to the variable’s symbol in generated code. Since you nominate the address, you have full control over where objects are positioned. But, you must also ensure that absolute variables do not overlap. Variables placed in the middle of banks can cause havoc with the allocation of other variables and lead to “Can’t find space” errors; see Section 3.7.6 “How Do I Fix a “Can’t find space...” Error?”. See also, Section 5.5.2.2.4 “Changing the Default Auto Variable Allocation”, for information on moving auto variables; Section 5.5.2.1.3 “Changing the Default Non-Auto Variable Allocation”, for moving non-auto variables; and Section 5.5.3.2 “Changing the Default Allocation”, for moving program-space variables. 3.4.3.2 HOW DO I POSITION FUNCTIONS AT AN ADDRESS I NOMINATE? The easiest way to do this is to make the functions absolute by using the @ address construct, see Section 5.8.4 “Changing the Default Function Allocation”. This means that the address you specify is used in preference to the function’s symbol in generated code. Since you nominate the address, you have full control over where functions are positioned, but must also ensure that absolute functions do not overlap. Functions placed in the middle of pages can cause havoc with the allocation of other functions and lead to "Can’t find space" errors, see Section 3.7.6 “How Do I Fix a “Can’t find space...” Error?”. 2012-2013 Microchip Technology Inc. DS50002053D-page 51 MPLAB® XC8 C Compiler User’s Guide 3.4.3.3 HOW DO I PLACE VARIABLES IN PROGRAM MEMORY? The const qualifier implies that the qualified variable is read-only. As a consequence of this, any variables (except for auto variables or function parameters) that are qualified const are placed in program memory, thus freeing valuable data RAM. See Section 5.5.3 “Variables in Program Space”, for more information. Variables that are qualified const can also be made absolute, so that they can be positioned at an address you nominate; see Section 5.5.4.2 “Absolute Objects in Program Memory”. 3.4.3.4 HOW DO I STOP THE COMPILER FROM USING CERTAIN MEMORY LOCATIONS? Memory can be reserved when you build. The --RAM and --ROM options allow you to adjust the ranges of data and program memory, respectively, when you build; see Section 4.8.49 “--RAM: Adjust RAM Ranges”, and Section 4.8.50 “--ROM: Adjust ROM Ranges”. By default, all the available on-chip memory is available for use. However, these options allow you to reserve parts of this memory. 3.4.4 Variables This sections examines questions that relate to the definition and usage of variables and types within a program. • Why Are My Floating-point Results Not Quite What I Am Expecting? • How Can I Access Individual Bits of a Variable? • How Long Can I Make My Variable and Macro Names? See, also, the following linked information in other sections. • • • • • • • • How Do I Share Data Between Interrupt and Main-line Code? How Do I Position Variables at an Address I Nominate? How Do I Place Variables in Program Memory? How Do I Place Variables in the PIC18 Device’s External Program Memory? How Can I Rotate a Variable? How Do I Utilize All the RAM Banks on My Device? How Do I Utilize the Linear Memory on Enhanced Mid-range PIC Devices? How Do I Find Out Where Variables and Functions Have Been Positioned? 3.4.4.1 WHY ARE MY FLOATING-POINT RESULTS NOT QUITE WHAT I AM EXPECTING? First, if you are watching floating-point variables in MPLAB X IDE, make sure that their type and size agree with the way in which they are defined. For 24-bit floating point variables (whether they have type float or double), ensure that in MPLAB X IDE the Display Column Value As popup menu to IEEE Float (24 bit). If the variable is a 32-bit floating point object, set the types to IEEE Float. The size of the floating point type can be adjusted for both float and double types; see Section 4.8.31 “--FLOAT: Select Kind of Float Types”, and Section 4.8.23 “--DOUBLE: Select Kind of Double Types”. Since floating-point variables only have a finite number of bits to represent the values they are assigned, they will hold an approximation of their assigned value; see Section 5.4.3 “Floating-Point Data Types”. A floating-point variable can only hold one of a set of discrete real number values. If you attempt to assign a value that is not in this set, it is rounded to the nearest value. The more bits used by the mantissa in the floating-point variable, the more values can be exactly represented in the set, and the average error due to the rounding is reduced. DS50002053D-page 52 2012-2013 Microchip Technology Inc. How To’s Whenever floating-point arithmetic is performed, rounding also occurs. This can also lead to results that do not appear to be correct. 3.4.4.2 HOW CAN I ACCESS INDIVIDUAL BITS OF A VARIABLE? There are several ways of doing this. The simplest and most portable way is to define an integer variable and use macros to read, set, or clear the bits within the variable using a mask value and logical operations, such as the following. #define #define #define testbit(var, bit) setbit(var, bit) clrbit(var, bit) ((var) & (1 <<(bit))) ((var) |= (1 << (bit))) ((var) &= ~(1 << (bit))) These, respectively, test to see if bit number, bit, in the integer, var, is set; set the corresponding bit in var; and clear the corresponding bit in var. Alternatively, a union of an integer variable and a structure with bit-fields (see Section 5.4.4.2 “Bit-Fields in Structures”) can be defined, e.g., union both { unsigned char byte; struct { unsigned b0:1, b1:1, b2:1, b3:1, b4:1, b5:1, b6:1, b7:1; } bitv; } var; This allows you to access byte as a whole (using var.byte), or any bit within that variable independently (using var.bitv.b0 through var.bitv.b7). Note that the compiler does support bit variables (see Section 5.4.2.1 “Bit Data Types and Variables”), as well as bit-fields in structures. 3.4.4.3 HOW LONG CAN I MAKE MY VARIABLE AND MACRO NAMES? The C Standard indicates that a only a specific number of initial characters in an identifier are significant, but it does not actually state what this number is and it varies from compiler to compiler. For XC8, the first 255 characters are significant, but this can be reduced using the -N option; see Section 4.8.8 “-N: Identifier Length”. The fewer characters there are in your variable names, the more portable your code. Using the -N option allows the compiler to check that your identifiers conform to a specific length. This option affects variable and function names, as well as preprocessor macro names. If two identifiers only differ in the non-significant part of the name, they are considered to represent the same object, which will almost certainly lead to code failure. 3.4.5 Functions This section examines questions that relate to functions. • What is the Optimum Size For Functions? • How Do I Stop An Unused Function Being Removed? • How Do I Make a Function Inline? See, also, the following linked information in other sections. • • • • • How Can I Tell How Big a Function Is? How Do I Position Functions at an Address I Nominate? How Do I Know What Resources Are Being Used by Each Function? How Do I Find Out Where Variables and Functions Have Been Positioned? How Do I Use Interrupts in C? 2012-2013 Microchip Technology Inc. DS50002053D-page 53 MPLAB® XC8 C Compiler User’s Guide 3.4.5.1 WHAT IS THE OPTIMUM SIZE FOR FUNCTIONS? Generally speaking, the source code for functions should be kept small, as this aids in readability and debugging. It is much easier to describe and debug the operation of a function that performs a small number of tasks. And, they typically have fewer side effects, which can be the source of coding errors. In the embedded programming world, a large number of small functions, and the calls necessary to execute them, can result in excessive memory and stack usage, so a compromise is often necessary. The PIC10/12/16 devices use pages in the program memory as the location where the function code is stored and executed. Although the compiler will allow, and can encode, functions whose size (the size of the assembly code they generate) exceeds that of a program memory page, functions of such a size should be avoided and split into smaller routines where possible. The assembly call and jump sequences to locations in other pages are much longer than those made to destinations in the same page. If a function is so large as to cross a page boundary, then loops (or other code constructs that require jumps within that function) can use the longer form of jump on each iteration; see Section 5.8.3 “Allocation of Executable Code”. PIC18 devices are less affected by internal memory paging and the instruction set allows for calls and jumps to any destination with no penalty. But you should still endeavor to keep functions as small as possible. With all devices, the smaller the function, the easier it is for the linker to allocate them to memory without errors. 3.4.5.2 HOW DO I STOP AN UNUSED FUNCTION BEING REMOVED? If a C function’s symbol is referenced in hand-written assembly code, the function will never be removed, even if it is not called or never had its address taken in C code. Create an assembly source file and add this file to your project. You only have to reference the symbol in this file; so, the file can contain the following GLOBAL _myFunc where myFunc is the C name of the function in question (note the leading underscore in the assembly name, see Section 5.12.3.1 “Equivalent Assembly Symbols”). This is sufficient to prevent the function removal optimization from being performed. 3.4.5.3 HOW DO I MAKE A FUNCTION INLINE? You can ask the compiler to inline a function by using the inline specifier (see Section 5.8.1.2 “Inline Specifier”) or #pragma inline. This is only a suggestion to the compiler and cannot always be obeyed. Do not confuse this specifier/pragma with the intrinsic pragma1 (see Section 5.14.4.4 “The #pragma Intrinsic Directive”), which is for functions that have no corresponding source code and which will be specifically expanded by the code generator during compilation. 1. This specifier was originally named in-line but was changed to avoid confusion. DS50002053D-page 54 2012-2013 Microchip Technology Inc. How To’s 3.4.6 Interrupts Interrupt and interrupt service routine questions are discussed in this section. • How Do I Use Interrupts in C? See, also, the following linked information in other sections. • How Can I Make My Interrupt Routine Faster? • How Do I Share Data Between Interrupt and Main-line Code? 3.4.6.1 HOW DO I USE INTERRUPTS IN C? First, be aware of what sort of interrupt hardware is available on your target device. Baseline PIC devices do not implement interrupts at all; mid-range devices utilize a single interrupt vector, and PIC18 devices implement two separate interrupt vector locations and use a simple priority scheme. In C source code, a function can be written to act as the interrupt service routine by using the interrupt qualifier; see Section 5.9.1 “Writing an Interrupt Service Routine”. Such functions save/restore program context before/after executing the function body code and a different return instruction is used; see Section 5.9.4 “Context Switching”. There must be no more than one interrupt function for each interrupt vector implemented on the target device. Aside from the interrupt qualifier, the function prototype must specify no parameters and a void return type. If you wish to implement the low priority interrupt function on PIC18 devices, use the low_priority keyword as well as the interrupt qualifier. Code inside the interrupt function can do anything you like, but see Section 3.6.7 “How Can I Make My Interrupt Routine Faster?” for suggestions to enhance real-time performance. Prior to any interrupt occurring, your program must ensure that peripherals are correctly configured and that interrupts are enabled; see Section 5.9.5 “Enabling Interrupts”. On PIC18 devices, you must specify the priority of interrupt sources by writing the appropriate SFRs. 2012-2013 Microchip Technology Inc. DS50002053D-page 55 MPLAB® XC8 C Compiler User’s Guide 3.4.7 Assembly Code This section examines questions that arise when writing assembly code as part of a C project. • • • • • How Should I Combine Assembly and C Code? What Do I Need Other than Instructions in an Assembly Source File? How Do I Access C Objects from Assembly Code? How Can I Access SFRs from Within Assembly Code? What Things Must I Manage When Writing Assembly Code? 3.4.7.1 HOW SHOULD I COMBINE ASSEMBLY AND C CODE? Ideally, any hand-written assembly should be written as separate routines that can be called. This offers some degree of protection from interaction between compiler-generated and hand-written assembly code. Such code can be placed into a separate assembly module that can be added to your project; see Section 5.12.1 “Integrating Assembly Language Modules”. If necessary, assembly code can be added in-line with C code using either of two methods; see Section 5.12.2 “#asm, #endasm and asm()”. The code added in-line should ideally be limited to instructions such as NOP, SLEEP or CLRWDT. Macros are already provided which in-line all these instructions; see Appendix A. “Library Functions”. More complex in-line assembly that changes register contents and the device state can cause code failure if precautions are not taken and should be used with caution. See Section 5.7 “Register Usage” for those registers used by the compiler. 3.4.7.2 WHAT DO I NEED OTHER THAN INSTRUCTIONS IN AN ASSEMBLY SOURCE FILE? Assembly code typically needs assembler directives as well as the instructions themselves. The operation of all the directives are described in the subsections of Section 6.4.9 “Assembler Directives”. Common directives required are mentioned below. All assembly code must be placed in a psect so it can be manipulated as a whole by the linker and placed in memory. See Section 5.15.1 “Program Sections” for general information on psects; see Section 6.4.9.3 “PSECT” for information on the directive used to create and specify psects. The other commonly used directive is GLOBAL, defined in Section 6.4.9.1 “GLOBAL” which is used to make symbols accessible across multiple source files. 3.4.7.3 HOW DO I ACCESS C OBJECTS FROM ASSEMBLY CODE? Most C objects are accessible from assembly code. There is a mapping between the symbols used in the C source and those used in the assembly code generated from this source. Your assembly should access the assembly-equivalent symbols which are detailed in Section 5.12.3 “Interaction between Assembly and C Code”. Instruct the assembler that the symbol is defined elsewhere by using the GLOBAL assembler directive; see Section 6.4.9.1 “GLOBAL”. This is the assembly equivalent of a C declaration, although no type information is present. This directive is not needed and should not be used if the symbol is defined in the same module as your assembly code. Any C variable accessed from assembly code will be treated as if it were qualified volatile; see Section 5.4.7.2 “Volatile Type Qualifier”. Specifically specifying the volatile qualifier in C code is preferred as it makes it clear that external code can access the object. DS50002053D-page 56 2012-2013 Microchip Technology Inc. How To’s 3.4.7.4 HOW CAN I ACCESS SFRS FROM WITHIN ASSEMBLY CODE? The safest way to gain access to SFRs in assembly code is to have symbols defined in your assembly code that equate to the corresponding SFR address. Header files are provided with the compiler so that you do not need to define these yourselves, and they are detailed in Section 5.12.3.2 “Accessing Registers from Assembly Code”. There is no guarantee that you will be able to access symbols generated by the compilation of C code, even code that accesses the SFR you require. 3.4.7.5 WHAT THINGS MUST I MANAGE WHEN WRITING ASSEMBLY CODE? If you are hand-writing assembly code there are several things that you must take control of. • Whenever you access a RAM variable, you must ensure that the bank of the variable is selected before you read or write the location. This is done by one or more assembly instructions. The exact code is based on the device you are using and the location of the variable. Bank selection is not be required if the object is in common memory, (which is called the access bank on PIC18 devices) or if you are using an instruction that takes a full address (such as the MOVFF instruction on PIC18 devices). Check your device data sheet to see the memory architecture of your device, and the instructions and registers which control bank selection. Failure to select the correct bank will lead to code failure. The BANKSEL pseudo instruction can be used to simplify this process; see Section 6.4.1.2 “Bank and Page Selection”. • You must ensure that the address of the RAM variable you are accessing has been masked so that only the bank offset is being used as the instruction’s file register operand. This should not be done if you are using an instruction that takes a full address (such as the MOVFF instruction on PIC18 devices). Check your device data sheet to see what address operand instructions requires. Failure to mask an address can lead to a fixup error (see Section 3.7.8 “How Do I Fix a Fixup Overflow Error?”) or code failure. The BANKMASK macro can truncate the address for you; see Section 5.12.3.2 “Accessing Registers from Assembly Code”. • Before you call or jump to any routine, you must ensure that you have selected the program memory page of this routine using the appropriate instructions. You can either use the PAGESEL pseudo instruction; see Section 6.4.1.2 “Bank and Page Selection”, or the FCALL or LJMP pseudo instructions (not required on PIC18 devices); see Section 6.4.1.7 “Long Jumps and Calls” which will automatically add page selection instructions, if required. • You must ensure that any RAM used for storage has memory reserved. If you are only accessing variables defined in C code, then reservation is already done by the compiler. You must reserve memory for any variables you only use in the assembly code using an appropriate directive such as DS or DABS; see Section 6.4.9.10 “DS” or Section 6.4.9.11 “DABS”. It is often easier to define objects in C code rather than in assembly. 2012-2013 Microchip Technology Inc. DS50002053D-page 57 MPLAB® XC8 C Compiler User’s Guide • You must place any assembly code you write in a psect (see Section 6.4.9.3 “PSECT”, for the directive to do this and Section 5.15.1 “Program Sections”, for general information about psects). A psect you define can need flags (options) to be specified. Pay particular note to the delta, space and class flags (see Section 6.4.9.3.4 “Delta”, and Section 6.4.9.3.15 “Space” and Section 6.4.9.3.3 “Class”). If these are not set correctly, compile errors or code failure will almost certainly result. If the psect specifies a class and you are happy with it being placed anywhere in the memory range defined by that class (see Section 7.2.1 “-Aclass =low-high,...”), it does not need any additional options to be linked; otherwise, you will need to link the psect using a linker option (see Section 7.2.19 “-Pspec” for the usual way to link psects and Section 4.8.6 “-L-: Adjust Linker Options Directly” which indicates how you can specify this option without running the linker directly). Assembly code that is placed in-line with C code will be placed in the same psect as the compiler-generated assembly and you should not place this into a separate psect. • You must ensure that any registers you write to in assembly code are not already in used by compiler-generated code. If you write assembly in a separate module, then this is less of an issue as the compiler will, by default, assume that all registers are used by these routines (see Section 5.7 “Register Usage”). No assumptions are made for in-line assembly (see Section 5.12.2 “#asm, #endasm and asm()”) and you must be careful to save and restore any resources that you use (write) and which are already in use by the surrounding compiler-generated code. DS50002053D-page 58 2012-2013 Microchip Technology Inc. How To’s 3.5 GETTING MY APPLICATION TO DO WHAT I WANT This section provides programming techniques, applications and examples. It also examines questions that relate to making an application perform a specific task. • • • • • • • • • • • • What Can Cause Glitches on Output Ports? How Do I Link Bootloaders and Downloadable Applications? What Do I Need to Do When Compiling to Use a Debugger? How Can I Have Code Executed Straight After Reset? How Do I Share Data Between Interrupt and Main-line Code? How Can I Prevent Misuse of My Code? How Do I Use Printf to Send Text to a Peripheral? How Do I Calibrate the Oscillator on My Device? How Do I Place Variables in the PIC18 Device’s External Program Memory? How Can I Implement a Delay in My Code? How Can I Rotate a Variable? How Can I Stop Variables Being Cleared at Startup? 3.5.1 What Can Cause Glitches on Output Ports? In most cases, this is caused by using ordinary variables to access port bits or the entire port itself. These variables should be qualified volatile. The value stored in a variable mapped over a port (hence the actual value written to the port) directly translates to an electrical signal. It is vital that the values held by these variables only change when the code intends them to, and that they change from their current state to their new value in a single transition. See Section 5.4.7.2 “Volatile Type Qualifier”. The compiler attempts to write to volatile variables in one operation. 3.5.2 How Do I Link Bootloaders and Downloadable Applications? Exactly how this is done depends on the device you are using and your project requirements, but the general approach when compiling applications that use a bootloader is to allocate discrete program memory space to the bootloader and application so they have their own dedicated memory. In this way the operation of one cannot affect the other. This will require that either the bootloader or the application is offset in memory. That is, the Reset and interrupt location are offset from address 0 and all program code is offset by the same amount. On PIC18 devices, typically the application code is offset, and the bootloader is linked with no offset so that it populates the Reset and interrupt code locations. The bootloader Reset and interrupt code merely contains code which redirects control to the real Reset and interrupt code defined by the application and which is offset. On mid-range devices, this is not normally possible to perform when interrupts are being used. Consider offsetting all of the bootloader with the exception of the code associated with Reset, which must always be defined by the bootloader. The application code can define the code linked at the interrupt location. The bootloader will need to remap any application code that attempts to overwrite the Reset code defined by the bootloader. The option --CODEOFFSET, (see Section 4.8.21 “--CODEOFFSET: Offset Program Code to Address”), allows the program code (Reset and vectors included) to be moved by a specified amount. The option also restricts the program from using any program memory from address 0 (Reset vector) to the offset address. Always check the map file; see Section 7.4.2 “Contents”, to ensure that nothing remains in reserved areas. 2012-2013 Microchip Technology Inc. DS50002053D-page 59 MPLAB® XC8 C Compiler User’s Guide The contents of the HEX file for the bootloader can be merged with the code of the application by adding the HEX file as a project file, either on the command line, or in MPLAB IDE. This results in a single HEX file that contains the bootloader and application code in the one image. HEX files are merged by the HEXMATE application; see Section 8.3 “HEXMATE”. Check for warnings from this application about overlap, which can indicate that memory is in use by both bootloader and the downloadable application. 3.5.3 What Do I Need to Do When Compiling to Use a Debugger? You can use debuggers, such as ICD3 or REALICE, to debug code built with the MPLAB XC8 compiler. These debuggers use some of the data and program memory of the device for its own use, so it is important that your code does not also use these resources. There is a command-line option; see Section 4.8.22 “--DEBUGGER: Select Debugger Type”, that can be used to tell the compiler which debugger is to be used. The compiler can then reserve the memory used by the debugger so that your code will not be located in these locations. In the MPLAB X IDE, the appropriate debugger option is specified if you perform a debug build. It will not be specified if you perform a regular Build Project or Clean and Build. Since some device memory is being used up by the debugger, there is less available for your program and it is possible that your code or data might not fit in the device when a debugger is selected. Note that which specific memory locations used by the debuggers is an attribute of MPLAB IDE, not the device. If you move a project to a new version of the IDE, the resources required can change. For this reason, you should not manually reserve memory for the debugger, or make any assumptions in your code as to what memory is used. A summary of the debugger requirements is available in the MPLAB IDE help files. To verify that the resources reserved by the compiler match those required by the debugger, do the following. Compile your code with and without the debugger selected and keep a copy of the map file produced for both builds. Compare the linker options in the map files and look for changes in the -A options; see Section 7.2.1 “-Aclass =low-high,...”. For example, the memory defined for the CODE class with no debugger might be specified by this option: -ACODE=00h-0FFh,0100h-07FFh,0800h-0FFFhx3 and with the ICD3 selected as the debugger, it becomes: -ACODE=00h-0FFh,0100h-07FFh,0800h-0FFFhx2,01800h-01EFFh This shows that a memory range from 1F00 to 1FFF has been removed by the compiler and cannot be used by your program. See also Section 3.6.16 “Why Are Some Objects Positioned Into Memory That I Reserved?”. 3.5.4 How Can I Have Code Executed Straight After Reset? A special hook has been provided so you can easily add special “powerup” assembly code which will be linked to the Reset vector; see Section 5.10.2 “The Powerup Routine”. This code will be executed before the runtime startup code is executed, which in turn is executed before the main function; see Section 5.10 “Main, Runtime Startup and Reset”. DS50002053D-page 60 2012-2013 Microchip Technology Inc. How To’s 3.5.5 How Do I Share Data Between Interrupt and Main-line Code? Variables accessed from both interrupt and main-line code can easily become corrupted or mis-read by the program. The volatile qualifier (see Section 5.4.7.2 “Volatile Type Qualifier”) tells the compiler to avoid performing optimizations on such variables. This will fix some of the issues associated with this problem. The other issues relates to whether the compiler/device can access the data atomically. With 8-bit PIC devices, this is rarely the case. An atomic access is one where the entire variable is accessed in only one instruction. Such access is uninterruptable. You can determine if a variable is being accessed atomically by looking at the assembly code the compiler produces in the assembly list file; see Section 6.6 “Assembly List Files”. If the variable is accessed in one instruction, it is atomic. Since the way variables are accessed can vary from statement to statement it is usually best to avoid these issues entirely by disabling interrupts prior to the variable being accessed in main-line code, then re-enable the interrupts afterwards; see Section 5.9.5 “Enabling Interrupts”. 3.5.6 How Can I Prevent Misuse of My Code? First, many devices with flash program memory allow all or part of this memory to be write protected. The device Configuration bits need to be set correctly for this to take place; see Section 5.3.5 “Configuration Bit Access” and your device data sheet. Second, you can prevent third-party code being programmed at unused locations in the program memory by filling these locations with a value rather than leaving them in their default unprogrammed state. You can chose a fill value that corresponds to an instruction or set all the bits so as the values cannot be further modified. (Consider what will happen if you program somehow reaches and starts executing from these filled values. What instruction will be executed?) The compiler’s HEXMATE utility (see Section 8.3 “HEXMATE”) has the capability to fill unused locations and this operation can be requested using a command-line driver option; see Section 4.8.30 “--FILL: Fill Unused Program Memory”. As HEXMATE only works with HEX files, this feature is only available when producing HEX/COF file outputs (as opposed to binary, for example), which is the default operation. And last, if you wish to make your library files or intermediate p-code files available to others but do not want the original source code to be viewable, then you can obfuscate the files using the --SHROUD option; see Section 4.8.55 “--SHROUD: Obfuscate P-code Files” 2012-2013 Microchip Technology Inc. DS50002053D-page 61 MPLAB® XC8 C Compiler User’s Guide 3.5.7 How Do I Use Printf to Send Text to a Peripheral? The printf function does two things: it formats text based on the format string and placeholders you specify, and sends (prints) this formatted text to a destination (or stream); see Appendix A. “Library Functions”. The printf function performs all the formatting; then it calls a helper function, called putch, to send each byte of the formatted text. By customizing the putch function you can have printf send data to any peripheral or location; see Section 5.11.1 “The printf Routine”. You can choose the printf output go to an LCD, SPI module or USART, for example. A stub for the putch function can be found in the compiler’s sources directory. Copy it into your project then modify it to send the single byte parameter passed to it to the required destination. Before you can use printf, peripherals that you use will need to be initialized in the usual way. Here is an example of putch for a USART on a mid-range device. void putch(char data) { while( ! TXIF) // check buffer continue; // wait till ready TXREG = data; // send data } You can get printf to send to one of several destinations by using a global variable to indicate your choice. Have the putch function send the byte to one of several destinations based on the contents of this variable. 3.5.8 How Do I Calibrate the Oscillator on My Device? Some devices allow for calibration of their internal oscillators; see your device data sheet. The runtime startup code generated by the compiler, (see Section 5.10.1 “Runtime Startup Code”), will by default provide code that performs oscillator calibration. This can be disabled, if required, using an option; see Section 4.8.51 “--RUNTIME: Specify Runtime Environment”. 3.5.9 How Do I Place Variables in the PIC18 Device’s External Program Memory? If all you mean to do is place read-only variables in program memory, qualify them as const; see Section 5.5.3 “Variables in Program Space”. If you intend the variables to be located in the external program memory then use the far qualifier and specify the memory using the --RAM option; see Section 4.8.49 “--RAM: Adjust RAM Ranges”. The compiler will allow far-qualified variables to be modified. Note that the time to access these variables will be longer than for variables in the internal data memory. The access mode to external memory can be specified with an option; see Section 4.8.25 “--EMI: Select External Memory Interface Operating Mode”. DS50002053D-page 62 2012-2013 Microchip Technology Inc. How To’s 3.5.10 How Can I Implement a Delay in My Code? If an accurate delay is required, or if there are other tasks that can be performed during the delay, then using a timer to generate an interrupt is the best way to proceed. If these are not issues in your code, then you can use the compiler’s in-built delay pseudo-functions: _delay, __delay_ms or __delay_us; see Appendix A. “Library Functions”. These all expand into in-line assembly instructions or a (nested) loop of instructions which will consume the specified number of cycles or time. The delay argument must be a constant and less than approximately 179,200 for PIC18 devices and approximately 50,659,000 for other devices. Note that these code sequences will only use the NOP instruction and/or instructions which form a loop. The alternate PIC18-only versions of these pseudo-functions, e.g., _delaywdt, can use the CLRWDT instruction as well. See also, Appendix A. “Library Functions”. 3.5.11 How Can I Rotate a Variable? The C language does not have a rotate operator, but rotations can be performed using the shift and bitwise OR operators. Since the PIC devices have a rotate instruction, the compiler will look for code expressions that implement rotates (using shifts and ORs) and use the rotate instruction in the generated output wherever possible; see Section 5.6.2 “Rotation”. 3.5.12 How Can I Stop Variables Being Cleared at Startup? Use the persistent qualifier (see Section 5.4.8.1 “Persistent Type Qualifier”), which will place the variables in a different psect that is not cleared by the runtime startup code. 2012-2013 Microchip Technology Inc. DS50002053D-page 63 MPLAB® XC8 C Compiler User’s Guide 3.6 UNDERSTANDING THE COMPILATION PROCESS This section tells you how to find out what the compiler did during the build process, how it encoded output code, where it placed objects, etc. It also discusses the features that are supported by the compiler. • • • • • • • • • • • • • • • • • • • • • • What’s the Difference Between the Free, Standard and PRO Modes? How Can I Make My Code Smaller? How Can I Reduce RAM Usage? How Can I Make My Code Faster? How Can I Speed Up Programming Times How Does the Compiler Place Everything in Memory? How Can I Make My Interrupt Routine Faster? How Big Can C Variables Be? How Do I Utilize All the RAM Banks on My Device? How Do I Utilize the Linear Memory on Enhanced Mid-range PIC Devices? What Devices are Supported by the Compiler? How Do I Know What Code the Compiler Is Producing? How Can I Tell How Big a Function Is? How Do I Know What Resources Are Being Used by Each Function? How Do I Find Out Where Variables and Functions Have Been Positioned? Why Are Some Objects Positioned Into Memory That I Reserved? How Do I Know How Much Memory Is Still Available? How Do I Use Library Files in My Project? What Optimizations Are Employed by the Compiler? Why Do I Get Out-of-memory Errors When I Select a Debugger? How Do I Know Which Stack Model the Compiler Has Assigned to a Function? How Do I Know What Value Has Been Programmed in the Configuration Bits or ID Location? See, also, the following linked information in other sections. • • • • How Do I Find Out What an Warning/error Message Means? What is Different About an MPLAB X IDE Debug Build? How Do I Stop An Unused Function Being Removed? How Do I Build Libraries? 3.6.1 What’s the Difference Between the Free, Standard and PRO Modes? These modes (see Section 1.2 “Compiler Description and Documentation”) mainly differ in the optimizations that are performed when compiling. Compilers operating in Free (formerly called Lite) and Standard mode can compile for all the same devices as supported by the Pro mode. The code compiled in Free and Standard mode can use all the available memory for the selected device. What will be different is the size and speed of the generated compiler output. Free mode output will be much less efficient when compared to that produced in Standard mode, which in turn will be less efficient than that produce when in Pro mode. All these modes use the OCG compiler framework, so the entire C program is compiled in one step and the source code does not need many non-standard extensions. DS50002053D-page 64 2012-2013 Microchip Technology Inc. How To’s There are a small number of command-line options disabled in Free mode, but these do not relate to code features; merely how the compiler can be executed. Most customers never need to use these options. The options are --GETOPTION (see Section 4.8.32 “--GETOPTION: Get Command-line Options”) and --SETOPTION (see Section 4.8.54 “--SETOPTION: Set the Command-line Options for Application”). 3.6.2 How Can I Make My Code Smaller? There are a number of ways that this can be done, but results vary from one project to the next. Use the assembly list file, (see Section 6.6 “Assembly List Files”), to observe the assembly code produced by the compiler to verify that the following tips are relevant to your code. Use the smallest data types possible as less code is needed to access these. (This also reduces RAM usage.) Note that a bit type and non-standard 24-bit integer type (short long) exists for this compiler. Avoid multi-bit bit-fields whenever possible. The code used to access these can be very large. See Section 5.4 “Supported Data Types and Variables”, for all data types and sizes. There are two sizes of floating-point type, as well, and these are discussed in the same section. Avoid floating-point if at all possible. Consider writing fixed-point arithmetic code. Use unsigned types, if possible, instead of signed types; particularly if they are used in expressions with a mix of types and sizes. Try to avoid an operator acting on operands with mixed sizes whenever possible. Whenever you have a loop or condition code, use a “strong” stop condition, i.e., the following: for(i=0; i!=10; i++) is preferable to: for(i=0; i<10; i++) A check for equality (== or !=) is usually more efficient to implement than the weaker < comparison. In some situations, using a loop counter that decrements to zero is more efficient than one that starts at zero and counts up by the same number of iterations. This is more likely to be the case if the loop index is a byte-wide type. So you might be able to rewrite the above as: for(i=10; i!=0; i--) There might be a small advantage in changing the order of function parameters so that the first parameter is byte sized. A register is used if the first parameter is byte-sized. For example consider: char calc(char mode, int value); over char calc(int value, char mode); Ensure that all optimizations are enabled; see Section 4.8.42 “--OPT: Invoke Compiler Optimizations”. Be aware of what optimizations the compiler performs (see Section 5.13 “Optimizations”, and Section 6.5 “Assembly-Level Optimizations”) so you can take advantage of them and don’t waste your time manually performing optimizations in C code that the compiler already handles, e.g., don’t turn a multiply-by-4 operation into a shift-by-2 operation as this sort of optimization is already detected. 2012-2013 Microchip Technology Inc. DS50002053D-page 65 MPLAB® XC8 C Compiler User’s Guide 3.6.3 How Can I Reduce RAM Usage? Use the smallest data types possible. (This also reduces code size as less code is needed to access these.) Note that a bit type and non-standard 24-bit integer type (short long) exists for this compiler. See Section 5.4 “Supported Data Types and Variables” for all data types and sizes. There are two sizes of floating-point type, as well, and these are discussed in the same section. Consider using auto variables over global or static variables as there is the potential that these can share memory allocated to other auto variables that are not active at the same time. Memory allocation of auto variables is made on a compiled stack, described in Section 5.5.2.2 “Auto Variable Allocation and access”. Rather than pass large objects to, or from, functions, pass pointers which reference these objects. This is particularly true when larger structures are being passed, but there might be RAM savings to be made even when passing long variables. Objects that do not need to change throughout the program can be located in program memory using the const qualifier; see Section 5.4.7.1 “Const Type Qualifier”, and Section 5.5.3 “Variables in Program Space”. This frees up precious RAM, but slows execution. Ensure that all optimizations are enabled; see Section 4.8.42 “--OPT: Invoke Compiler Optimizations”. Be aware of which optimizations the compiler performs (see Section 5.13 “Optimizations”), so that you can take advantage of them and don’t waste your time manually performing optimizations in C code that the compiler already handles. 3.6.4 How Can I Make My Code Faster? To a large degree, smaller code is faster code, so efforts to reduce code size often decrease execution time; see Section 3.6.2 “How Can I Make My Code Smaller?”. See also, Section 3.6.7 “How Can I Make My Interrupt Routine Faster?”. However, there are ways some sequences can be sped up at the expense of increased code size. One of the compiler optimization settings is for speed (the alternate setting is for space), so ensure this is selected; see Section 4.8.42 “--OPT: Invoke Compiler Optimizations”. This will use alternate output in some instances that is faster, but larger. Some library multiplication routines operate faster when one of their operands is a smaller value. See Section 5.3.9 “Hardware Multiply Instructions” for more information on how to take advantage of this. Generally, the biggest gains to be made in terms of speed of execution come from the algorithm used in a project. Identify which sections of your program need to be fast. Look for loops that might be linearly searching arrays and choose an alternate search method such as a hash table and function. Where results are being recalculated, consider if they can be cached. 3.6.5 How Can I Speed Up Programming Times The linker may build up some program output at low addresses in the HEX file, working their way to higher address; and some output at high address, working their way down in memory. This does not affect code operation and makes linking easier, but it can produce a HEX file covering the entire device memory space. Programming this HEX file into the device may take a long time. To reduce programming times in this situation, instruct the linker to not use all the device’s program memory. Use the --ROM option to reserve the upper part of program memory, see Section 4.8.50 “--ROM: Adjust ROM Ranges”. DS50002053D-page 66 2012-2013 Microchip Technology Inc. How To’s 3.6.6 How Does the Compiler Place Everything in Memory? In most situations, assembly instructions and directives associated with both code and data are grouped into sections, called psects, and these are then positioned into containers which represent the device memory. An introductory explanation into this process is given in Section 5.15.1 “Program Sections”. The exception is for absolute variables (see Section 5.5.4 “Absolute Variables”), which are placed at a specific address when they are defined and which are not placed in a psect. 3.6.7 How Can I Make My Interrupt Routine Faster? Consider suggestions made in Section 3.6.2 “How Can I Make My Code Smaller?” (code size) for any interrupt code. Smaller code is often faster code. In addition to the code you write in the ISR there is the code the compiler produces to switch context. This is executed immediately after an interrupt occurs and immediately before the interrupt returns, so must be included in the time taken to process an interrupt; see Section 5.9.4 “Context Switching”. This code is optimal in that only registers used in the ISR will be saved by this code. Thus, the less registers used in your ISR will mean potentially less context switch code to be executed. Mid-range devices have only a few registers that are used by the compiler, and there is little context switch code. Even fewer registers are considered for saving when compiling for enhanced mid-range device. PIC18 devices will benefit most from the above suggestion as they use a larger set of registers in generated code; see Section 5.7 “Register Usage”. Generally simpler code will require less resources than more complicated expressions. Use the assembly list file to see which registers are being used by the compiler in the interrupt code; see Section 6.6 “Assembly List Files”. Consider having the ISR simply set a flag and return. The flag can then be checked in main-line code to handle the interrupt. This has the advantage of moving the complicated interrupt-processing code out of the ISR so that it no longer contributes to its register usage. Always use the volatile qualifier (see Section 5.4.7.2 “Volatile Type Qualifier”for variables shared by the interrupt and main-line code; see Section 3.5.5 “How Do I Share Data Between Interrupt and Main-line Code?”. 3.6.8 How Big Can C Variables Be? This question specifically relates to the size of individual C objects, such as arrays or structures. The total size of all variables is another matter. To answer this question you need to know in which memory space the variable will be located. Objects qualified const will be located in program memory; other objects will be placed in data memory. Program memory object sizes are discussed in Section 5.5.3.1 “Size Limitations of Const Variables”. Objects in data memory are broadly grouped into autos and non-autos and the size limitations of these objects, respectively, are discussed in Section 5.5.2.2.2 “Software Stack Operation” and Section 5.5.2.1.2 “Non-Auto Variable Size Limits”. 3.6.9 How Do I Utilize All the RAM Banks on My Device? The compiler will automatically use all the available RAM banks on the device you are programming. It is only if you wish to alter the default memory allocation that you need take any action. Special bank qualifiers; see Section “--RAM=default,+20000-2FFFF.”, and an option (see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”) to indicate how these qualifiers are interpreted are used to manually allocate variables. Note that there is no guarantee that all the memory on a device can be utilized as data and code is packed in sections, or psects. 2012-2013 Microchip Technology Inc. DS50002053D-page 67 MPLAB® XC8 C Compiler User’s Guide 3.6.10 How Do I Utilize the Linear Memory on Enhanced Mid-range PIC Devices? The linear addressing mode is a means of accessing the banked data memory as one contiguous and linear block; see Section 5.5.1 “Address Spaces”. Use of the linear memory is fully automatic. Objects that are larger than a data bank can be defined in the usual way and will be accessed using the linear addressing mode; see Section 5.5.2.2.2 “Software Stack Operation”, and Section 5.5.2.1.2 “Non-Auto Variable Size Limits”. If you define absolute objects at a particular location in memory, you can use a linear address, if you prefer, or the regular banked address; see Section 5.5.4.1 “Absolute Variables in Data Memory”. 3.6.11 What Devices are Supported by the Compiler? Support for new devices usually takes place with each compiler release. To find whether a device is supported by your compiler, you can do several things; see also, Section 5.3.1 “Device Support”. • HTML listings are provided in the compiler’s docs directory. Open these in your favorite web browser. They are called pic_chipinfo.html and pic18_chipinfo.html. • Run the compiler driver on the command line (see Section 4.2 “Invoking the Compiler”) with the --CHIPINFO option; see Section 4.8.19 “--CHIPINFO: Display List of Supported Devices”. A full list of all devices is printed to the screen. 3.6.12 How Do I Know What Code the Compiler Is Producing? The assembly list file (see Section 6.6 “Assembly List Files”) shows the assembly output for almost the entire program, including library routines linked in to your program, as well a large amount of the runtime startup code; see Section 5.10.1 “Runtime Startup Code”. The list file is produced by default if you are using MPLAB IDE. If you are using the command-line, the option --ASMLIST will produce this file for you; see Section 4.8.16 “--ASMLIST: Generate Assembler List Files”. The assembly list file will have a .lst extension. The list file shows assembly instructions, some assembly directives and information about the program, such as the call graph (see Section 6.6.6 “Call Graph”), pointer reference graph (see Section 6.6.5 “Pointer Reference Graph”), and information for every function. Not all assembly directives are shown in the list file if the assembly optimizers are enabled (they are produced in the intermediate assembly file). Temporarily disable the assembly optimizers (see Section 4.8.42 “--OPT: Invoke Compiler Optimizations”), if you wish to see all the assembly directives produced by the compiler. 3.6.13 How Can I Tell How Big a Function Is? This size of a function (the amount of assembly code generated for that function) can be determined from the assembly list file; see Section 6.6 “Assembly List Files”, or a ‘funclist’ file generated by the compiler. Recent compilers define a symbol whose assigned value is equal to the size of the function. The symbol has the form __size_of_func, where func is the name of the function. The units of this symbol will be the same as the addressability of the program memory for the particular device: words for PIC10/12/16 and bytes for PIC18. You can also search for the labels that mark the beginning and end of the function. The function starts at the label _func:, where func is the name of the function, and ends just prior to the label __end_of_func. For example, the function main can have associated symbols __size_of_main, _main and __end_of_main. These will be found in the symbol table at the end of the assembly list file. DS50002053D-page 68 2012-2013 Microchip Technology Inc. How To’s The list of functions, memory location, and size is available in a file called funclist. Each function will have a line similar to the following. _main: CODE, 2012 0 30 This indicates that generated assembly code associated with the function, main, was placed in the CODE linker class (see Section 6.4.9.3.3 “Class”), was located at address 2012 (decimal), in address space number 0 (see Section 6.4.9.3.15 “Space”), and was 30 (again decimal) words/bytes long. An introduction to psects is given in Section 5.15.1 “Program Sections”. 3.6.14 How Do I Know What Resources Are Being Used by Each Function? In the assembly list file there is information printed for every C function, including library functions; see Section 6.6 “Assembly List Files”. This information indicates what registers the function used, what functions it calls (this is also found in the call graph; see Section 6.6.6 “Call Graph”), and how many bytes of data memory it requires. Note that auto, parameter and temporary variables used by a function can overlap with those from other functions as these are placed in a compiled stack by the compiler; see Section 5.5.2.2.1 “Compiled Stack Operation”. 3.6.15 How Do I Find Out Where Variables and Functions Have Been Positioned? You can determine where variables and functions have been positioned from either the assembly list file, see Section 6.6 “Assembly List Files”; or the map file, see Section 7.4 “Map Files”. Only global symbols are shown in the map file; all symbols (including locals) are listed in the assembly list file, but only for the code represented by that list file. (Each assembly module has its own list file.) There is a mapping between C identifiers and the symbols used in assembly code, which are the symbols shown in both of these files; see Section 5.12.3.1 “Equivalent Assembly Symbols”. The symbol associated with a variable is assigned the address of the lowest byte of the variable; for functions it is the address of the first instruction generated for that function. 3.6.16 Why Are Some Objects Positioned Into Memory That I Reserved? The memory reservation options (see Section 3.4.3.4 “How Do I Stop the Compiler From Using Certain Memory Locations?”) will adjust the range of addresses associated with classes used by the linker. Most variables and function are placed into psects (see Section 5.15.1 “Program Sections”) that are linked anywhere inside these class ranges and so are affected by these reservation options. Some psects are explicitly placed at an address rather than being linked anywhere in an address range, e.g., the psect that holds the code to be executed at Reset is always linked to address 0 because that is where the Reset location is defined to be for 8-bit devices. Such a psect will not be affected by the --ROM option, even if you use it to reserve memory address 0. Psects that hold code associated with Reset and interrupts can be shifted using the --CODEOFFSET option; see Section 4.8.20 “--CLIST: Generate C Listing File”. Check the assembly list file (see Section 6.6 “Assembly List Files”) to determine the names of psects that hold objects and code. Check the linker options in the map file; see Section 7.4 “Map Files”, to see if psects have been linked explicitly or if they are linked anywhere in a class. See also, the linker options -p (Section 7.2.19 “-Pspec”) and -A (Section 7.2.1 “-Aclass =low-high,...”). 2012-2013 Microchip Technology Inc. DS50002053D-page 69 MPLAB® XC8 C Compiler User’s Guide 3.6.17 How Do I Know How Much Memory Is Still Available? Although the memory summary printed by the compiler after compilation, (see Section 4.8.58 “--SUMMARY: Select Memory Summary Output Type” options), or the memory gauge available in MPLAB IDE both indicate the amount of memory used and the amount still available, neither of these features indicate whether this memory is one contiguous block or broken into many small chunks. Small blocks of free memory cannot be used for larger objects and so out-of-memory errors can be produced even though the total amount of memory free is apparently sufficient for the objects to be positioned. (See Section 3.7.6 “How Do I Fix a “Can’t find space...” Error?”.) The “UNUSED ADDRESS RANGES” section (see Section 7.4.2.5 “Unused Address Ranges”) in the map file indicates exactly what memory is still available in each linker class. It also indicated the largest contiguous block in that class if there are memory bank or page divisions. 3.6.18 How Do I Use Library Files in My Project? See Section 3.3.6 “How Do I Build Libraries?” for information on how you build your own library files. The compiler will automatically include any applicable standard library into the build process when you compile, so you never need to control these files. To use one or more library files that were built by yourself or a colleague, include them in the list of files being compiled on the command line. The library files can be specified in any position in the file list relative to the source files, but if there is more than one library file, they will be searched in the order specified in the command line. The LPP libraries do not need to be specified if you are compiling to an intermediate file, i.e., using the --PASS1 option (see Section 4.8.45 “--PARSER: Specify Parser Mode”). For example: xc8 --chip=16f1937 main.c int.c lcd.lpp If you are using MPLAB X IDE to build a project, add the library file(s) to the Libraries folder that will shown in your project, in the order in which they should be searched. The IDE will ensure that they are passed to the compiler at the appropriate point in the build sequence. 3.6.19 What Optimizations Are Employed by the Compiler? Optimizations are employed at both the C and assembly level of compilation. This is described in Section 5.13 “Optimizations” and Section 6.5 “Assembly-Level Optimizations”, respectively. The options that control optimization are described in Section 4.8.42 “--OPT: Invoke Compiler Optimizations”. 3.6.20 Why Do I Get Out-of-memory Errors When I Select a Debugger? If you use a hardware tool debugger, such as the REAL ICE or ICD3, these require memory for the on-board debug executive. When you select a debugger using the compiler’s --DEBUGGER option (Section 4.8.22 “--DEBUGGER: Select Debugger Type”), or the IDE equivalent, the memory required for debugging is removed from that available to your project. See Section 3.5.3 “What Do I Need to Do When Compiling to Use a Debugger?” DS50002053D-page 70 2012-2013 Microchip Technology Inc. How To’s 3.6.21 How Do I Know Which Stack Model the Compiler Has Assigned to a Function? Look in the function information section in the assembly list file, see Section 6.6.3 “Function Information”. The last line of this block will indicate whether the function uses a reentrant or non-reentrant model. 3.6.22 How Do I Know What Value Has Been Programmed in the Configuration Bits or ID Location? Check the file startup.as. This contains the output of the #pragma config directive. You will see the numerical value programmed to the appropriate locations. In the following example, the configuration value programmed is 0xFFBF. A breakdown of what this value means is also printed. ; Config register CONFIG @ 0x2007 ; BOREN = OFF, BOR disabled ; ... ; PWRTE = 0x1, unprogrammed default psect 2012-2013 Microchip Technology Inc. config org 0x0 dw 0xFFBF DS50002053D-page 71 MPLAB® XC8 C Compiler User’s Guide 3.7 FIXING CODE THAT DOES NOT WORK This section examines issues relating to projects that do not build due to compiler errors, or those that build, but do not work as expected. • How Do I Find Out What an Warning/error Message Means? • How Do I Find the Code that Caused Compiler Errors or Warnings in My Program? • How Can I Stop Spurious Warnings From Being Produced? • Why Can’t I Even Blink an LED? • How Do I Know If the Hardware Stack Has Overflowed? • How Do I Fix a “Can’t find space...” Error? • How Do I Fix a “Can’t generate code...” Error? • How Do I Fix a Fixup Overflow Error? • What Can Cause Corrupted Variables and Code Failure When Using Interrupts? 3.7.1 How Do I Find Out What an Warning/error Message Means? Each warning or error message has a description, and possibly sample code that might trigger such an error, listed in the messages chapter, see Appendix C. “Error and Warning Messages”. The compiler prints with each message a unique ID number in brackets. Use this number to look up the message in the manual. This number also allows you to control message behavior using options and pragmas, see Section 4.6.5 “Changing Message Behavior”. 3.7.2 How Do I Find the Code that Caused Compiler Errors or Warnings in My Program? In most instances, where the error is a syntax error relating to the source code, the message produced by the compiler indicates the offending line of code, see Section 4.6 “Compiler Messages”. If you are compiling in MPLAB IDE, then you can double-click the message and have the editor take you to the offending line. But identifying the offending code is not always so easy. In some instances, the error is reported on the line of code following the line that needs attention. This is because a C statement is allowed to extend over multiple lines of the source file. It is possible that the compiler cannot be able to determine that there is an error until it has started to scan to statement following. So in the following code input = PORTB if(input>6) // ... // oops - forgot the semicolon The missing semicolon on the assignment statement will be flagged on the following line that contains the if() statement. In other cases, the error might come from the assembler, not the code generator. If the assembly code was derived from a C source file then the compiler will try to indicate the line in the C source file that corresponds to the assembly that is at fault. If the source being compiled is an assembly module, the error directly indicates the line of assembly that triggered the error. In either case, remember that the information in the error relates to some problem is the assembly code, not the C code. Finally, there are errors that do not relate to any particular line of code at all. An error in a compiler option or a linker error are examples of these. If the program defines too many variables, there is no one particular line of code that is at fault; the program as a whole uses too much data. Note that the name and line number of the last processed file and source can be printed in some situations even though that code is not the direct source of the error. DS50002053D-page 72 2012-2013 Microchip Technology Inc. How To’s To determine the application that generated the error or warning, take a note of its unique number printed in the message, see Section 4.6.1 “Messaging Overview”, and check the message section of the manual, see Appendix C. “Error and Warning Messages”. At the top of each message description, on the right in brackets, is the name of the application that produced this message. Knowing the application that produced the error makes it easier to track down the problem. The compiler application names are indicated in Section 4.3 “The Compilation Sequence”. If you need to see the assembly code generated by the compiler, look in the assembly list file, see Section 6.6 “Assembly List Files”. For information on where the linker attempted to position objects, see the map file discussed in Section 7.4 “Map Files”. 3.7.3 How Can I Stop Spurious Warnings From Being Produced? Warnings indicate situations that could possibly lead to code failure. In many situations the code is valid and the warning is superfluous. Always check your code to confirm that it is not a possible source of error and in cases where this is so, there are several ways that warnings can be hidden. • The warning level threshold can be adjusted so that only warnings of a certain importance are printed, see Section 4.6.5.1 “Disabling Messages” • All warnings with a specified ID can be inhibited • In some situations, a pragma can be used to inhibit a warning with a specified ID for certain lines of source code, see Section 5.14.4.11 “The #pragma warning Directive”. 3.7.4 Why Can’t I Even Blink an LED? Even if you have set up the TRIS register and written a value to the port, there are several things that can prevent such a seemingly simple program from working. • Make sure that the device’s Configuration registers are set up correctly, see Section 5.3.5 “Configuration Bit Access”. Make sure that you explicitly specify every bit in these registers and don’t just leave them in their default state. All the configuration features are described in your device data sheet. If the Configuration bits that specify the oscillator source are wrong, for example, the device clock cannot even be running. • If the internal oscillator is being used, in addition to Configuration bits there can be SFRs you need to initialize to set the oscillator frequency and modes, see Section 5.3.6 “Using SFRs From C Code” and your device data sheet. • Either turn off the Watch Dog Timer in the Configuration bits or clear the Watch Dog Timer in your code (see Section Appendix A. “Library Functions”) so that the device does not reset. If the device is resetting, it can never reach the lines of code in your program that blink the LED. Turn off any other features that can cause device Reset until your test program is working. • The device pins used by the port bits are often multiplexed with other peripherals. A pin might be connected to a bit in a port, or it might be an analog input, or it might the output of a comparator, for example. If the pin connected to your LED is not internally connected to the port you are using, then your LED will never operate as expected. The port function tables shown in your device data sheets will show other uses for each pin that will help you identify peripherals to investigate. 2012-2013 Microchip Technology Inc. DS50002053D-page 73 MPLAB® XC8 C Compiler User’s Guide • Make sure you do not have a “read-modify-write” problem. If the device you are using does not have a separate “latch” register (as is the case with mid-range PIC devices) this problem can occur, particularly if the port outputs are driving large loads, such as an LED. You can see that setting one bit turns off another or other unusual events. Create your own latch by using a temporary variable. Rather than read and write the port directly, make modifications to the latch variable. After modifications are complete, copy the latch as a whole to the port. This means you are never reading the port to modify it. Check the device literature for more detailed information. 3.7.5 How Do I Know If the Hardware Stack Has Overflowed? The 8-bit PIC devices have a limited hardware stack that is only used for function (and interrupt function) return addresses, see Section 5.3.4 “Stacks”. If the nesting of function calls and interrupts is too deep, the stack will overflow (wraps around and overwrites previous entries). Code will then fail at a later point — sometimes much later in the call sequence — when it accesses the corrupted return address. The compiler attempts to track stack depth and, when required, swap to a method of calling that does not need the hardware stack (PIC10/12/16 devices only). You have some degree of control over what happens when the stack depth has apparently overflowed, see Section 4.8.51 “--RUNTIME: Specify Runtime Environment” and the stackcall suboption. A call graph shows the call hierarchy and depth that the compiler has determined. This graph is shown in the assembly list file. To understand the information in this graph, see Section 6.6.6 “Call Graph”. Since the runtime behavior of the program cannot be determined by the compiler, it can only assume the worst case and can report that overflow is possible even though it is not. However, no overflow should go undetected if the program is written entirely in C. Assembly code that uses the stack is not considered by the compiler and this must be taken into account. 3.7.6 How Do I Fix a “Can’t find space...” Error? There are a number of different variants of this message, but all essentially imply a similar situation. They all relate to there being no free space large enough to place a block of data or instructions. Due to memory paging, banking or other fragmentation, this message can be issued when seemingly there is enough memory remaining. See Appendix C. “Error and Warning Messages” for more information on your particular error number. 3.7.7 How Do I Fix a “Can’t generate code...” Error? This is a catch-all message which is generated if the compiler has exhausted all possible means of compiling a C expression, see Appendix C. “Error and Warning Messages”. It does not usually indicate a fault in your code. The inability to compile the code can be a deficiency in the compiler, or an expression that requires more registers or resources than are available at that point in the code. This is more likely to occur on baseline devices. In any case, simplifying the offending expression, or splitting a statement into several smaller statements, usually allows the compilation to continue. You can need to use another variable to hold the intermediate results of complicated expressions. DS50002053D-page 74 2012-2013 Microchip Technology Inc. How To’s 3.7.8 How Do I Fix a Fixup Overflow Error? Fixup – the process of replacing a symbolic reference with an actual address – can overflow if the address assigned to the symbol is too large to fit in the address field of the assembly instruction. Most 8-bit PIC assembly instructions specify a file address that is an offset into the currently selected memory bank. If a full unmasked address is specified with these instructions, the linker will be unable to encode the large address value into the instruction and this error will be generated. For example, a mid-range device instruction only allows for file addresses in the range of 0 to 0x7F. However, if such a device has 4 data banks of RAM, the address of variables can range from 0 to 0x1FF. If the symbol of a variable that will be located at address 1D0, for example, is specified with one of these instructions, when the symbol is replaced with its final value, this value will not fit in the address field of the instruction. In most cases, these errors are caused by hand-written assembly code. When writing assembly, it is the programmer’s responsibility to add instructions to select the destination bank and then mask the address being used in the instruction, see Section 3.4.7.5 “What Things Must I Manage When Writing Assembly Code?”. It is important to remember that this is an issue with an assembly instruction and you need to find the instruction at fault before you can proceed. See the relevant error number in Appendix C. “Error and Warning Messages” for specific details on to how to track down the offending instruction. 3.7.9 What Can Cause Corrupted Variables and Code Failure When Using Interrupts? This is usually caused by having variables used by both interrupt and main-line code. If the compiler optimizes access to a variable or access is interrupted by an interrupt routine, then corruption can occur. See Section 3.5.5 “How Do I Share Data Between Interrupt and Main-line Code?” for more information. 2012-2013 Microchip Technology Inc. DS50002053D-page 75 MPLAB® XC8 C Compiler User’s Guide NOTES: DS50002053D-page 76 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Chapter 4. XC8 Command-line Driver 4.1 INTRODUCTION The name of the command-line driver is xc8. MPLAB XC8 can be invoked to perform all aspects of compilation, including C code generation, assembly, and link steps. Even if an IDE is used to assist with compilation, the IDE will ultimately call xc8. Although the internal compiler applications can be called explicitly from the command line, the xc8 driver is the recommended way to use the compiler as it hides the complexity of all the internal applications used and provides a consistent interface for all compilation steps. This chapter describes the steps that the driver takes during compilation, the files that the driver can accept and produce, as well as the command-line options that control the compiler’s operation. The relationship between these command-line options and the controls in the MPLAB IDE Build Options dialog is also described. The following topics are examined in this chapter of the MPLAB XC8 C Compiler User’s Guide: • • • • • • • Invoking the Compiler The Compilation Sequence Runtime Files Compiler Output Compiler Messages MPLAB XC8 Driver Options MPLAB X Option Equivalents 2012-2013 Microchip Technology Inc. DS50002053D-page 77 MPLAB® XC8 C Compiler User’s Guide 4.2 INVOKING THE COMPILER This section explains how to invoke xc8 on the command line, as well as the files that it can read. 4.2.1 Driver Command-line Format The xc8 driver has the following basic command format: xc8 [options] files [libraries] Throughout this manual, it is assumed that the compiler applications are in the console’s search path or that the full path is specified when executing an application. The compiler’s location can be added to the search path when installing the compiler by selecting the Add to environment checkbox at the appropriate time during the installation. It is customary to declare options (identified by a leading dash “-” or double dash “–”) before the files’ names. However, this is not mandatory. The formats of the options are supplied in Section 4.7 “MPLAB XC8 Driver Options”, along with corresponding descriptions of the options. The files can be an assortment of C and assembler source files, and precompiled intermediate files, such as relocatable object (.obj) files or p-code (.p1) files. While the order in which the files are listed is not important, it can affect the order in which code or data appears in memory, and can affect the name of some of the output files. Libraries is a list of user-defined object code or p-code library files that will be searched by the code generator (in the case of p-code libraries) or the linker (for object code libraries), in addition to the standard C libraries. The order of these files will determine the order in which they are searched. It is customary to insert the Libraries list after the list of source file names. However, this is not mandatory. If you are building code using a make system, familiarity with the unique intermediate p-code file format, as described in Section 4.3.3 “Multi-Step Compilation”, is recommended. Object files are seldom used with the MPLAB XC8 C Compiler, unless assembly source modules are in the project. 4.2.1.1 LONG COMMAND LINES The xc8 driver is capable of processing command lines exceeding any operating system limitation if the driver is passed options via a command file. The command file is specified by the @ symbol, which should be immediately followed (i.e., no intermediate space character) by the name of the file containing the command-line arguments that are intended for the driver. Each command-line argument must be separated by one or more spaces and can extended to several lines by using a space and backslash character to separate lines. The file can contain blank lines, which are simply skipped by the driver. The use of a command file means that compiler options and source code filenames can be permanently stored for future reference without the complexity of creating a make utility. In the following example, a command file xyz.xc8 was constructed in a text editor to contain both the options and the file names that are required to compile a project. --chip=16F877A -m \ --opt=all -g \ main.c isr.c After it is saved, the compiler can be invoked with the following command: xc8 @xyz.xc8 DS50002053D-page 78 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.2.2 Environment Variables When hosted on a Windows environment, the compiler uses the registry to store information relating to the compiler installation directory and activation details, along with other configuration settings. That information is required whether the compiler is run on the command line or from within an IDE. Under Linux® and Mac OS® X environments, the registry is replaced by an XML file that stores the same information. On non-Windows hosts, the compiler searches for the XML file in the following ways: 1. The compiler looks for the presence of an environment variable called XC_XML. If present, this variable should contain the full path to the XML file (including the file’s name). 2. If this variable is not defined, the compiler then searches for an environment variable called HOME. This variable typically contains the path to the user’s home directory. The compiler looks for the XML with a name .xc.xml in the directory indicated by the HOME variable. 3. If the HOME environment variable is not defined, the compiler tries to open the file /etc/xc.xml. 4. If none of these methods finds the XML file, an error is generated. When running the compiler on the command line, you can wish to set the PATH environment variable. This allows you to run the compiler driver without specifying the full compiler path with the driver name. Note that the directories specified by the PATH variable are only used to locate the compiler driver. Once the driver is running, it uses the registry or XML file, described above, to locate the internal compiler applications, such as the parser, assembler and linker, etc. The directories specified in the PATH variable do not override the information contained in the registry or XML file. The MPLAB IDE allows the compiler to be selected via a dialog and execution of the compiler does not depend on the PATH variable. 2012-2013 Microchip Technology Inc. DS50002053D-page 79 MPLAB® XC8 C Compiler User’s Guide 4.2.3 Input File Types xc8 distinguishes source files, intermediate files, and library files solely by the file type, or extension. Recognized file types are listed in Table 4-1. Alphabetic case of the extension is not important from the compiler’s point of view, but most operating system shells are case sensitive. TABLE 4-1: xc8 INPUT FILE TYPES File Type Meaning .c C source file .p1 p-code file .lpp p-code library file .as or .asm Assembler source file .obj Relocatable object code file .lib Relocatable object library file .hex Intel HEX file This means, for example, that a C source file must have a .c extension. Assembler files can use either .as or .asm extensions. There are no compiler restrictions imposed on the names of source files, but be aware of case, name-length, and other restrictions that are imposed by your operating system. If you are using an IDE, avoid assembly source files whose base name is the same as the base name of any project in which the file is used. This can result in the source file being overwritten by a temporary file during the build process. The terms “source file” and “module” are often used when talking about computer programs. They are often used interchangeably, but they refer to the source code at different points in the compilation sequence. A source file is a file that contains all or part of a program. They can contain C code, as well as preprocessor directives and commands. Source files are initially passed to the preprocessor by the driver. A module is the output of the preprocessor, for a given source file, after inclusion of any header files (or other source files) which are specified by #include preprocessor directives. All preprocessor directives and commands (with the exception of some commands for debugging) have been removed from these files. These modules are then passed to the remainder of the compiler applications. Thus, a module can be the amalgamation of several source and header files. A module is also often referred to as a translation unit. These terms can also be applied to assembly files, as they can include other header and source files. DS50002053D-page 80 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.3 THE COMPILATION SEQUENCE When you compile a project, many internal applications are called to do the work. This section looks at when these internal applications are executed, and how this relates to the build process of multiple source files. This section should be of particular interest if you are using a make system to build projects. 4.3.1 The Compiler Applications The main internal compiler applications and files are illustrated in Figure 4-1. You can consider the large underlying box to represent the whole compiler, which is controlled by the command line driver, xc8. You can be satisfied just knowing that C source files (shown on the far left) are passed to the compiler and the resulting output files (shown here as a HEX and COFF debug file on the far right) are produced; however, internally there are many applications and temporary files being produced. An understanding of the internal operation of the compiler, while not necessary, does assist with using the tool. To simplify the compiler design, some of the internal applications come in a PIC18 and PIC10/12/16 variant. The appropriate application is executed based on the target device. In fact, the xc8 driver delegates the build commands to one of two command-line drivers: PICC or PICC18. This operation is transparent and xc8 can be considered as “the driver” which does all the work. The driver will call the required compiler applications. These applications are shown as the smaller boxed inside the large driver box. The temporary file produced by each application can also be seen in this diagram. FIGURE 4-1: COMPILER APPLICATIONS AND FILES p-code libraries Command-line driver .p1 p-code files .as assembly source files processed files (module) p-code files assembly file relocatable object file .pre .p1 .as .obj C source files .c .lpp preprocessor p or code generator parser assembler debug file linker relocatable object files 2012-2013 Microchip Technology Inc. .obj cromwell .cof .c hexmate .hhex objtohex .obj .hex absolute object file hex file .lib object libraries hex file hex .hex files DS50002053D-page 81 MPLAB® XC8 C Compiler User’s Guide Table 4-2 lists the compiler applications. The names shown are the names of the executables, which can be found in the bin directory under the compiler’s installation directory. TABLE 4-2: COMPILER APPLICATION NAMES Name Description xc8 (calls PICC or PICC18) Command line driver; the interface to the compiler CLIST Text file formatter CPP The C preprocessor P1 C code parser CGPIC or CGPIC18 Code generator (based on the target device) ASPIC or ASPIC18 Assembler (based on the target device) HLINK Linker OBJTOHEX Conversion utility to create HEX files CROMWELL Debug file converter HEXMATE HEX file utility LIBR Librarian DUMP Object file viewer For example, C source files (.c files) are first passed to the C preprocessor, CPP. The output of this application is .pre files. These files are then passed to the parser application, P1, which produces a p-code file output with extension .p1. The applications are executed in the order specified and temporary files are used to pass the output of one application to the next. The compiler can accept more than just C source files. Table 4-1 lists all the possible input file types, and these files can be seen in this diagram, on the top and bottom, being passed to different compilation applications. They are processed by these applications and then the application output joins the normal flow indicated in the diagram. For example, assembly source files are passed straight to the assembler application1 and are not processed at all by the code generator. The output of the assembler (an object file with .obj extension) is passed to the linker in the usual way. You can see that any p-code files (.p1 extension) or p-code libraries (.lpp extension) that are supplied on the command line are initially passed to the code generator. Other examples of input files include object files (.obj extension) and object libraries (.lib extension), both of which are passed initially to the linker, and even HEX files (.hex extension), which are passed to one of the utility applications, called HEXMATE, which is run right at the end of the compilation sequence. Some of the temporary files shown in this diagram are actually preserved and can be inspected after compilation has concluded. There are also driver options to request that the compilation sequence stop after a particular application and the output of that application becomes the final output. 1. Assembly file will be preprocessed before being passed to the assembler if the -P option is selected. DS50002053D-page 82 2012-2013 Microchip Technology Inc. XC8 Command-line Driver FIGURE 4-2: MULTI-FILE COMPILATION Intermediate files C file preprocess & parse pcode C file preprocess & parse pcode code generation assemble link library files First stage of compilation 4.3.2 Second stage of compilation Single-Step Compilation Figure 4-1 showed us the files that are generated by each application and the order in which these applications are executed. However this does not indicate how these applications are executed when there is more than one source file being compiled. Consider the case when there are two C source files that form a complete project and that are to be compiled, as is the case shown in Figure 4-2. If these files are called main.c and io.c, these could be compiled with a single command, such as: xc8 --chip=16F877A main.c io.c This command will compile the two source files all the way to the final output, but internally we can consider this compilation as consisting of two stages. The first stage involves processing of each source file separately, and generating some sort of intermediate file for each source file. The second stage involves combining all these intermediate files and further processing to form the final output. An intermediate file is a particular temporary file that is produced and marks the midpoint between the first and second stage of compilation. The intermediate file used by xc8 is the p-code (.p1 extension) file output by the parser, so there will be one p-code file produced for each C source file. As indicated in the diagram, CPP and then P1 are executed to form this intermediate file. (For clarity, the CPP and P1 applications have been represented by the same block in the diagram.) In the second stage, the code generator reads in all the intermediate p-code files and produces a single assembly file output, which is then passed to the subsequent applications that produce the final output. The desirable attribute of this method of compilation is that the code generator, which is the main application that transforms from the C to the assembly domain, sees the entire project source code via the intermediate files. Traditional compilers have always used intermediate files that are object files output by the assembler. These intermediate object files are then combined by the linker and further processed to form the final output. This method of compilation is shown in Figure 4-3. It shows that the code generator is executed once for each source file. So, the code generator can only analyze that part of the project that is contained in the source file that is currently being compiled. The MPLAB XC16 and XC32 compilers work in this fashion. Using object files as the intermediate file format with MPLAB XC8 C Compiler will defeat many features the compiler uses to optimize code. Always use p-code files as the intermediate file format if you are using a make system to build projects. 2012-2013 Microchip Technology Inc. DS50002053D-page 83 MPLAB® XC8 C Compiler User’s Guide FIGURE 4-3: THE TRADITIONAL COMPILATION SEQUENCE Intermediate files C file preprocess & parse code generation assemble .obj files C file preprocess & parse code generation assemble .obj files link library files Second stage of compilation First stage of compilation When compiling files of mixed types, this can still be achieved with just one invocation of the compiler driver. As discussed in Section 4.3 “The Compilation Sequence”, the driver will pass each input file to the appropriate compiler application. For example, the files, main.c, io.c, mdef.as and c_sb.lpp are to be compiled. To perform this in a single step, the following command line could be used. xc8 --chip=16F877A main.c io.c mdef.as c_sb.lpp As shown in Figure 4-1 and Figure 4-2, the two C files (main.c and io.c) will be compiled to intermediate p-code files; these, along with the p-code library file (c_sb.lpp) will be passed to the code generator. The output of the code generator, as well as the assembly source file (mdef.as), will be passed to the assembler. The driver will recompile all source files, regardless of whether they have changed since the last build. IDEs (such as MPLAB® IDE) and make utilities must be employed to achieve incremental builds. See also, Section 4.3.3 “Multi-Step Compilation”. Unless otherwise specified, a HEX file and Microchip COFF file are produced as the final output. All intermediate files remain after compilation has completed, but most other temporary files are deleted, unless you use the --NODEL option (see Section 4.8.40 “--NODEL: Do Not Remove Temporary Files”) which preserves all generated files except the run-time start-up file. Note that some generated files can be in a different directory than your project source files. See Section 4.8.43 “--OUTDIR: Specify a Directory for Output Files”, and Section 4.8.41 “--OBJDIR: Specify a Directory for Intermediate Files”, which can both control the destination for some output files. DS50002053D-page 84 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.3.3 Multi-Step Compilation Make utilities and IDEs, such as MPLAB IDE, allow for an incremental build of projects that contain multiple source files. When building a project, they take note of which source files have changed since the last build and use this information to speed up compilation. For example, if compiling two source files, but only one has changed since the last build, the intermediate file corresponding to the unchanged source file need not be regenerated. MPLAB IDE is aware of the different compilation sequence employed by xc8 and takes care of this for you. From MPLAB IDE you can select an incremental build (Build Project icon), or fully rebuild a project (Clean and Build Project icon). If the compiler is being invoked using a make utility, the make file will need to be configured to recognized the different intermediate file format and the options used to generate the intermediate files. Make utilities typically call the compiler multiple times: once for each source file to generate an intermediate file, and once to perform the second stage compilation. You can also wish to generate intermediate files to construct your own library files. However, xc8 is capable of constructing libraries in a single step, so this is typically not necessary. See Section 4.8.44 “--OUTPUT= type: Specify Output File Type”, for more information on library creation. The option --PASS1 (see Section 4.8.45 “--PARSER: Specify Parser Mode”) is used to tell the compiler that compilation should stop after the parser has executed. This will leave the p-code intermediate file behind on successful completion. For example, the files main.c and io.c are to be compiled using a make utility. The command lines that the make utility should use to compile these files might be something like: xc8 --chip=16F877A --pass1 main.c xc8 --chip=16F877A --pass1 io.c xc8 --chip=16F877A main.p1 io.p1 If is important to note that the code generator needs to compile all p-code or p-code library files associated with the project in the one step. When using the --PASS1 option the code generator is not being invoked, so the above command lines do not violate this requirement. Using object files as the intermediate file format with MPLAB XC8 C Compiler will defeat many features the compiler uses to optimize code. Always use p-code files as the intermediate file format if you are using a make system to build projects. 2012-2013 Microchip Technology Inc. DS50002053D-page 85 MPLAB® XC8 C Compiler User’s Guide 4.3.4 Compilation of Assembly Source Since the code generator performs many tasks that were traditionally performed by the linker, there could be complications when assembly source is present in a project. Assembly files are traditionally processed after C code, but it is necessary to have this performed first so that specific information contained in the assembly code can be conveyed to the code generator. The specific information passed to the code generator is discussed in more detail in Section 5.12.3 “Interaction between Assembly and C Code”. When assembly source is present, the order of compilation is as shown in Figure 4-4. FIGURE 4-4: C file C file COMPILATION SEQUENCE WITH ASSEMBLY FILES preprocess & parse preprocess & parse pcode pcode library files ASM file code generation assemble link assemble driver OBJ file First, any assembly source files are assembled to form object files. These files, along with any other objects files that are part of the project, are scanned by the command-line driver and the information is passed to the code generator; where it subsequently builds the C files, as has been described earlier. 4.3.4.1 INTERMEDIATE FILES AND ASSEMBLY SOURCE The intermediate file format associated with assembly source files is the same as that used in traditional compilers; i.e., an object file (.obj extension). Assembly files are never passed to the code generator and so the code generator technology does not alter the way these files are compiled. The -C option (see Section 4.8.1 “-C: Compile to Object File”) is used to generate object files and to halt compilation after the assembly step. 4.3.5 Printf Check An extra execution of the code generator is performed prior to the actual code generation phase. This pass is part of the process by which the printf library function is customized; see Section 5.11.1 “The printf Routine”, for more details. This pass is only associated with scanning the C source code for printf placeholder usage and you will see the code generator being executed if you select the verbose option when you build; see Section 4.8.14 “-V: Verbose Compile”. DS50002053D-page 86 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.4 RUNTIME FILES In addition to the C and assembly source files specified on the command line, there are also compiler-generated source files and pre-compiled library files which might be compiled into the project by the driver. These files contain: • • • • • • C Standard library routines Implicitly called arithmetic routines User-defined library routines The runtime startup code The powerup routine The printf routine. Strictly speaking, the powerup routine is neither a compiler-generated source, nor a library routine. It is fully defined by the user. Because it is very closely associated with the runtime startup module, it is discussed with the other runtime files in the following sections. 4.4.1 Library Files The names of the C standard library files appropriate for the selected target device, and other driver options, are determined by the driver and passed to the code generator and linker. You do not need to include library files into your project manually. P-code libraries (.lpp libraries) are used by the code generator, and object code libraries (.lib files) are used by the linker. Most library routines are derived from p-code libraries. By default, xc8 will search the lib directory under the compiler installation directory for library files that are required during compilation. 4.4.1.1 STANDARD LIBRARIES The C standard libraries contain a standardized collection of functions, such as string, math and input/output routines. The range of these functions is described in Appendix A. “Library Functions”. Although it is considered a library function, the printf function’s code is not found in these library files. C source code for this function is generated from a special C template file that is customized after analysis of the user’s C code. See “PRINTF, VPRINTF”, for more information on using the printf library function, and Section 5.11.1 “The printf Routine”, for information on how the printf function is customized when you build a project. The libraries also contain C routines that are implicitly called by the output code of the code generator. These are routines that perform tasks such as floating-point operations, integer division and type conversions, and that cannot directly correspond to a C function call in the source code. The library name format is family-type-options.lpp, where the following apply. • family can be pic18 for PIC18 devices, or pic for all other 8-bit PIC devices • type indicates the sort of library functionality provided and can be stdlib for the standard library functions, or trace, etc. • options indicates hyphen-separated names to indicate variants of the library to accommodate different compiler options or modes, e.g., htc for HI-TECH C compatibility, d32 for 32-bit doubles, etc. For example, the standard library for baseline and midrange devices using 24-bit double types is pic-stdlib-d24.lpp. All the libraries are present in the lib directory of the compiler installation. Search this directory for the full list of all libraries supplied. 2012-2013 Microchip Technology Inc. DS50002053D-page 87 MPLAB® XC8 C Compiler User’s Guide 4.4.1.2 USER-DEFINED LIBRARIES User-defined libraries can be created and linked in with programs as required. Library files are easier to manage and can result in faster compilation times, but must be compatible with the target device and options for a particular project. Several versions of a library might need to be created to allow it to be used for different projects. Libraries can be created manually using the compiler and the librarian, LIBR. See Section 8.2 “Librarian” for more information on the librarian and creating library files using this application. Alternatively, library files can be created directly from the compiler by specifying a library output using the --OUTPUT option; see Section 4.8.44 “--OUTPUT= type: Specify Output File Type”. User-created libraries that should be searched when building a project can be listed on the command line along with the source files. As with Standard C library functions, any functions contained in user-defined libraries should have a declaration added to a header file. It is common practice to create one or more header files that are packaged with the library file. These header files are then included into source code when required. Library files specified on the command line are initially scanned for unresolved symbols; so, these files can redefine anything that is defined in the C standard libraries. See also, Section 5.15.4 “Changing and Linking the Allocated Section”. 4.4.2 Startup and Initialization A C program requires certain objects to be initialized and the device to be in a particular state before it can begin execution of its function main. It is the job of the runtime startup code to perform these tasks. Section 5.10.1 “Runtime Startup Code” details the specific actions taken by this code and how it interacts with programs you write. Rather than the traditional method of linking in a generic, precompiled routine, the MPLAB XC8 C Compiler determines what runtime startup code is required from the user’s program and then generates this code each time you build. Both the driver and code generator are involved in generating the runtime startup code. The driver creates the code that handles device setup. This code is placed into a separate assembly startup module. The code generator produces code that initializes the C environment, such as clearing uninitialized C variables and copying initialized C variables. This code is output along with the rest of the C program. The runtime startup code is regenerated every time you build a project. The file created by the driver can be deleted after compilation, and this operation can be controlled with the keep suboption to the --RUNTIME option. The default operation of the driver is to keep the startup module; however, if using MPLAB IDE to build, the file will be deleted unless you indicate otherwise in the Project Properties dialog. If the startup module is kept, it will be called startup.as and will be located in the current working directory. If you are using an IDE to perform the compilation, the destination directory can be dictated by the IDE itself. MPLAB X IDE stores this file in the dist/default/production directory in your project directory. Generation of the runtime startup code is an automatic process that does not require any user interaction; however, some aspects of the runtime code can be controlled, if required, using the --RUNTIME option. Section 4.8.51 “--RUNTIME: Specify Runtime Environment” describes the use of this option. See Section 5.10.1 “Runtime Startup Code”, which describes the functional aspects of the code contained in this module and its effect on program operation. The runtime startup code is executed before main. However, if you require any special initialization to be performed immediately after Reset, you should use the powerup feature described later in Section 5.10.2 “The Powerup Routine”. DS50002053D-page 88 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.5 COMPILER OUTPUT There are many files created by the compiler during the compilation. A large number of these are intermediate files. Some are deleted after compilation is complete, but many remain and are used for programming the device, or for debugging purposes. 4.5.1 Output Files The names of many output files use the same base name as the source file from which they were derived. For example, the source file input.c will create a p-code file called input.p1. Some of the output files contain project-wide information and are not directly associated with any one particular input file, e.g., the map file. If the names of these output files are not specified by a compiler option, their base name is derived from the first C source file listed on the command line. If there are no files of this type specified, the name is based on the first input file (regardless of type) on the command line. If you are using an IDE, such as MPLAB X IDE, to specify options to the compiler, there is typically a project file that is created for each application. The name of this project is used as the base name for project-wide output files, unless otherwise specified by the user. However, check the manual for the IDE you are using, for more details. Note: Throughout this manual, the term project name will refer to either the name of the project created in the IDE, or the base name (file name without extension) of the first C source file specified on the command line. The compiler is directly able to produce a number of the output file formats that are used by the 8-bit PIC development tools. The default behavior of xc8 is to produce a Microchip format COFF and Intel HEX output. Unless changed by a driver option, the base names of these files will be the project name. The default output file types can be controlled by compiler options, e.g., the --OUTPUT option. The extensions used by these files are fixed and are listed together with this option’s description in Section 4.8.44 “--OUTPUT= type: Specify Output File Type”. The COFF file is used by debuggers to obtain debugging information about the project. The compiler can produce ELF/DWARF debugger files, although these are not compatible with MPLAB IDE v8 and early versions of MPLAB X IDE. You must specifically select ELF output for these files to be produced. ELF/DWARF files allow for more accurate debugging. Use of these files correct several COFF-related issues that prevent you from correctly viewing objects, in particular pointer variables, in the IDE. Ensure the IDE version you are using supports ELF before selecting this option. Table 4-16 shows all output format options available with xc8 using the --OUTPUT option. The File Type column lists the filename extension that is used for the output file. 4.5.1.1 SYMBOL FILES By default, xc8 creates symbol files that are used to generate the debug output files, such as COFF and ELF files. These files include a SYM file (.sym extension) and a CMF file (.cmf extension), and both are produced by the linker. In addition, there is a SDB file (.sdb extension) produced by the code generator. The SDB file contains type information, and the SYM and CMF files contain address information. The SDB and SYM/CMF files, in addition to the HEX file, are combined by the CROMWELL application to produce the output debug files, such as the COFF file. The CMF file largely replaces the older SYM file format. They contain similar information, but CMF files are more detailed and enable more accurate debug files to be generated. 2012-2013 Microchip Technology Inc. DS50002053D-page 89 MPLAB® XC8 C Compiler User’s Guide 4.5.2 Diagnostic Files Two valuable files produced by the compiler are the assembly list file, produced by the assembler, and the map file, produced by the linker. The compiler options --ASMLIST (see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”) generates a list file, and the -M option (see Section 4.8.7 “-M: Generate Map File”) specifies generation of a map file. The assembly list file contains the mapping between the original source code and the generated assembly code. It is useful for information such as how C source was encoded, or how assembly source can have been optimized. It is essential when confirming if compiler-produced code that accesses objects is atomic, and shows the psects in which all objects and code are placed. For an introductory guide to psects, see Section 5.15.1 “Program Sections”. Also, see Section 6.5 “Assembly-Level Optimizations”, for more information on the contents of this file. There is one list file produced for the entire C program, including C library files. It is assigned the project name and the extension .lst. One additional list file is produced for each assembly source file compiled in the project. The map file shows information relating to where objects were positioned in memory. It is useful for confirming if user-defined linker options were correctly processed, and for determining the exact placement of objects and functions. It also shows all the unused memory areas in a device and memory fragmentation. See Section 7.4 “Map Files”, for complete information on the contents of this file. There is one map file produced when you build a project, assuming the linker was executed and ran to completion. The file is assigned the project name and a .map extension. DS50002053D-page 90 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.6 COMPILER MESSAGES All compiler applications, including the command-line driver, xc8, use textual messages to report feedback during the compilation process. A centralized messaging system is used to produce the messages, which allows consistency during all stages of the compilation process. The messaging system is described in this section and a complete list of all warning and error messages can be found in Appendix C. “Error and Warning Messages”. 4.6.1 Messaging Overview A message is referenced by a unique number that is passed to the messaging system by the compiler application that needs to convey the information. The message string corresponding to this number is obtained from Message Description Files (MDF), which are stored in the dat directory in the compiler’s installation directory. When a message is requested by a compiler application, its number is looked up in the MDF that corresponds to the currently selected language. The language of messages can be altered as discussed in Section 4.6.2 “Message Language”. Once found, the alert system can read the message type and the string to be displayed from the MDF. Several different message types are described in Section 4.6.3 “Message Type”; and the type can be overridden by the user, as described in that same section. The user is also able to set a threshold for warning message importance, so that only those that the user considers significant will be displayed. In addition, messages with a particular number can be disabled. A pragma can also be used to disable a particular message number within specific lines of code. These methods are explained in Section 4.6.5.1 “Disabling Messages”. Provided the message is enabled and it is not a warning message whose level is below the current warning threshold, the message string will be displayed. In addition to the actual message string, there are several other pieces of information that can be displayed, such as the message number, the name of the file for which the message is applicable, the file’s line number and the application that issued the message, etc. If a message is an error, a counter is incremented. After a specific amount of errors has been reached, compilation of the current module will cease. The default number of errors that will cause this termination can be adjusted by using the --ERRORS option; see Section 4.8.28 “--ERRORS: Maximum Number of Errors”. This counter is reset for each internal compiler application, thus specifying a maximum of five errors will allow up to five errors from the parser, five from the code generator, five from the linker, five from the driver, etc. Although the information in the MDF can be modified with any text editor, this is not recommended. Message behavior should only be altered using the options and pragmas described in the following sections. 2012-2013 Microchip Technology Inc. DS50002053D-page 91 MPLAB® XC8 C Compiler User’s Guide 4.6.2 Message Language The xc8 driver supports more than one language for displayed messages. There is one MDF for each language supported. Under Windows®, the default language can be specified when installing the compiler. The default language can be changed on the command line using the --LANG option; see Section 4.8.35 “--LANG: Specify the Language for Messages”. Alternatively, it can be changed permanently by using the --LANG option together with the --SETUP option which will store the default language in either the registry, under Windows, or in the XML configuration file on other systems. On subsequent builds, the default language used will be that specified. Table 4-3 shows the MDF applicable for the currently supported languages. TABLE 4-3: SUPPORTED LANGUAGES Language MDF name English en_msgs.txt German de_msgs.txt French fr_msgs.txt If a language other than English is selected, and the message cannot be found in the appropriate non-English MDF, the alert system tries to find the message in the English MDF. If an English message string is not present, a message is displayed that is similar to this one: error/warning (*) generated, but no description available where * indicates the message number that was generated that will be printed; otherwise, the message in the requested language will be displayed. 4.6.3 Message Type There are four types of messages. These are described below along with the compiler’s behavior when encountering a message of each type. Advisory Messages convey information regarding a situation the compiler has encountered or some action the compiler is about to take. The information is being displayed “for your interest”, and typically requires no action to be taken. Compilation will continue as normal after such a message is issued. Warning Messages indicate source code or some other situation that can be compiled, but is unusual and can lead to a runtime failure of the code. The code or situation that triggered the warning should be investigated; however, compilation of the current module will continue, as will compilation of any remaining modules. Error Messages indicate source code that is illegal or that compilation of this code cannot take place. Compilation will be attempted for the remaining source code in the current module, but no additional modules will be compiled and the compilation process will then conclude. Fatal Error Messages indicate a situation in which the compilation cannot proceed and requires that the compilation process to stop immediately. DS50002053D-page 92 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.6.4 Message Format By default, messages are printed in a human-readable format. This format can vary from one compiler application to another, since each application reports information about different file formats. Some applications (for example, the parser) are typically able to pinpoint the area of interest down to a position on a particular line of C source code, whereas other applications, such as the linker, can at best only indicate a module name and record number, which is less directly associated with any particular line of code. Some messages relate to issues in driver options that are in no way associated with any source code. There are several ways of changing the format in which message are displayed, which are discussed below. The driver option -E (with or without a filename) alters the format of all displayed messages. See Section 4.8.3 “-E: Redirect Compiler Errors to a File”, for details. Using this option produces messages that are better suited to machine parsing, and are less user-friendly. Typically, each message is displayed on a single line. The general form of messages produced when using the -E option is: filename line: (message number) message string (type) The -E option also has another effect. When used, the driver first checks to see if special environment variables have been set. If so, the format dictated by these variables is used as a template for all messages that will be produced by all compiler applications. The names of these environment variables are given in Table 4-4. TABLE 4-4: MESSAGING ENVIRONMENT VARIABLES Variable Effect HTC_MSG_FORMAT All advisory messages HTC_WARN_FORMAT All warning messages HTC_ERR_FORMAT All error and fatal error messages The value of these environment variables are strings that are used as templates for the message format. Printf-like placeholders can be placed within the string to allow the message format to be customized. The placeholders, and what they represent, are presented in Table 4-5. TABLE 4-5: MESSAGING PLACEHOLDERS Placeholder Replacement %a Application name %c Column number %f Filename %l Line number %n Message number %s Message string (from MDF) If these options are used in a DOS batch file, two percent characters will need to be used to specify the placeholders, as DOS interprets a single percent character as an argument and will not pass this on to the compiler. For example: SET HTC_ERR_FORMAT="file %%f: line %%l" 2012-2013 Microchip Technology Inc. DS50002053D-page 93 MPLAB® XC8 C Compiler User’s Guide Environment variables, in turn, can be overridden by the driver options: --MSGFORMAT, --WARNFORMAT and --ERRFORMAT; see Section 4.8.27 “--ERRFORMAT: Define Format for Compiler Messages”. These options take a string as their argument. The option strings are formatted, and can use the same placeholders as their variable counterparts. For example, a project is compiled, but, as shown, produces a warning from the parser and an error from the linker (numbered 362 and 492, respectively). main.c: main() 17: ip = &b; ^ (362) redundant "&" applied to array (warning) (492) attempt to position absolute psect "text" is illegal Notice that the parser message format identifies the particular line and position of the offending source code. If the -E option is now used and the compiler issues the same messages, the compiler will output: main.c: 12: (362) redundant "&" applied to array (warning) (492) attempt to position absolute psect "text" is illegal (error) The user now uses the --WARNFORMAT in the following fashion: --WARNFORMAT="%a %n %l %f %s" When recompiled, the following output will be displayed: parser 362 12 main.c redundant "&" applied to array (492) attempt to position absolute psect "text" is illegal (error) Notice that the format of the warning was changed, but that of the error message was not. The warning format now follows the specification of the environment variable. The application name (parser) was substituted for the %a placeholder, the message number (362) substituted the %n placeholder, etc. 4.6.5 Changing Message Behavior Both the attributes of individual messages and general settings for the messaging system can be modified during compilation. There are both driver options and C pragmas that can be used to achieve this. 4.6.5.1 DISABLING MESSAGES Each warning message has a default number indicating a level of importance. This number is specified in the MDF and ranges from -9 to 9. The higher the number, the more important the warning. Warning messages can be disabled by adjusting the warning level threshold using the --WARN driver option; see Section 4.8.61 “--WARN: Set Warning Level”. Any warnings whose level is below that of the current threshold are not displayed. The default threshold is 0 which implies that only warnings with a warning level of 0 or higher will be displayed by default. The information in this option is propagated to all compiler applications, so its effect will be observed during all stages of the compilation process. Warnings can also be disabled by using the --MSGDISABLE option; see Section 4.8.38 “--MSGDISABLE: Disable Warning Messages”. This option takes a comma-separated list of warning numbers. The warnings corresponding to the numbers listed are disabled and will never be issued, regardless of the current warning level threshold. If the special message number 0 is specified, then all warning messages are disabled. DS50002053D-page 94 2012-2013 Microchip Technology Inc. XC8 Command-line Driver Some warning messages can also be disabled by using the warning pragma. This pragma will only affect warnings that are produced by either the parser or the code generator; i.e., errors directly associated with C code. See Section 5.14.4.11 “The #pragma warning Directive” for more information on this pragma. Error messages can also be disabled; however, a more verbose form of the above command is required to confirm the action. To specify an error message number in the --MSGDISABLE command, each error number must be followed by :off to ensure that it is disabled. For example: --MSGDISABLE=1257,195:off,194:off will disable warning 1257, and errors 195 and 194. Note: 4.6.5.2 Disabling error or warning messages in no way fixes the condition that triggered the message. Always use extreme caution when exercising these options. CHANGING MESSAGE TYPES It is also possible to change the type of some messages. This can only be done for messages generated by the parser or code generator. See Section 5.14.4.11 “The #pragma warning Directive”, for more information on this pragma. 2012-2013 Microchip Technology Inc. DS50002053D-page 95 MPLAB® XC8 C Compiler User’s Guide 4.7 MPLAB XC8 DRIVER OPTIONS This section looks at the general form of xc8 command-line options and what action the compiler will perform if no option is specified for a certain feature. 4.7.1 General Option Formats All single letter options are identified by a leading dash character, “-”, for example: -C. Some single letter options specify an additional data field that follows the option name immediately and without any whitespace, for example: -Ddebug. In this manual, options are written in upper case and suboptions are written in lower case. Multi-letter, or word, options have two leading dash characters, for example: --ASMLIST. (Because of the double dash, the driver can determine that the option --DOUBLE, for example, is not a -D option followed by the argument OUBLE.) Some of these word options use suboptions which typically appear as a comma-separated list following an equal character, =, for example: --OUTPUT=hex,cof. The exact formats of the options vary. The options and formats are described in detail in the following sections. Some commonly used suboptions include default, which represent the default specification that would be used if this option was absent altogether; all, which indicates that all the available suboptions should be enabled as if they had each been listed; and none, which indicates that all suboptions should be disabled. For example: --OPT=none will turn off all optimizers. Some suboptions can be prefixed with a plus character, +, to indicate that they are in addition to the other suboptions present; or a minus character “-”, to indicate that they should be excluded. For example: --OPT=default,-asm indicates that the default optimization be used, but that the assembler optimizer should be disabled. If the first character after the equal sign is + or -, then the default keyword is implied. For example: --OPT=-asm is the same as the previous example. See the –-HELP option, Section 4.8.33 “--HELP: Display Help”, for more information about options and suboptions. 4.7.2 Default Options If you run the compiler driver from the command line and do not specify the option for a feature, it will default to a certain state. You can also specify the default suboption (to double-dash options) that will also invoke the default behavior. You can check what the default behavior is by using the --HELP=option on the command line; see Section 4.8.33 “--HELP: Display Help”. If you are compiling from within the MPLAB X IDE, it will, by default, issue explicit options to the compiler (unless changed in the Project Properties dialog), and these options can be different to those that are the default on the command line. For example, unless you specify the --ASMLIST option on the command line, the default operation of the compiler is not to produce an assembly list file. But, if you are compiling from within the MPLAB X IDE, the default operation (this, in fact, cannot be disabled) is to always produce an assembly list file. If you are compiling the same project from the command line and from the MPLAB X IDE, always check that all options are explicitly specified. DS50002053D-page 96 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8 OPTION DESCRIPTIONS Most aspects of the compilation can be controlled using the command-line driver, xc8. The driver will configure and execute all required applications, such as the code generator, assembler and linker. xc8 recognizes the compiler options which are tabled below and are explained in detail in the sections following. The case of the options is not important; however, command shells in most operating systems are case sensitive when it comes to the names of files. TABLE 4-6: DRIVER OPTIONS Option Meaning -C Compile to object file and stop -Dmacro Define preprocessor macro symbol -Efilename Redirect compile errors -G[filename] Generate symbolic debug information -Ipath Specify include path -Largument Set linker option -M[filename] Generate map file -Nnumber Specify identifier length -Ofile Specify output filename and type -P Preprocess assembly source -Q Quiet mode -S Compile to assembly file and stop -Umacro Undefine preprocessor macro symbol -V Verbose mode --ADDRQUAL=qualifier Specify address space qualifier handling --ASMLIST Generate assembly list file --CHAR=type Default character type (defunct) --CHECKSUM=specification Calculate a checksum and store the result in program memory --CHIP=device Select target device --CHIPINFO Print device information --CODEOFFSET=value Specify ROM offset address --DEBUGGER=type Set debugger environment --DOUBLE=size Size of double type --ECHO Echo command line --EMI=mode Select external memory interface operating mode --ERRATA=type Specify errata workarounds --ERRFORMAT=format Set error format --ERRORS=number Set maximum number of errors --EXT=extensions Specify C language extensions --FILL=specification Specify a ROM-fill value for unused memory --FLOAT=size Size of float type --GETOPTION=argument Get advanced options --HELP=option Help --HTML=file Generate HTML debug files --LANG=language Specify language --MEMMAP=mapfile Display memory map 2012-2013 Microchip Technology Inc. DS50002053D-page 97 MPLAB® XC8 C Compiler User’s Guide TABLE 4-6: DRIVER OPTIONS (CONTINUED) Option Meaning --MODE=mode Choose operating mode --MSGDISABLE=list Disable warning messages --MSGFORMAT=specification Set advisory message format --NODEL Do not remove temporary files --NOEXEC Do not execute compiler applications --OBJDIR=path Set object files directory --OPT=optimizations Control optimization --OUTDIR=path Set output directory --OUTPUT=path Set output formats --PARSER=mode Specify parser mode --PASS1 Produce intermediate p-code file and stop --PRE Produce preprocessed source files and stop --PROTO Generate function prototypes --RAM=ranges Adjust RAM ranges --ROM=ranges Adjust ROM ranges --RUNTIME=options Specify runtime options --SCANDEP Scan for dependencies --SERIAL=specification Insert a hexadecimal code or serial number --SETOPTION=argument Set advanced options --SETUP=specification Setup the compiler --SHROUD Shroud (obfuscate) generated p-code files --STACK=type[:sizes] Specify data stack type and sizes --STRICT Use strict ANSI keywords --SUMMARY=type Summary options --TIME Report compilation times --VER Show version information --WARN=number Set warning threshold level --WARNFORMAT=specification Set warning format 4.8.1 -C: Compile to Object File The -C option is used to halt compilation after executing the assembler, leaving a relocatable object file as the output. It is frequently used when compiling assembly source files using a make utility. It cannot be used unless all C source files are present on the command line. Use --PASS1 to generate intermediate files from C source, see Section 4.8.45 “--PARSER: Specify Parser Mode”. See Section 4.3.3 “Multi-Step Compilation”, for more information on generating and using intermediate files. DS50002053D-page 98 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.2 -D: Define Macro The -D option is used to define a preprocessor macro on the command line, exactly as if it had been defined using a #define directive in the source code. This option can take one of two forms, -Dmacro which is equivalent to: #define macro 1 placed at the top of each module compiled using this option, or -Dmacro= text which is equivalent to: #define macro text where text is the textual substitution required. Thus, the command: xc8 --CHIP=16F877AA -Ddebug -Dbuffers=10 test.c will compile test.c with macros defined exactly as if the C source code had included the directives: #define #define debug 1 buffers 10 Defining macros as C string literals requires bypassing any interpretation issues in the operating system that is being used. To pass the C string, "hello world", (including the quote characters) in the Windows environment, use: "-DMY_STRING=\\\"hello world\\\"" (you must include the quote characters around the entire option, as there is a space character in the macro definition). Under Linux or Mac OS X, use: -DMY_STRING=\"hello\ world\". See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 4.8.3 -E: Redirect Compiler Errors to a File This option has two purposes. The first is to change the format of displayed messages. The second is to optionally allow messages to be directed to a file, as some editors do not allow the standard command line redirection facilities to be used when invoking the compiler. The general form of messages produced with the -E option in force is: filename line_number: (message number) message string (type) If a filename is specified immediately after -E, it is treated as the name of a file to which all messages (errors, warnings, etc.) will be printed. For example, to compile x.c and redirect all errors to x.err, use the command: xc8 --CHIP=16F877AA -Ex.err x.c The -E option also allows errors to be appended to an existing file by specifying an addition character, +, at the start of the error filename, for example: xc8 --CHIP=16F877AA -E+x.err y.c If you wish to compile several files and combine all of the errors generated into a single text file, use the -E option to create the file then use -E+ when compiling all the other source files. For example, to compile a number of files with all errors combined into a file called project.err, you could use the - E option as follows: xc8 --CHIP=16F877AA -Eproject.err -O --PASS1 main.c xc8 --CHIP=16F877AA -E+project.err -O --PASS1 part1.c xc8 --CHIP=16F877AA -E+project.err -C asmcode.as Section 4.6 “Compiler Messages” has more information regarding this option as well as an overview of the messaging system and other related driver options. 2012-2013 Microchip Technology Inc. DS50002053D-page 99 MPLAB® XC8 C Compiler User’s Guide 4.8.4 -I: Include Search Path Use -I to specify an additional directory to search for header files which have been included using the #include directive. The directory can either be an absolute or relative path. The -I option can be used more than once if multiple directories are to be searched. The compiler’s include directory containing all standard header files is always searched, even if no -I option is present. If header filenames are specified using quote characters rather than angle brackets, as in #include "lcd.h", then the current working directory is searched in addition to the compiler’s include directory. Note that if compiling within MPLAB IDE, the search path is relative to the output directory, not the project directory. These default search paths are searched after any user-specified directories have been searched. For example, the following code: xc8 --CHIP=16F877AA -C -Ic:\include -Id:\myapp\include test.c will search the directories c:\include and d:\myapp\include for any header files included into the source code, then search the default include directory. Under Windows OS, be aware that use of the directory backslash character may unintentionally form an escape sequence. For example, to specify an include file path that ends with a directory separator character and which is quoted, use -I"E:\\" instead of -I"E:\", to avoid the escape sequence \". Note that MPLAB X IDE will quote any include file path you specify in the project properties. This option has no effect for files that are included into assembly source using the assembly INCLUDE directive. See Section 6.4.10.4 “INCLUDE”, for details. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 4.8.5 -L: Scan Library The -L option is used to specify additional libraries that are to be scanned by the linker. Libraries specified using the -L option are scanned before the standard C library, allowing additional versions of standard library functions to be accessed. The argument to -L is a library keyword to which the prefix pic; numbers representing the device range, number of ROM pages and the number of RAM banks; and the suffix .lib are added. In this way, the option -Ll, when compiling for a 16F877A, will, for example, scan the library pic42c-l.lib and the option -Lxx will scan a library called pic42c-xx.lib. All libraries must be located in the lib directory of the compiler installation directory. As indicated, the argument to the -L option is not a complete library filename. If you wish the linker to scan libraries whose names do not follow the naming convention previously mentioned or whose locations are not in the lib subdirectory, simply include the libraries’ names on the command line along with your source files, or add these to your project. DS50002053D-page 100 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.6 -L-: Adjust Linker Options Directly The -L driver option can be used to specify an option that will be passed directly to the linker. If -L is followed immediately by text starting with a dash character “-”, the text will be passed directly to the linker without being interpreted by the xc8 command-line driver. In the event that the -L option is not followed immediately by a dash character, it is assumed the option is the library scan option; see Section 4.8.5 “-L: Scan Library” for more information. For example, if the option -L-N is specified, the -N option will be passed on to the linker without any subsequent interpretation by the driver. The linker will then process this option, when, and if, it is invoked, and perform the appropriate operation. Take care with command-line options. The linker cannot interpret command-line driver options; similarly, the driver cannot interpret linker options. In most situations, it is always the command-line driver, xc8, that is being executed. If you need to add alternate linker settings in the Linker category of the Project Properties dialog, you must add driver options (not linker options). These driver options will be used by the driver to generate the appropriate linker options during the linking process. The -L option is a means of allowing a linker option to be specified via a driver option. The -L option is especially useful when linking code which contains non-standard program sections (or psects), as can be the case if the program contains hand-written assembly code which contains user-defined psects (see 6.4.9.3 “PSECT”, and Section 5.15.1 “Program Sections”), or C code which uses the #pragma psect directive (see 5.14.4.8 “The #pragma psect Directive”). Without this -L option, it would be necessary to invoke the linker manually to allow the linker options to be adjusted. This option can also be used to replace default linker options. If the string starting from the first character after the -L option, up to the first equal character, “=”, matches a psect or class name in the default options, then (the reference to the psect or class name in the default option, and the remainder of that option, are deleted) that default linker option is replaced by the option specified by the -L. For example, if a default linker option was: -preset_vec=00h,intentry,init,end_init the driver option -L-pinit=100h would result in the following options being passed to the linker: -pinit=100h -preset_vec=00h. Note that the end_init linker option has been removed entirely. If there are no characters following the first equal character in the -L option, then no replacement will be made for the default linker options that will be deleted. For example, the driver option -L-pinit= will adjust the default options passed to the linker, as above; but, the -pinit linker option would be removed entirely. No warning is generated if such a default linker option cannot be found. The default option that you are deleting or replacing must contain an equal character. 2012-2013 Microchip Technology Inc. DS50002053D-page 101 MPLAB® XC8 C Compiler User’s Guide 4.8.7 -M: Generate Map File The -M option is used to request the generation of a map file. The map file is generated by the linker and includes detailed information about where objects are located in memory. See Section 7.4 “Map Files” for information regarding the content of these files. If no filename is specified with the option, then the name of the map file will have the project name (see Section 4.3 “The Compilation Sequence”), with the extension .map. This option is on by default when compiling from within MPLAB X IDE and using the Universal Toolsuite. 4.8.8 -N: Identifier Length This option allows the significant C identifier length (used by functions and variables) to be decreased from the default value of 255. Valid sizes for this option are from 32 to 255. The option has no effect for all other values. This option also controls the significant length of identifiers used by the preprocessor, such as macro names. The default length is also 255, and can be adjusted to a minimum of 31. If the --STRICT option is used, the default significant identifier length is reduced to 31. Code that uses a longer identifier length will be less portable. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 4.8.9 -O: Specify Output File This option allows the base name of the output file(s) to be specified. If no -O option is given, the base name of output file(s) will be the same as the project name, see Section 4.3 “The Compilation Sequence”. The files whose names are affected by this option are those files that are not directly associated with any particular source file, such as the HEX file, MAP file and SYM file. The -O option can also change the directory in which the output file is located by including the required path before the filename. This will then also specify the output directory for any files produced by the linker or subsequently run applications. Any relative paths specified are with respect to the current working directory. For example, if the option -Oc:\project\output\first is used, the MAP and HEX file, etc., will use the base name first, and will be placed in the directory c:\project\output. Any extension supplied with the filename will be ignored. If a path is specified with the option that enables MAP file creation, -M, (see Section 4.8.7 “-M: Generate Map File”), this overrides any name or path information provided by -O. To change the directory in which all output and intermediate files are written, use the --OUTDIR option; see Section Section 4.8.43 “--OUTDIR: Specify a Directory for Output Files”. Note that if -O specifies a path that is inconsistent with the path specified in the --OUTDIR option, it will result in an error. DS50002053D-page 102 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.10 -P: Preprocess Assembly Files The -P option causes assembler source files to be preprocessed before they are assembled, thus allowing the use of preprocessor directives, such as #include, and C-style comments with assembler code. By default, assembler files are not preprocessed. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 4.8.11 -Q: Quiet Mode This option places the compiler in a quiet mode that suppresses the Microchip Technology Incorporated copyright notice from being displayed. 4.8.12 -S: Compile to Assembler Code The -S option stops compilation after generating an assembly output file. One assembly file will be generated for all the C source code, including p-code library code. The command: xc8 --CHIP=16F877A -S test.c will produce an assembly file called test.as, which contains the assembly code generated from test.c. The generated file is valid assembly code that could be passed to xc8 as a source file, however this should only be done for exploratory reasons. To take advantage of the benefits of the compilation technology in the compiler, it must compile and link all the C source code in a single step. See the --PASS1 option (Section 4.8.45 “--PARSER: Specify Parser Mode”) to generate intermediate files if you wish to compile code using a two-step process or use intermediate files. This option is useful for checking assembly code output by the compiler. The file produced by this option differs to that produced by the --ASMLIST option (see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”) in that it does not contain op-codes or addresses and it can be used as a source file in subsequent compilations. The assembly list file is more human readable, but is not a valid assembly source file. 4.8.13 -U: Undefine a Macro The -U option, the inverse of the -D option, is used to undefine predefined macros. This option takes the form -Umacro, where macro is the name of the macro to be undefined The option, -Udraft, for example, is equivalent to: #undef draft placed at the top of each module compiled using this option. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 2012-2013 Microchip Technology Inc. DS50002053D-page 103 MPLAB® XC8 C Compiler User’s Guide 4.8.14 -V: Verbose Compile The -V option specifies verbose compilation. When used, the compiler will display the command lines used to invoke each of the compiler applications described in Section 4.3 “The Compilation Sequence”. The name of the compiler application being executed will be displayed, plus all the command-line arguments to this application. This option is useful for confirming options and files names passed to the compiler applications. If this option is used twice (-V -V), it will display the full path to each compiler application as well as the full command-line arguments. This would be useful to ensure that the correct compiler installation is being executed, if there is more than one compiler installed. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 4.8.15 --ADDRQUAL: Set Compiler Response to Memory Qualifiers The --ADDRQUAL option indicates the compiler’s response to non-standard memory qualifiers in C source code. By default, these qualifiers are ignored; i.e., they are accepted without error, but have no effect. Using this option allows these qualifiers to be interpreted differently by the compiler. The near qualifier is affected by this option. On PIC18 devices, this option also affects the far qualifier; and for other 8-bit devices, the bankx qualifiers (bank0, bank1, bank2, etc.) are affected. The suboptions are detailed in Table 4-7. TABLE 4-7: Selection COMPILER RESPONSES TO MEMORY QUALIFIERS Response require The qualifiers will be honored. If they cannot be met, an error will be issued. request The qualifiers will be honored, if possible. No error will be generated if they cannot be followed. ignore The qualifiers will be ignored and code compiled as if they were not used. reject If the qualifiers are encountered, an error will be immediately generated. For example, when using the option --ADDRQUAL=request the compiler will try to honor any non-standard qualifiers, but silently ignore them if they cannot be met. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. DS50002053D-page 104 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.16 --ASMLIST: Generate Assembler List Files The --ASMLIST option tells xc8 to generate assembler listing files for the C and assembly source modules being compiled. One assembly list file is produced for the entire C program, including code from the C library functions. Additionally, one assembly list file is produced for each assembly source file in the project, including the runtime startup code. For more information, see Section 4.4.2 “Startup and Initialization”. Assembly list files use a .lst extension and, due to the additional information placed in these files, cannot be used as assembly source files. In the case of listings for C source code, the list file shows both the original C code and the corresponding assembly code generated by the compiler. See Section 6.5 “Assembly-Level Optimizations”, for full information regarding the content of these files. The same information is shown in the list files for assembly source code. This option is on by default when compiling under MPLAB IDE. 2012-2013 Microchip Technology Inc. DS50002053D-page 105 MPLAB® XC8 C Compiler User’s Guide 4.8.17 --CHECKSUM: Calculate a Checksum This option will perform a checksum over the address range specified and store the result at the destination address specified. The general form of this option is as follows. -CHECKSUM=start-end@destination[,offset=][,width=w][,code=c][,algorithm=a] Additional specifications are appended as a comma-separated list to this option. Such specifications are: width=n selects the width of the checksum result in bytes for non-Fletcher algorithms. A negative width will store the result in little-endian byte order. Result widths from one to four bytes are permitted. offset=nnnn specifies an initial value or offset to be added to this checksum. algorithm=n selects one of the checksum algorithms implemented in HEXMATE. The selectable algorithms are described in Table 8-4. code=nn is a hexadecimal code that will trail each byte in the checksum result. This can allow each byte of the checksum result to be embedded within an instruction. The start, end and destination attributes are, by default, hexadecimal constants. If an accompanying --FILL option has not been specified, unused locations within the specified address range will be filled with FFFh for baseline devices, 3FFFh for mid-range devices, or FFFF for PIC18 devices. This is to remove any unknown values from the equation and ensure the accuracy of the checksum result. For example: --checksum=800-fff@20,width=1,algorithm=2 will calculate a 1 byte checksum from address 0x800 to 0xfff and store this at address 0x20. A 16-bit addition algorithm will be used. See Table 4-8, for the available algorithms. TABLE 4-8: CHECKSUM ALGORITHM SELECTION Selector Algorithm description -4 Subtraction of 32 bit values from initial value -3 Subtraction of 24 bit values from initial value -2 Subtraction of 16 bit values from initial value -1 Subtraction of 8 bit values from initial value 1 Addition of 8 bit values from initial value 2 Addition of 16 bit values from initial value 3 Addition of 24 bit values from initial value 4 Addition of 32 bit values from initial value 7 Fletcher’s checksum (8 bit calculation, 2-byte result width) 8 Fletcher’s checksum (16 bit calculation, 4-byte result width) The checksum calculations are performed by the HEXMATE application. The information in this driver option is passed to the HEXMATE application when it is executed. 4.8.18 --CHIP: Define Device This option must be used to specify the target device, or device, for the compilation. This is the only compiler option that is mandatory when compiling code. To see a list of supported devices that can be used with this option, use the --CHIPINFO option described in Section 4.8.19 “--CHIPINFO: Display List of Supported Devices”. DS50002053D-page 106 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.19 --CHIPINFO: Display List of Supported Devices The --CHIPINFO option displays a list of devices the compiler supports. The names listed are those chips that are defined in the chipinfo file and which can be used with the --CHIP option. Compiler execution will terminate after this list has been printed. 4.8.20 --CLIST: Generate C Listing File Use of this option will generate a C listing file for each C source file specified on the command line. The listing files produced consist of the original C source code prepended with a line number. Do not confuse these files with assembly list files; see Section 4.8.16 “--ASMLIST: Generate Assembler List Files”. 4.8.21 --CODEOFFSET: Offset Program Code to Address In some circumstances, such as bootloaders, it is necessary to shift the program image to an alternative address. This option is used to specify a base address for the program code image and to reserve memory from address 0 to that specified in the option. When using this option, all code psects (including Reset and interrupt vectors and constant data) will be adjusted to the address specified. The address is assumed to be a hexadecimal constant. A leading 0x, or a trailing h hexadecimal specifier can be used, but is not necessary. This option differs from the --ROM option in that it will move the code associated with the Reset and interrupt vectors. That cannot be done using the --ROM option; see Section 4.8.50 “--ROM: Adjust ROM Ranges”. For example, if the option --CODEOFFSET=600 is specified, the Reset vector will be moved from address 0 to address 0x600; the interrupt vector will be moved from address 4 to 0x604, in the case of mid-range PIC devices, or to the addresses 0x608 and 0x618 for PIC18 devices. No code will be placed between address 0 and 0x600. As the Reset and interrupt vector locations are fixed by the PIC device, it is the programmer’s responsibility to ensure code that can redirect control to the offset Reset and interrupt routines is written and located at the original locations. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 2012-2013 Microchip Technology Inc. DS50002053D-page 107 MPLAB® XC8 C Compiler User’s Guide 4.8.22 --DEBUGGER: Select Debugger Type This option is intended for use for compatibility with development tools that can act as a debugger. xc8 supports several debuggers and using this option will configure the compiler to conform to the requirements of that selected. The possible selections for this option are defined in Table 4-9. TABLE 4-9: SELECTABLE DEBUGGERS Suboption Debugger selected none No debugger (default) icd2 MPLAB® ICD 2 icd3 MPLAB ICD 3 pickit2 PICkit™ 2 pickit3 PICkit 3 realice MPLAB REAL ICE™ in-circuit emulator For example: xc8 --CHIP=16F877AA --DEBUGGER=icd2 main.c Choosing the correct debugger is important as they can use memory resources that could otherwise be used by the project if this option is omitted. Such a conflict would lead to runtime failure. If the debugging features of the development tool are not to be used (for example, if the MPLAB ICD 3 is only being used as a programmer), then the debugger option can be set to none, because memory resources are not being used by the tool. MPLAB X IDE will automatically apply this option for debug builds once you have indicated the hardware tool in the Project Preferences. 4.8.23 --DOUBLE: Select Kind of Double Types This option allows the kind of double-precision floating-point types to be selected. By default, the compiler will choose the truncated IEEE754 24-bit implementation for double types. With this option, it can be changed to the full 32-bit IEEE754 implementation. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 4.8.24 --ECHO: Echo Command Line Before Processing Use of this option will result in the driver command line being echoed to the stdout stream before compilation commences. Each token of the command line will be printed on a separate line and they will appear in the order in which they are placed on the command line. DS50002053D-page 108 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.25 --EMI: Select External Memory Interface Operating Mode The external memory interface available on some PIC18 devices can be operated in several modes. The interface can operate in 16-bit modes; Word-write and Byte-select modes or in an 8-bit mode: Byte-write mode. These modes are represented by those specified in Table 4-10. TABLE 4-10: EXTERNAL MEMORY INTERFACE MODES Mode Operation wordwrite 16-bit Word-write mode (default) byteselect 16-bit Byte-select mode bytewrite 8-bit Byte-write mode The selected mode will affect the code generated when writing to the external data interface. In word write mode, dummy reads and writes can be added to ensure that an even number of bytes are always written. In Byte-select or Byte-write modes, dummy reads and writes are not generated and can result in more efficient code. Note that this option does not pre-configure the device for operation in the selected mode. See your device data sheet for the registers and settings that are used to configure the device’s external interface mode. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 2012-2013 Microchip Technology Inc. DS50002053D-page 109 MPLAB® XC8 C Compiler User’s Guide 4.8.26 --ERRATA: Specify Errata Workarounds This option allows specification of software workarounds to documented silicon errata issues. A default set of errata issues apply to each device, but this set can be adjusted by using this option and the arguments presented in Table 4-11. TABLE 4-11: Symbol ERRATA WORKAROUNDS Bit pos. Workaround 4000 0 Program memory accesses/jumps across 4000h address boundary fastints 1 Fast interrupt shadow registers corruption lfsr 2 Broken LFSR instruction minus40 3 Program memory reads at -40 degrees reset 4 GOTO instruction cannot exist at Reset vector bsr15 5 Flag problems when BSR holds value 15 daw 6 Broken DAW instruction eedatard 7 Read EEDAT in immediate instruction after RD set eeadr 8 Don't set RD bit immediately after loading EEADR ee_lvd 9 LVD must stabilize before writing EEPROM fl_lvd 10 LVD must stabilize before writing Flash tblwtint 11 Clear interrupt registers before tblwt instruction fw4000 12 Flash write exe must act on opposite side of 4000h boundary resetram 13 RAM contents can corrupt if async. Reset occurs during write access fetch 14 Corruptible instruction fetch. – apply FFFFh (NOP) at required locations clocksw 15 Code corruption if switching to external oscillator clock source – apply switch to HFINTOSC high-power mode first At present, workarounds are mainly employed for PIC18 devices, but the clocksw errata is only applicable for some enhanced mid-range devices. To disable all software workarounds, use the following. --ERRATA=none To apply the default set of workarounds, but to specifically disable the jump across 4000 errata, for example, use the following. --ERRATA=default,-4000 A preprocessor macro ERRATA_TYPES (see Section 5.14.3 “Predefined Macros”) is set to a value to indicate the errata applied. Each errata listed in Table 4-11 represents a bit position in the macro’s value, with the topmost errata in the table being the least significant. The bit position is indicated in the table and is set if the corresponding errata is applied. That is, if the 4000, reset and bsr15 errata were applied, the value assigned to the ERRATA_TYPES macro would be 0x31. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. DS50002053D-page 110 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.27 --ERRFORMAT: Define Format for Compiler Messages If the --ERRFORMAT option is not used, the default behavior of the compiler is to display any errors in a “human readable” form. This standard format is perfectly acceptable to a person reading the error output, but is not generally usable with environments that support compiler error handling. This option allows the exact format of printed error messages to be specified using special placeholders embedded within a message template. See Section 4.6 “Compiler Messages” for full details of the messaging system employed by xc8, and the placeholders which can be used with this option. This section is also applicable to the --WARNFORMAT and --MSGFORMAT options, which adjust the format of warning and advisory messages, respectively. If you are compiling using MPLAB IDE, the format of the compiler messages is automatically configured to what the IDE expects. It recommended that you do not adjust the message formats if compiling using this IDE. 4.8.28 --ERRORS: Maximum Number of Errors This option sets the maximum number of errors each compiler application, as well as the driver, will display before execution is terminated. By default, up to 20 error messages will be displayed by each application. See Section 4.6 “Compiler Messages” for full details of the messaging system employed by xc8. 4.8.29 --EXT: Specify C Language Extensions The compiler can accept several different sets of non-standard C language extensions. The suboption to --EXT indicates the set and these are shown in Table 4-12. TABLE 4-12: ACCEPTABLE C LANGUAGE EXTENSIONS Suboption Meaning xc8 The native XC8 extensions (default) cci A common C interface acceptable by all MPLAB XC compilers iar Extensions defined by the IAR C/C++ Compiler for ARM Enabling the cci suboption requests the compiler to check all source code and compiler options for compliance with the Common C Interface (CCI). Code that complies with this interface is portable across all MPLAB XC compilers. Code or options that do not conform to the CCI will be flagged by compiler warnings. See Chapter 2. “Common C Interface”, for information on the features that are covered by this interface. The iar suboption enables conformance with the non-standard extensions supported by the IAR C/C++ Compiler™ for ARM. This is discussed further in Appendix B. “Embedded Compiler Compatibility Mode”. 2012-2013 Microchip Technology Inc. DS50002053D-page 111 MPLAB® XC8 C Compiler User’s Guide 4.8.30 --FILL: Fill Unused Program Memory This option allows specification of a hexadecimal opcode that can be used to fill all unused program memory locations. This option utilizes the features of the HEXMATE application, so it is only available when producing a HEX output file, which is the default operation. This driver feature allows you to compile and fill unused locations in one step. If you prefer not to use the driver option and you prefer to fill unused locations after compilation, then you need to use the HEXMATE application. Note that the corresponding option in HEXMATE is -FILL (one leading dash) as opposed to the drivers --FILL option. Note, also, that the unused tag that can be specified with the driver option does not exist in the HEXMATE options. The usage of the driver option is: --FILL=[const_width:]fill_expr[@address[:end_address]] where: • const_width has the form wn and signifies the width (n bytes) of each constant in fill_expr. If const_width is not specified, the default value is the native width of the architecture. That is, --FILL=w1:1 with fill every byte with the value 0x01. • fill_expr can use the syntax (where const and increment are n-byte constants): - const fill memory with a repeating constant; i.e., --FILL=0xBEEF becomes 0xBEEF, 0xBEEF, 0xBEEF, 0xBEEF - const+=increment fill memory with an incrementing constant; i.e., --fill=0xBEEF+=1 becomes 0xBEEF, 0xBEF0, 0xBEF1, 0xBEF2 - const-=increment fill memory with a decrementing constant; i.e., --fill=0xBEEF-=0x10 becomes 0xBEEF, 0xBEDF, 0xBECF, 0xBEBF - const,const,...,const fill memory with a list of repeating constants; i.e., --FILL=0xDEAD,0xBEEF becomes 0xDEAD,0xBEEF,0xDEAD,0xBEEF • The options following fill_expr result in the following behavior: - @unused (or nothing) fill all unused memory with fill_expr; i.e., --FILL=0xBEEF@unused fills all unused memory with 0xBEEF. The driver will expand this to the appropriate ranges and pass these to HEXMATE. - @address fill a specific address with fill_expr; i.e., --FILL=0xBEEF@0x1000 puts 0xBEEF at address 1000h - @address:end_address fill a range of memory with fill_expr; i.e., --FILL=0xBEEF@0:0xFF puts 0xBEEF in unused addresses between 0 and 255 All constants can be expressed in (unsigned) binary, octal, decimal or hexadecimal, as per normal C syntax. For example, 1234 is a decimal value, 0xFF00 is hexadecimal, and FF00 is illegal. See Section 4.9 “MPLAB X Option Equivalents”, or Section 4.9 “MPLAB X Option Equivalents”, for information on using this option in MPLAB IDE. DS50002053D-page 112 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.31 --FLOAT: Select Kind of Float Types This option allows the size of float types to be selected. The types available to be selected are given in Table 4-13. See also, the --DOUBLE option in Section 4.8.23 “--DOUBLE: Select Kind of Double Types”. TABLE 4-13: FLOATING-POINT SELECTIONS Suboption double Effect Size of float matches size of double type 24 24-bit float (default) 32 32-bit float (IEEE754 format) See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 4.8.32 --GETOPTION: Get Command-line Options This option is used to retrieve the command line options that are used for named compiler application. The options are then saved into the given file. This option is not required for most projects, and is disabled when the compiler is operating in Free mode (see Section 4.8.37 “--MODE: Choose Compiler Operating Mode”). The options take an application name and a filename to store the options, for example: --GETOPTION=hlink,options.txt 4.8.33 --HELP: Display Help This option displays information on the xc8 compiler options. The option --HELP will display all options available. To find out more about a particular option, use the option’s name as a parameter. For example: xc8 --help=warn will display more detailed information about the --WARN option, the available suboptions, and which suboptions are enabled by default. 4.8.34 --HTML: Generate HTML Diagnostic Files This option will generate a series of HTML files that can be used to explore the compilation results of the latest build. The files are stored in a directory called html, located in the output directory. The top-level file (which can be opened with your favorite web browser) is called index.html. Use this option at all stages of compilation to ensure files associated with all compilation stages are generated. The index page is a graphical representation of the compilation process. Each file icon is clickable and will open to show the contents of that file. This is available for all intermediate files, and even binary files will open in a human-readable form. Each application icon can also be clicked to show a page that contains information about the application’s options and build results. The list of all preprocessor macros (preprocessor icon) and a graphical memory usage map (Linker icon) provide information that is not otherwise readily accessible. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 2012-2013 Microchip Technology Inc. DS50002053D-page 113 MPLAB® XC8 C Compiler User’s Guide 4.8.35 --LANG: Specify the Language for Messages This option allows the compiler to be configured to produce error, warning and some advisory messages in languages other than English. English is the default language unless this has been changed at installation, or by the use of the --SETUP option. Some messages are only ever printed in English regardless of the language specified with this option. For more information, see Section 4.6.2 “Message Language”. Table 4-14 shows those languages currently supported. TABLE 4-14: SUPPORTED LANGUAGES Suboption Language en, english English (default) fr, french, francais French de, german, deutsch German 4.8.36 --MEMMAP: Display Memory Map This option will display a memory map for the map file specified with this option. The information printed is controlled by the --SUMMARY option, see Section 4.8.58 “--SUMMARY: Select Memory Summary Output Type”, for example: xc8 --memmap=size.map --summary=psect,class,file This option is seldom required, but would be useful if the linker is being driven explicitly; i.e., instead of in the normal way through the command-line driver. This command would display the memory summary that is normally produced at the end of compilation by the driver. 4.8.37 --MODE: Choose Compiler Operating Mode This option selects the basic operating mode of the compiler. The available types are pro, std, and free. For legacy projects, the mode lite is accepted to mean the Free operating mode. A compiler operating in PRO mode uses full optimization and produces the smallest code size. Standard mode uses limited optimizations, and Free mode only uses a minimum optimization level and will produce relatively large code. See Section 5.13 “Optimizations” and Section 6.5 “Assembly-Level Optimizations”, for more information on which optimizations are available in each mode. Only those modes permitted by the compiler license status will be accepted. For example if you have purchased a Standard compiler license, that compiler can be run in Standard or Free mode, but not the PRO mode. See Section 4.9 “MPLAB X Option Equivalents” for use of this option in MPLAB IDE. DS50002053D-page 114 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.38 --MSGDISABLE: Disable Warning Messages This option allows error, warning or advisory messages to be disabled during compilation of a project. The option is passed a comma-separated list of message numbers that are to be disabled. If the number of an error message is specified in this list, it must be followed by :off; otherwise, it will be ignored. If the message list is specified as 0, then all warnings are disabled. For full information on the compiler’s messaging system and use of this option; see Section 4.6 “Compiler Messages”, and; see Section 4.6.5 “Changing Message Behavior”, for other ways to disable messages. 4.8.39 --MSGFORMAT: Set Advisory Message Format This option sets the format of advisory messages produced by the compiler. Warning and error messages are controlled separately by other options. See Section 4.8.27 “--ERRFORMAT: Define Format for Compiler Messages” and Section 4.8.62 “--WARNFORMAT: Set Warning Message Format” for information on changing the format of these sorts of messages. See Section 4.6 “Compiler Messages” for full information on the compiler’s messaging system. If you are compiling using MPLAB IDE, the format of the compiler messages is automatically configured to what the IDE expects. It recommended that you do not adjust the message formats if compiling using this IDE. 4.8.40 --NODEL: Do Not Remove Temporary Files Specifying --NODEL when building will instruct xc8 not to remove the intermediate and temporary files that were created during the build process. 4.8.41 --OBJDIR: Specify a Directory for Intermediate Files This option allows a directory to be nominated in xc8 to locate its intermediate files. If this option is omitted, intermediate files will be created in the current working directory. This option will not set the location of output files, instead use --OUTDIR. See Section 4.8.43 “--OUTDIR: Specify a Directory for Output Files” and Section 4.8.9 “-O: Specify Output File” for more information. 2012-2013 Microchip Technology Inc. DS50002053D-page 115 MPLAB® XC8 C Compiler User’s Guide 4.8.42 --OPT: Invoke Compiler Optimizations The --OPT option allows control of all the compiler optimizers. If this option is not specified, or it is specified as --OPT=all, the space and asm optimizations are enabled (see below). Optimizations can be disabled by using --OPT=none, or individual optimizers can be controlled, for example: --OPT=asm will only enable some assembler optimizations. Table 4-15 lists the available optimization types. TABLE 4-15: OPTIMIZATION OPTIONS Option name Function asm Select optimizations of assembly code derived from C source (default) asmfile Select optimizations of assembly source files debug Favor accurate debugging over optimization speed Favor optimizations that result in faster code space Favor optimizations that result in smaller code (default) all Enable all compiler optimizations none Do not use any compiler optimizations Note that different suboptions control assembler optimizations of assembly source files and intermediate assembly files produced from C source code. The speed and space suboptions are contradictory. Space optimizations are the default. If speed and space suboptions are both specified, then speed optimizations takes precedence. If all optimizations are requested, the space optimization is enabled. These optimizations affect procedural abstraction, which is performed by the assembler, and other optimizations at the code generation stage. The asmfile selection optimizes assembly source files, which are otherwise not optimized by the compiler. By contrast, the asm control allows for optimization of assembly code that was derived from C code, an optimization that is enabled by default. Some compiler optimizations can affect the ability to debug code. Enabling the debug suboption can restrict some optimizations that would otherwise take place, and which would affect debugging. See Section 4.9 “MPLAB X Option Equivalents” for use of this option in MPLAB IDE. DS50002053D-page 116 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.43 --OUTDIR: Specify a Directory for Output Files This option allows a directory to be nominated for xc8 to locate its output files. If this option is omitted, output files will be created in the current working directory. See also Section 4.8.41 “--OBJDIR: Specify a Directory for Intermediate Files” and Section 4.8.9 “-O: Specify Output File” for more information. 4.8.44 --OUTPUT= type: Specify Output File Type This option allows the type of the output file(s) to be specified. If no --OUTPUT option is specified, the output file’s name will be the same as the project name (see Section 4.3 “The Compilation Sequence”). The available output file formats are shown in Table 4-16. More than one output format can be specified by supplying a comma-separated list of tags. Not all formats are supported by Microchip development tools. For debugging, the ELF/DWARF format is preferred, but this format is not supported by MPLAB IDE v8 or early versions of MPLAB X IDE. Before selecting the ELF file output, ensure your IDE version has support for this format. Microchip COFF is the default debugging file output. Output file types that specify library formats stop the compilation process before the final stages of compilation are executed. So, specifying an output file format list that contains, for example: lib or all, will prevent the other formats from being created. TABLE 4-16: OUTPUT FILE FORMATS Type tag File format lib Object library file (for assembly source) lpp P-code library file (for C source) intel, inhx32 Intel HEX (default) inhx032 Intel Hex with initialization of upper extended linear address to zero tek Tektronix Hex aahex American Automation symbolic HEX file mot, motorola, s19 Motorola S19 HEX file bin, binary Binary file mcof, mcoff, mpcoff Microchip COFF (default) elf ELF/DWARF file So, for example: xc8 --CHIP=16F877AA --OUTPUT=lpp lcd_init.c lcd_data.c lcd_msgs.c will compile the three names files into an LPP (p-code) library. 2012-2013 Microchip Technology Inc. DS50002053D-page 117 MPLAB® XC8 C Compiler User’s Guide 4.8.45 --PARSER: Specify Parser Mode This option controls which symbols are stripped from intermediate (p-code) files produced by the parser. The default is to remove unused symbols, which can also be specified using the --PARSER=lean mode. The rich mode will not remove any symbols. Note that the rich mode will generate larger intermediate files and will considerably slow down the compilation, particularly if there are many SFRs defined for a device. PIC18 devices typically have a large number of SFRs. Use the rich setting if you have in-line assembly code that accesses symbols that are not referenced by C code otherwise undefined symbol errors can be produced by the assembler application. 4.8.46 --PASS1: Compile to P-code The --PASS1 option is used to generate p-code intermediate files (.p1 files) from the parser, and then stop compilation. Such files need to be generated if creating p-code library files, however the compiler is able to generate library files in one step, if required. See Section 4.8.44 “--OUTPUT= type: Specify Output File Type” for specifying a library output file type.) 4.8.47 --PRE: Produce Preprocessed Source Code The --PRE option is used to generate preprocessed C source files (also called modules or translation units) with an extension .pre. This can be useful to ensure that preprocessor macros have expanded to what you think they should. Use of this option can also create C source files that do not require any separate header files. If the .pre files are renamed to .c files, they can be passed to the compiler for subsequent processing. This is useful when sending files to a colleague or to obtain technical support without sending all the header files, which can reside in many directories. If you wish to see the preprocessed source for the printf() family of functions, do not use this option. The source for this function is customized by the compiler, but only after the code generator has scanned the project for printf() usage. Thus, as the –-PRE option stops compilation after the preprocessor stage, the code generator will not execute and no printf() code will be processed. If this option is omitted, the preprocessed source for printf() will be automatically retained in the file doprnt.pre. DS50002053D-page 118 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.48 --PROTO: Generate Prototypes The --PROTO option is used to generate .pro files containing both ANSI C and K&R style function declarations for all functions within the specified source files. Each .pro file produced will have the same base name as the corresponding source file. Prototype files contain both ANSI C-style prototypes and old-style C function declarations within conditional compilation blocks. The extern declarations from each .pro file should be edited into a global header file, which can then be included into all the C source files in the project. The .pro files can also contain static declarations for functions that are local to a source file. These static declarations should be edited into the start of the source file. To demonstrate the operation of the --PROTO option, enter the following source code as file test.c: #include add(arg1, arg2) int * arg1; int * arg2; { return *arg1 + *arg2; } void printlist(int * list, int count) { while (count--) printf("d " *list++); putchar(’\n’); } If compiled with the command: xc8 --CHIP=16F877AA --PROTO test.c xc8 will produce test.pro containing the following declarations, which can then be edited as necessary: /* Prototypes from test.c */ /* extern functions - include these in a header file */ #if PROTOTYPES extern int add(int *, int *); extern void printlist(int *, int); #else /* PROTOTYPES */ extern int add(); extern void printlist(); #endif /* PROTOTYPES */ 2012-2013 Microchip Technology Inc. DS50002053D-page 119 MPLAB® XC8 C Compiler User’s Guide 4.8.49 --RAM: Adjust RAM Ranges This option is used to adjust the default RAM, which is specified for the target device. The default memory will include all the on-chip RAM specified for the target PIC10/12/16 device, thus this option only needs be used if there are special memory requirements. Typically this option is used to reserve memory (reduce the amount of memory available). Specifying additional memory that is not in the target device will typically result in a successful compilation, but can lead to code failures at runtime. The default RAM memory for each target device is specified in the chipinfo file, picc.ini. Strictly speaking, this option specifies the areas of memory that can be used by writable (RAM-based) objects; but, not necessarily those areas of memory that contain physical RAM. The output that will be placed in the ranges specified by this option are typically variables that a program defines. For example, to specify an additional range of memory to that already present on-chip, use: --RAM=default,+100-1ff This will add the range from 100h to 1ffh to the on-chip memory. To only use an external range and ignore any on-chip memory, use: --RAM=0-ff This option can also be used to reserve memory ranges already defined as on-chip memory in the chipinfo file. To do this, supply a range prefixed with a minus character, -, for example: --RAM=default,-100-103 will use all the defined on-chip memory, but not use the addresses in the range from 100h to 103h for allocation of RAM objects. This option is also used to specify RAM for far objects on PIC18 devices. These objects are stored in the PIC18 extended memory. Any additional memory specified with this option whose address is above the on-chip program memory is assumed to be extended memory implemented as RAM. For example, to indicate that RAM has been implemented in the extended memory space at addresses 0x20000 to 0x20fff, use the following option. --RAM=default,+20000-20fff This option will adjust the memory ranges used by linker classes; see Section 7.2.1 “-Aclass =low-high,...”; so, any object that is in a psect is placed in this class. Any objects contained in a psect that are explicitly placed at a memory address by the linker (see Section 7.2.19 “-Pspec”) i.e., are not placed into a memory class, are not affected by the option. For an introductory guide to psects, see Section 5.15.1 “Program Sections”. See Section 4.9 “MPLAB X Option Equivalents” for use of this option in MPLAB IDE. DS50002053D-page 120 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.50 --ROM: Adjust ROM Ranges This option is used to change the default ROM that is specified for the target device. The default memory will include all the on-chip ROM specified for the target PIC10/12/16 device, thus this option only needs to be used if there are special memory requirements. Typically this option is used to reserve memory (reduce the amount of memory available). Specifying additional memory that is not in the target device will typically result in a successful compilation, but can lead to code failures at runtime. The default ROM memory for each target device is specified in the chipinfo file, picc.ini. Strictly speaking, this option specifies the areas of memory that can be used by read-only (ROM-based) objects; but, not necessarily those areas of memory that contain physical ROM. When producing code that can be downloaded into a system via a bootloader, the destination memory can be some sort of (volatile) RAM. The output that will be placed in the ranges specified by this option are typically executable code and any data variables that are qualified as const. For example, to specify an additional range of memory to that on-chip, use: --ROM=default,+100-2ff This will add the range from 100h to 2ffh to the on-chip memory. To only use an external range and ignore any on-chip memory, use: --ROM=100-2ff This option can also be used to reserve memory ranges already defined as on-chip memory in the chip configuration file. To do this supply a range prefixed with a minus character, -, for example: --ROM=default,-100-1ff will use all the defined on-chip memory, but not use the addresses in the range from 100h to 1ffh for allocation of ROM objects. This option will adjust the memory ranges used by linker classes, see Section 7.2.1 “-Aclass =low-high,...”; so, any object that is in a psect is placed in this class. Any objects which are contained in a psect that are explicitly placed at a memory address by the linker (see Section 7.2.19 “-Pspec”), i.e., are not placed into a memory class, are not affected by the option. For an introductory guide to psects, see Section 5.15.1 “Program Sections”. Note that some psects must be linked above a threshold address, most notably some psects that hold const data. Using this option to remove the upper memory ranges can make it impossible to place these psects. See Section 4.9 “MPLAB X Option Equivalents” for use of this option in MPLAB IDE. 2012-2013 Microchip Technology Inc. DS50002053D-page 121 MPLAB® XC8 C Compiler User’s Guide 4.8.51 --RUNTIME: Specify Runtime Environment The --RUNTIME option is used to control what is included as part of the runtime environment. The runtime environment encapsulates any code that is present at runtime that has not been defined by the user, instead supplied by the compiler, typically as library code or compiler-generated source files. All required runtime features are enabled by default and this option is not required for normal compilation. Note that the code that clears or initializes variables, which is included by default, will clobber the contents of the STATUS register. For mid-range and baseline devices, if you need to check the cause of Reset using the TO or PD bits in this register, then you must enable the resetbits suboption as well. See Section 5.10.1.4 “STATUS Register Preservation”, for how this feature is used. The usable suboptions include those shown in Table 4-17. TABLE 4-17: RUNTIME ENVIRONMENT SUBOPTIONS Suboption init Controls the code present in the main program module that copies the ROM-image of initial values to RAM variables On (+) Implies The ROM image is copied into RAM and initialized variables will contain their initial value at main(). Default State On clib the inclusion of library files into the output Library files are linked into the output. code by the linker. On clear the code present in the main program module that clears uninitialized variables Uninitialized variables are cleared and will contain 0 at main(). On config programming the device with default config bytes Configuration bits not specified will be assigned a default value. (PIC18 only) Off download conditioning of the Intel HEX file for use with bootloaders Data records in the Intel HEX file are padded out to 16-byte lengths and will align on 16-byte boundaries. Startup code will not assume Reset values in certain registers. Off flp additional code to provide function pro filing Diagnostic code will be embedded into the output to allow function profiling. Off no_startup whether the startup module is linked in with user-defined code Startup module will not be linked in. osccal initialize the oscillator with the oscillator constant Oscillator will be calibrated (PIC10/12/16 only). oscval: value set the internal clock oscillator calibration Oscillator will be calibrated with value supvalue plied (PIC10/12/16 only). n/a keep whether the startup module source file (startup.as) is deleted after compilation On plib whether the peripheral library is linked in. The peripheral library will be linked in to the build (PIC18 only). resetbits Preserve Power-down and Time-out STATUS bits at start up stackcall The startup module is not deleted. STATUS bits are preserved (PIC10/12/16 only). Allow function calls to use a table look-up Functions called via CALL instruction while the stack was not exhausted, then called via a method after the hardware stack has look-up table (PIC10/12/16 devices only). filled. (Ignored if reentrant or hybrid function model is used, or if function uses the reentrant specifier.) Off On On Off Off See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. DS50002053D-page 122 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.8.52 --SCANDEP: Scan for Dependencies When this option is used, .dep and .d dependency files are generated. The dependency file lists those files on which the source file is Dependant. Dependencies result when one file is #included into another. The .d file format is used by GCC-based compilers and it contains the same information as the .dep file. Compilation will stop after the preprocessing stage if this option is used. 4.8.53 --SERIAL: Store a Value at this Program Memory Address This option allows a hexadecimal code to be stored at a particular address in program memory. A typical task for this option might be to position a serial number in program memory. The byte-width of data to store is determined by the byte-width of the hexcode parameter in the option. For example, to store a one-byte value, 0, at program memory address 1000h, use --SERIAL=00@1000. To store the same value as a four byte quantity use --SERIAL=00000000@1000. This option is functionally identical to the corresponding HEXMATE option. For more detailed information and advanced controls that can be used with this option, refer to Section 8.3.1.15 “-SERIAL”. The driver will also define a label at the location where the value was stored, and which can be referenced from C code as _serial0. To enable access to this symbol, remember to declare it, for example: extern const int _serial0; See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 4.8.54 --SETOPTION: Set the Command-line Options for Application This option is used to supply alternative command line options for the named application when compiling. The general form of this option is shown below. --SETOPTION=app,file where the app component specifies the application that will receive the new options, and the file component specifies the name of the file that contains the additional options that will be passed to the application. This option is not required for most projects. If specifying more than one option to a component, each option must be entered on a new line in the option file. This option can also be used to remove an application from the build sequence. If the file parameter is specified as off, execution of the named application will be skipped. In most cases, this is not desirable as almost all applications are critical to the success of the build process. Disabling a critical application will result in catastrophic failure. However, it is permissible to skip a non-critical application such as CLIST or HEXMATE, if the final results are not relying on their function. 2012-2013 Microchip Technology Inc. DS50002053D-page 123 MPLAB® XC8 C Compiler User’s Guide 4.8.55 --SHROUD: Obfuscate P-code Files This option should be used in situations where either p-code files or p-code libraries are to be distributed and are built from confidential source code. C comments, which are normally included into these files, as well as line numbers and variable names will be removed, or obfuscated, so that the original source code cannot be reconstructed from the distributed files. 4.8.56 --STACK: Specify Data Stack Type For Entire Program This option allows selection of the stack type to be used by a program’s stack-based (auto and parameter) variables. The data stacks available are called a compiled stack and a software stack, and they are described in Section 5.3.4.2 “Data Stacks”. The stack types that can be used with this option are described in Table 4-18. TABLE 4-18: --STACK SUBOPTIONS Stack types Default Allocation for Stack-based Variables compiled or nonreentrant Use the compiled stack for all functions; functions are non-reentrant (default). software or reentrant Use the software stack for eligible functions and devices; such functions are reentrant. hybrid Use the compiled stack for functions not called reentrantly; use the software stack for all other eligible functions and devices; functions are only reentrant if required. Suboptions that specify reentrancy only affect target devices that support a software stack. Functions encoded for baseline and mid-range devices always use the compiled stack. In addition, not all functions can use a software stack. Interrupt functions must use the compiled stack, but functions they call may use the software stack. The hybrid setting forces the compiler to consider both a compiled and software stack for the program’s stack-based variables. The software stack will only be used if the functions and device supports reentrancy. This mode allows for reentrancy, when required, but takes advantage of the efficiency of the compiled stack for the majority of the program’s functions. A function is compiled to use the software stack if it is called reentrantly in the program; otherwise, it will use a compiled stack. Any of these option settings can be overridden for individual functions by using function specifiers, described in Section 5.8.1.3 “Reentrant and nonreentrant Specifiers”. Note: Use the software (reentrant) setting with caution. The maximum runtime size of the software stack is not accurately known at compile time, so the compiler cannot warn of memory overwrites. The stack can overflow and corrupt objects or data memory used by something outside the program (such as hardware or another independently-compiled applications). When all functions are forced to use the software stack, the stack size will increase substantially. In addition to the stack type, this option can be used to specify the maximum size of memory reserved by the compiler for the software stack. This option configuration only affects the software stack; there are no controls for the size of the compiled stack. Distinct memory areas are allocated for the software stack used by main-line code and each interrupt function. In essence there are separate stacks for each interrupt and main-line code, but this is transparent at the program level. The compiler automatically manages the allocation of memory to each stack. DS50002053D-page 124 2012-2013 Microchip Technology Inc. XC8 Command-line Driver You can manually specify the maximum space allocated for each stack by following the stack type with a colon-separated list of decimal values, each value being the maximum size, in bytes, of the memory to be reserved. The sizes specified correspond to the main-line code, the lowest priority interrupt through the highest priority interrupt. (PIC18 devices have two separate interrupts; other devices have only one.) Alternatively, you can explicitly state you have no size preference by using a size of auto or leave the numeric field blank. For PIC18 devices, the following example: --STACK=reentrant:auto:30:50 will arrange the stack starting locations so that the low-priority interrupt stack can grow to, at most, 30 bytes (before overflow); the high-priority interrupt stack can grow to, at most, 50 bytes (before overflow); and the main-line code stack can consume the remainder of the free memory that can be allocated to the stack (before overflow). If you do specify the stack sizes using this option, each size must be specified numerically or you can use the auto token. Do not leave a size field empty. If you try to use this option to allocate more stack memory than is available, a warning is issued and only the available memory will be utilized. 4.8.57 --STRICT: Strict ANSI Conformance The --STRICT option is used to enable strict ANSI C conformance of all special, non-standard keywords. The MPLAB XC8 C compiler supports various special keywords (for example the persistent type qualifier). If the --STRICT option is used, these keywords are changed to include two underscore characters at the beginning of the keyword (for example, __persistent) so as to strictly conform to the ANSI standard. Thus if you use this option, you will need to use the qualifier __persistent in your code, not persistent. Be warned that use of this option can cause problems with some standard header files (e.g., ) as they contain special keywords. See Section 4.9 “MPLAB X Option Equivalents”, for use of this option in MPLAB IDE. 4.8.58 --SUMMARY: Select Memory Summary Output Type Use this option to select the type of memory summary that is displayed after compilation. By default, or if the mem suboption is selected, a memory summary is shown. This shows the total memory usage for all memory spaces. A psect summary can be shown by enabling the psect suboption. This shows individual psects, after they have been grouped by the linker, and the memory ranges they cover. Table 4-19 shows what summary types are available. The output printed when compiling normally corresponds to the mem setting. TABLE 4-19: MEMORY SUMMARY SUBOPTIONS Suboption Controls psect A summary of psect names and the addresses where they were linked will be shown. mem A concise summary of memory used will be shown. (default) class A summary of all classes in each memory space will be shown. hex A summary of addresses and HEX files that make up the final output file will be shown. file Summary information will be shown on screen and saved to a file. See Section 4.9 “MPLAB X Option Equivalents” for use of this option in MPLAB IDE. 2012-2013 Microchip Technology Inc. DS50002053D-page 125 MPLAB® XC8 C Compiler User’s Guide 4.8.59 --TIME: Report Time Taken For Each Phase of Build Process Adding the --TIME option when building generates a summary that shows how much time it took for each stage of the build process to complete. 4.8.60 --VER: Display the Compiler’s Version Information The --VER option will display what version of the compiler is running and then exit the compiler. 4.8.61 --WARN: Set Warning Level The --WARN option is used to set the compiler warning level threshold. Allowable warning levels range from -9 to 9. The warning level determines how pedantic the compiler is about dubious type conversions and constructs. Each compiler warning has a designated warning level; the higher the warning level, the more important the warning message. If the warning message’s warning level exceeds the set threshold, the warning is printed by the compiler. The default warning level threshold is 0 and will allow all normal warning messages. Use this option with care as some warning messages indicate code that is likely to fail during execution, or compromise portability. Warning message can be individually disabled with the --MSGDISABLE option, see Section 4.8.38 “--MSGDISABLE: Disable Warning Messages”. See also Section 4.6 “Compiler Messages” for full information on the compiler’s messaging system. See Section 4.9 “MPLAB X Option Equivalents” for use of this option in MPLAB IDE. 4.8.62 --WARNFORMAT: Set Warning Message Format This option sets the format of warning messages produced by the compiler. See Section 4.8.27 “--ERRFORMAT: Define Format for Compiler Messages” for more information on this option. For full information on the compiler’s messaging system, see Section 4.6 “Compiler Messages”. If you are compiling using MPLAB IDE, the format of the compiler messages is automatically configured to what the IDE expects. It recommended that you do not adjust the message formats if compiling using this IDE. DS50002053D-page 126 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.9 MPLAB X OPTION EQUIVALENTS When compiling under the MPLAB X IDE, it is still the compiler’s command-line driver that is being executed and compiling the program. The MPLAB XC8 compiler plugins control the MPLAB X IDE Properties dialog that is used to access the compiler options, however these graphical controls ultimately adjust the command-line options passed to the command-line driver when compiling. You can see the command-line options being used when building in MPLAB X IDE in the Output window. The following dialogs and descriptions identify the mapping between the MPLAB X IDE dialog controls and command-line options. Click any option to see online help and examples shown in the Option Description field in the lower part of the Project Properties dialog. 4.9.1 Global Category The options in the panel in this category control the final output of the compiler. See Figure 4-5 in conjunction with the following command-line option equivalent. FIGURE 4-5: GLOBAL OPTIONS 1 Output file format This selector specifies the output source-level debug format that will be used by debuggers; see Section 4.8.44 “--OUTPUT= type: Specify Output File Type”. 4.9.2 Compiler Category The panels in this category control aspects of compilation of C source. 4.9.2.1 PREPROCESSING AND MESSAGES These options relate to the C preprocessor and messages produced by the compiler; see Section 4.6 “Compiler Messages” for more information. See Figure 4-6 in conjunction with the following command-line option equivalents. 2012-2013 Microchip Technology Inc. DS50002053D-page 127 MPLAB® XC8 C Compiler User’s Guide FIGURE 4-6: PREPROCESSING AND MESSAGES OPTIONS 1 2 3 4 5 6 7 8 9 10 11 1. Define macros The button and field on this line can be used to define preprocessor macros; see Section 4.8.2 “-D: Define Macro”. 2. Undefine macros The button and field on this line can be used to undefine preprocessor macros; see Section 4.8.13 “-U: Undefine a Macro”. 3. Preprocess assembly files This checkbox controls whether assembly source files are scanned by the preprocessor; see Section 4.8.10 “-P: Preprocess Assembly Files”. 4. Identifier length Not implemented; see Section 4.8.8 “-N: Identifier Length”. 5. Include directories This selection uses the buttons and fields grouped in the bracket to specify include (header) file search directories; see Section 4.8.4 “-I: Include Search Path”. 6. Strict ANSI Conformance This forces the compiler to reject any non-standard keywords; see Section 4.8.56 “--STACK: Specify Data Stack Type For Entire Program”. 7. Verbose This checkbox controls whether the full command lines for the compiler applications are displayed when building; see Section 4.8.14 “-V: Verbose Compile”. 8. Warning level This selector allows the warning level print threshold to be set; see Section 4.8.61 “--WARN: Set Warning Level”. 9. Use CCI Syntax This option indicates that the compiler should use the Common C Interface compiler extensions; see Section 4.8.29 “--EXT: Specify C Language Extensions”, and Chapter 2. “Common C Interface”, for more information. DS50002053D-page 128 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 10. Use IAR Syntax This option indicates that the compiler should use the IAR compiler extensions; see Section 4.8.29 “--EXT: Specify C Language Extensions” for more information. 11. Generate the ASM listing file This option indicates that the compiler should generate an assembly listing file. This file should be used to examine the assembly code produced by the compiler; see Section 4.8.16 “--ASMLIST: Generate Assembler List Files”. 4.9.2.2 OPTIMIZATIONS These options, shown in Figure 4-7, relate to optimizations performed by the compiler, 1. Optimization controls These controls adjust the optimizations employed by the compiler (see Section 4.8.42 “--OPT: Invoke Compiler Optimizations”). The Optimization set widget, if present, is not used; and, you can choose any setting for this. Select the custom options below (--OPT suboptions: asm, asmfile, speed/space and debug). The Speed checkbox indicates your preference for any applicable optimizations to be focused on speed or space. 2. Address qualifiers This selector allows the user to select the behavior of the address qualifiers; see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”. 3. Operation mode This selector allows the user to force another available operating mode (free, std, or pro) other than the default; see Section 4.8.37 “--MODE: Choose Compiler Operating Mode”. The operating mode will affect OCG-optimizations FIGURE 4-7: OPTIMIZATIONS OPTIONS 1 { 2 3 2012-2013 Microchip Technology Inc. DS50002053D-page 129 MPLAB® XC8 C Compiler User’s Guide 4.9.3 Linker Category The options in this dialog control the aspects of the second stage of compilation including code generation and linking. 4.9.3.1 RUNTIME All the widgets in Figure 4-8 correspond to suboptions of the --RUNTIME option, see Section 4.8.51 “--RUNTIME: Specify Runtime Environment”. Respectively, these map to the clear, init, keep, no_startup, osccal, oscval, resetbits, download, stackcall, config, clib and plib suboptions of the --RUNTIME option. FIGURE 4-8: DS50002053D-page 130 RUNTIME OPTIONS 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.9.3.2 MEMORY MODEL The panel in this category, shown in Figure 4-9, controls settings that apply to the entire project. FIGURE 4-9: MEMORY MODEL OPTIONS 1 2 3 4 5 1. Size of Double This selector allows the size of the double type to be selected; see Section 4.8.23 “--DOUBLE: Select Kind of Double Types”. 2. Size of Float This selector allows the size of the float type to be selected; see Section 4.8.31 “--FLOAT: Select Kind of Float Types”. 3. External memory This option allows specification of how external memory access is performed. This only affects those devices that can access external memory; see Section 4.8.25 “--EMI: Select External Memory Interface Operating Mode”. 4. RAM ranges This field allows the default RAM (data space) memory used to be adjusted; see Section 4.8.49 “--RAM: Adjust RAM Ranges”. 5. ROM ranges This field allows the default ROM (program memory space) memory used to be adjusted; see Section 4.8.50 “--ROM: Adjust ROM Ranges”. 2012-2013 Microchip Technology Inc. DS50002053D-page 131 MPLAB® XC8 C Compiler User’s Guide 4.9.3.3 FILL FLASH MEMORY All the controls shown in Figure 4-10 relate to options associated with filling unused program memory. See Section 4.8.30 “--FILL: Fill Unused Program Memory”, for more information on the different fields. FIGURE 4-10: FILL FLASH MEMORY OPTIONS 1 DS50002053D-page 132 2012-2013 Microchip Technology Inc. XC8 Command-line Driver 4.9.3.4 ADDITIONAL The options shown in Figure 4-11 relate to miscellaneous options. 1. Extra linker options This field allows you to enter additional options to the linker. Enter the entire -Loption in this field, for example -L-pmytest=300h. See Section 4.8.6 “-L-: Adjust Linker Options Directly”. 2. Serial This option allows you to specify a string that can be inserted into your output HEX file. See Section 4.8.53 “--SERIAL: Store a Value at this Program Memory Address”, for details. 3. Codeoffset This field allows an offset for the program to be specified; see Section 4.8.20 “--CLIST: Generate C Listing File”. 4. Checksum This field allows the checksum specification to be specified; see Section 4.8.17 “--CHECKSUM: Calculate a Checksum”. 5. Errata This allows customization of the errata workarounds applied by the compiler; see Section 4.8.26 “--ERRATA: Specify Errata Workarounds”. 6. Trace type Not implemented. Native trace supported. 7. Extend address 0 in HEX file This option specifies that the Intel HEX file should have initialization to zero of the upper address; see Section 4.8.44 “--OUTPUT= type: Specify Output File Type”. 8. Use response file to link This option allows a file name to be specified. The file must contain command-line options which are then used by MPLAB XC8 during the link step and in preference to the other link-step settings in the project properties; see Section 4.2.1.1 “Long Command Lines”. This option is only relevant when running MPLAB X IDE under Windows. FIGURE 4-11: ADDITIONAL OPTIONS 1 3 5 7 2 4 6 8 2012-2013 Microchip Technology Inc. DS50002053D-page 133 MPLAB® XC8 C Compiler User’s Guide 4.9.3.5 REPORTING These options, shown in Figure 4-12 relate to information produced during and after compilation. 1. Display memory usage after compilation These checkboxes allow you to specify what information is displayed after compilation. The correspond to the psect, class, mem and hex suboptions to the --SUMMARY option; see Section 4.8.58 “--SUMMARY: Select Memory Summary Output Type”. 2. Create summary file Selecting this checkbox will send the information you have selected above to a file, as well as to the standard output. This corresponds to the file suboption to the --SUMMARY option; see Section 4.8.58 “--SUMMARY: Select Memory Summary Output Type”. 3. Create html files This will create HTML files summarizing the previous build; see Section 4.8.34 “--HTML: Generate HTML Diagnostic Files”. FIGURE 4-12: REPORTING OPTIONS 1 { 2 3 DS50002053D-page 134 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Chapter 5. C Language Features 5.1 INTRODUCTION MPLAB XC8 C Compiler supports a number of special features and extensions to the C language which are designed to ease the task of producing ROM-based applications for 8-bit PIC devices. This chapter documents the special language features which are specific to these devices. • • • • • • • • • • • • • • 5.2 ANSI C Standard Issues Device-Related Features Supported Data Types and Variables Memory Allocation and Access Operators and Statements Register Usage Functions Interrupts Main, Runtime Startup and Reset Library Routines Mixing C and Assembly Code Optimizations Preprocessing Linking Programs ANSI C STANDARD ISSUES This compiler conforms to the ISO/IEC 9899:1990 Standard for programming languages. This is commonly called the C90 Standard. It is referred to as the ANSI C Standard in this manual. Some violations to the ANSI C Standard are indicated below in Section 5.2.1 “Divergence from the ANSI C Standard”. Some features from the later standard C99 are also supported. 5.2.1 Divergence from the ANSI C Standard The C language implemented on MPLAB XC8 C Compiler can diverge from the ANSI C Standard in several areas. Due to limited memory and no hardware implementation of a data stack, recursion is not supported and functions are not reentrant on baseline and some mid-range devices. Functions can be encoded reentrantly for enhanced mid-range and PIC18 devices. See Section 5.3.4 “Stacks” for more information on the stack models used by the compiler for each device family. For those devices that do not support reentrancy, the compiler can make functions called from main-line and interrupt code appear to be reentrant via a duplication feature. See Section 5.9.6 “Function Duplication”. 2012-2013 Microchip Technology Inc. DS50002053D-page 135 MPLAB® XC8 C Compiler User’s Guide You cannot reliably use the C sizeof operator with pointer types; however, this operator may be used with pointer variable identifiers. This is a result of the dynamic size of pointers assigned by the compiler. So for the following code: char * size_t size = size = cp; size; sizeof(char *); sizeof(cp); size in the first example will be assigned the maximum size a pointer can be for the particular target device you have chosen. In the second example, size will be assigned the actual size of the pointer variable, cp. 5.2.2 Implementation-Defined behavior Certain features of the ANSI C standard have implementation-defined behavior. This means that the exact behavior of some C code can vary from compiler to compiler. The exact behavior of the compiler is detailed throughout this manual, and is fully summarized in Appendix D. “Implementation-Defined Behavior”. 5.2.3 Common C Interface Standard This compiler conforms to the Microchip XC compiler Common C Interface standard (CCI). This is a further refinement of the ANSI standard that attempts to standardize implementation-defined behavior and non-standard extensions across the entire MPLAB XC compiler family. It is described in Chapter 2. “Common C Interface”. If you choose to write code which conforms to this standard, a compiler option (see Section 4.8.29 “--EXT: Specify C Language Extensions”) should be enabled. This will to indicate that the compiler should enforce conformance. Alternatively, you can continue to write code using the non-standard ANSI extensions provided by the compiler. DS50002053D-page 136 2012-2013 Microchip Technology Inc. C Language Features 5.3 DEVICE-RELATED FEATURES MPLAB XC8 has several features which relate directly to the 8-bit PIC architectures and instruction sets. These are detailed in the following sections. 5.3.1 Device Support MPLAB XC8 C Compiler aims to support all 8-bit PIC devices. However, new devices in these families are frequently released. There are several ways you can check whether the compiler you are using supports a particular device. From the command line, run the compiler you wish to use and pass it the option --CHIPINFO (See Section 4.8.19 “--CHIPINFO: Display List of Supported Devices”). A list of all devices will be printed. If you use the -V option in addition to the --CHIPINFO option, more detailed information my be shown about each device. You can also see the supported devices in your favorite web browser. Open the files pic_chipinfo.html for a list of all supported baseline or mid-range device, or pic18_chipinfo.html for all PIC18 devices. Both these files are located in the DOCS directory under your compiler’s installation directory. 5.3.2 Instruction Set Support The compiler support all instruction sets for PIC10/12/16 devices as well as the standard (legacy) PIC18 instruction set. The extended instruction mode available on some PIC18 devices is not currently supported. Ensure you set the Configuration bits to use the PIC18 legacy instruction mode when appropriate. 5.3.3 Device Header Files There is one header file that is typically included into each C source file you write. The file is and is a generic header file that will include other device- and architecture-specific header files when you build your project. Inclusion of this file will allow access to SFRs via special variables, as well as macros which allow special memory access or inclusion of special instructions, like CLRWDT. Legacy projects can continue to use the header file. Avoid including chip-specific header files into your code as this will reduce portability. The header files shipped with the compiler are specific to that compiler version. Future compiler versions can ship with modified header files. If you copy compiler header files into your project, particularly if you modify these files, be aware that they cannot be compatible with future versions of the compiler. For information about assembly include files (.inc), see Section 5.12.3.2 “Accessing Registers from Assembly Code”. 2012-2013 Microchip Technology Inc. DS50002053D-page 137 MPLAB® XC8 C Compiler User’s Guide 5.3.4 Stacks Stacks are used for two different purposes by programs running on 8-bit devices: one stack is for storing function return addresses, and one or two other stacks are used for data allocation. 5.3.4.1 FUNCTION RETURN ADDRESS STACK The 8-bit PIC devices use what is referred to in this user’s guide as a hardware stack. This stack is limited in depth and cannot be manipulated directly. It is only used for function return addresses and cannot be used for program data. You must ensure that the maximum hardware stack depth is not exceeded; otherwise, code can fail. Nesting function calls too deeply will overflow the stack. It is important to take into account implicitly called library functions and interrupts, which also use levels of the stack. The compiler can be made to manage stack usage for some devices using the stackcall suboption to the --RUNTIME compiler option, see Section 4.8.51 “--RUNTIME: Specify Runtime Environment”. This enables an alternate means of calling functions to prevent stack overflow. A call graph is provided by the code generator in the assembler list file, see Section 6.6.6 “Call Graph”. This will indicate the stack levels at each function call and can be used as a guide to stack depth. The code generator can also produce warnings if the maximum stack depth is exceeded. The warnings and call graphs are guides to stack usage. Optimizations and the use of interrupts can decrease or increase the program’s stack depth over that determined by the compiler. 5.3.4.2 DATA STACKS The compiler can implement two types of data stack: a compiled stack and a software stack. Both these stacks are for storing stack-based variables, such as a function’s auto, parameter, and temporary variables. Either one or both of these types of stacks may be used by a program. Compiler options, specifiers, and how the functions are called will dictate which stacks are used. See Section 5.5.2.2 “Auto Variable Allocation and access”, for more information on how the compiler allocates a function’s stack-based objects. A compiled stack is a static allocation of memory for stack-based objects that can be built up in multiple data banks. See Section 5.5.2.2.1 “Compiled Stack Operation”, for information about how objects are allocated to this stack. Objects in the stack are in fixed locations and can be accessed using an identifier (hence it is a static allocation). Thus, there is no stack pointer. The size of the compiled stack is known at compile time, and so available space can be confirmed by the compiler. The compiled stack is allocated to psects that use the basename cstack; for example, cstackCOMMON, cstackBANK0. See Section 5.15.2 “Compiler-Generated Psects”, for more information on the naming convention for compiler-generated psects. By contrast, the software stack has a size that is dynamic and varies as the program is executed. The maximum size of the stack is not exactly known at compile time and the compiler typically reserves as much space as possible for the stack to grow during program execution. The stack is always allocated a single memory range, which may cross bank boundaries, but within this range it may be segregated into one area for main-line code and an area for each interrupt routine, if required. A stack pointer is used to indicate the current position in the stack. This pointer is permanently allocated to FSR1. A psect is used as a placeholder to reserve the memory used by the stack. This psect is called stack. DS50002053D-page 138 2012-2013 Microchip Technology Inc. C Language Features 5.3.5 Configuration Bit Access The PIC devices have several locations which contain the Configuration bits or fuses. These bits specify fundamental device operation, such as the oscillator mode, watchdog timer, programming mode and code protection. Failure to correctly set these bits can result in code failure, or a non-running device. These bits can be set using a configuration pragma. The pragma has the following forms. #pragma config setting = state|value #pragma config register = value where setting is a configuration setting descriptor, e.g., WDT, and state is a textual description of the desired state, e.g., OFF. The value field is a numerical value that can be used in preference to a descriptor. Consider the following examples. #pragma config WDT = ON #pragma config WDTPS = 0x1A // turn on watchdog timer // specify the timer postscale value One pragma can be used to program several settings by separating each setting-value pair with a comma. For example, the above could be specified with one pragma, as in the following. #pragma config WDT=ON, WDTPS = 0x1A The setting-value pairs can also be quoted to ensure that the preprocessor does not perform substitution of these tokens, for example: #pragma config "BOREN=OFF" Without the quotes and with the preprocessor macro OFF defined, for example, substitution within the pragma would take place. You should never assume that the OFF and ON tokens used in configuration macros equate to 0 and 1, respectively, as that is often not the case. Rather than specify individual settings, the entire register can be programmed with one numerical value, for example: #pragma config CONFIG1L = 0x8F The upper and lower half of each register must be programmed separately. The settings and values associated with each device can be determined from an HTML guide. Open the file pic_chipinfo.html or pic18_chipinfo.html, which are located in the DOCS directory of your compiler installation. Click the link to your target device, and the page will show you the settings and values that are appropriate with this pragma. Check your device data sheet for more information. 2012-2013 Microchip Technology Inc. DS50002053D-page 139 MPLAB® XC8 C Compiler User’s Guide 5.3.5.1 CONFIGURATION BIT LEGACY SUPPORT You can continue to use the configuration macros for legacy projects, but use the pragma for new projects. The compiler supports the __CONFIG and __PROG_CONFIG macros, which allow configuration bit symbols or a Configuration Word value, respectively, to be specified, for example: #include __CONFIG(WDTDIS & HS & UNPROTECT); For PIC10/12/16 devices that have more than one Configuration Word, each subsequent invocation of __CONFIG() will modify the next Configuration Word in sequence. When using the legacy macros for these devices, the order of the macros must match the order of the Configuration Words. Typically this might look like: #include __CONFIG(WDTDIS & XT & UNPROTECT); // Program config. word 1 __CONFIG(FCMEN); The __CONFIG macro used for PIC18 devices takes an additional argument being the number of the Configuration Word location. For example: __CONFIG(2, BW8 & PWRTDIS & WDTPS1 & WDTEN); // specify symbols If you want to use a literal value to program the entire Configuration Word, you must use the __PROG_CONFIG macro. For PIC10/12/16 devices, that might appear as follows: __PROG_CONFIG(0xFFFA); and with PIC18 devices, you must again specify the word being programmed, as in the following: __PROG_CONFIG(1, 0xFE57); // specify a literal constant value You cannot use the setting symbols in the __PROG_CONFIG macro, nor can you use a literal value in the __CONFIG macro. The configuration locations do not need to be programmed in order, except as noted above for multi-word PIC10/12/16 devices using the legacy macros. To use the legacy macros, ensure you include in your source file. Symbols for the macros can be found in the .cfgmap files contained in the dat/cfgmap directory of your compiler installation. 5.3.5.2 CONFIGURATION CONSIDERATIONS Neither the config pragma nor the __CONFIG macro produce executable code. They should both be placed outside function definitions so as not to interfere with the operation of the function’s code. MPLAB X IDE does not allow the Configuration bits to be adjusted. They must be specified in your source code using the pragma (or legacy macro). All the bits in the Configuration Words should be programmed to prevent erratic program behavior. Do not leave them in their default/unprogrammed state. Not all Configuration bits have a default state of logic high; some have a logic low default state. Consult your device data sheet for more information. DS50002053D-page 140 2012-2013 Microchip Technology Inc. C Language Features 5.3.6 Using SFRs From C Code The Special Function Registers (SFRs) are registers which control aspects of the MCU operation or that of peripheral modules on the device. Most of these registers are memory mapped, which means that they appear at, and can be accessed using, specific addresses in the device’s data memory space. Individual bits within some registers control independent features. Some registers are read-only; some are write-only. See your device data sheet for more information. Memory-mapped SFRs are accessed by special C variables that are placed at the address of the register. (Variables that are placed at specific addresses are called absolute variables and are described in Section 5.5.4 “Absolute Variables”.) These variables can be accessed like any ordinary C variable so that no special syntax is required to access SFRs. The SFR variables are predefined in header files and are accessible once you have included the header file (see Section 5.3.3 “Device Header Files”) into your source code. Both bit variables and structures with bit-fields are defined, so you can use either of them in your source code to access bits within a register. The names given to the C variables that map over registers and bits within those registers are based on the names specified in the device data sheet. However, as there can be duplication of some bit names within registers, there can be differences in the nomenclature. The names of the structures that hold the bit-fields will typically be those of the corresponding register followed by bits. For example, the following shows code that includes the generic header file, clears PORTA as a whole, sets bit 0 of PORTA using a bit variable and sets bit 2 of PORTA using the structure/bit-field definitions. #include void main(void) { PORTA = 0x00; RA0 = 1; PORTAbits.RA2 = 1; } To confirm the names that are relevant for the device you are using, check the device-specific header file that will include for the definitions of each variable. These files will be located in the include directory of the compiler and will have a name that represents the device. There is a one-to-one correlation between device and header file name that will be included by , e.g., when compiling for a PIC16LF1826 device, will include the header file . Remember that you do not need to include this chip-specific file into your source code; it is automatically included by . Care should be taken when accessing some SFRs from C code or from assembly in-line with C code. Some registers are used by the compiler to hold intermediate values of calculations, and writing to these registers directly can result in code failure. The compiler does not detect when SFRs have changed as a result of C or assembly code that writes to them directly. The list of registers used by the compiler and further information can be found in Section 5.7 “Register Usage”. SFRs associated with peripherals are not used by the compiler to hold intermediate results and can be changed as you require. Always ensure that you confirm the operation of peripheral modules from the device data sheet. 2012-2013 Microchip Technology Inc. DS50002053D-page 141 MPLAB® XC8 C Compiler User’s Guide 5.3.6.1 SPECIAL BASELINE/MID-RANGE REGISTER ISSUES Some SFRs are not memory mapped, do not have a corresponding variable defined in the device specific header file, and cannot be directly accessed from C code. For example, the W register is not memory mapped on baseline devices. Some devices use OPTION and TRIS registers, that are only accessible via special instructions and that are also not memory mapped. See Section 5.3.9 “Hardware Multiply Instructions” on how these registers are accessed by the compiler. 5.3.6.2 SPECIAL PIC18 REGISTER ISSUES Some of the SFRs associated with the PIC18 can be grouped to form multi-byte values, e.g., the TMRxH and TMRxL register combined form a 16-bit timer count value. Depending on the device and mode of operation, there can be hardware requirements to read these registers in certain ways, e.g., often the TMRxL register must be read before trying to read the TMRxH register to obtain a valid 16-bit result. Although it is possible to define a word-sized C variable to map over such registers, i.e., an int variable TMRx that maps over both TMRxL and TMRxH, the order in which the compiler would read the bytes of such an object will vary from expression to expression. Some expressions require that the Most Significant Byte (MSB) is read first; others start with the Least Significant Byte (LSB) first. It is recommended that the existing SFR definitions in the chip header files be used. Each byte of the SFR should be accessed directly, and in the required order, as dictated by the device data sheet. This results in a much higher degree of portability. The following code copies the two timer registers into a C unsigned variable count for subsequent use. count = TMR0L; count += TMR0H << 8; Macros are also provided to perform reading and writing of the more common timer registers. See the macros READTIMERx and WRITETIMERx in Appendix A. “Library Functions”. These guarantee the correct byte order is used. 5.3.7 ID Locations The 8-bit PIC devices have locations outside the addressable memory area that can be used for storing program information, such as an ID number. The config pragma is also used to place data into these locations by using a special register name. The pragma is used as follows. #pragma config IDLOCx = value where x is the number (position) of the ID location, and value is the nibble or byte which is to be positioned into that ID location. If value is larger than the maximum value allowable for each location on the target device, the value will be truncated and a warning message issued. The size of each ID location varies from device to device. See your device data sheet for more information. For example: #pragma config IDLOC0 = 1 #pragma config IDLOC1 = 4 will attempt fill the first two ID locations with 1 and 4. One pragma can be used to program several locations by separating each register-value pair with a comma. For example, the above could also be specified as shown below. #pragma config IDLOC0 = 1, IDLOC1 = 4 The config pragma does not produce executable code and so should ideally be placed outside function definitions. DS50002053D-page 142 2012-2013 Microchip Technology Inc. C Language Features 5.3.7.1 ID LOCATION LEGACY SUPPORT The compiler also has legacy support for the __IDLOC macro. The macro is used in a manner similar to: #include __IDLOC(x); where x is a list of hexadecimal digits, which are positioned into the ID locations. Do not use the usual 0x hexadecimal radix specifier with these values. If an invalid character is encountered, the value 0 will be programmed into the corresponding location. Only the lower four bits of each ID location are programmed, so the following: __IDLOC(15F0); will attempt to fill ID locations with the hexadecimal values: 1, 5, F and 0. To use this macro, ensure you include in your source file. The __IDLOC macro does not produce executable code and so should ideally be placed outside function definitions. Some devices permit programming up to seven bits within each ID location. The __IDLOC() macro is not suitable for such devices and the __IDLOC7(a,b,c,d) macro should be used instead. The parameters a to d must be constants which represent the values to be programmed. The values can be entered in either decimal or hexadecimal format, such as: __IDLOC7(0x7f,1,70,0x5a); It is not appropriate to use the __IDLOC7() macro on a device that does not permit seven-bit programming of ID locations. The __IDLOC7 macro does not produce executable code and so should ideally be placed outside function definitions. 5.3.8 Bit Instructions Wherever possible, the MPLAB XC8 C Compiler will attempt to use bit instructions, even on non-bit integer values. For example, when using a bitwise operator and a mask to alter a bit within an integral type, the compiler will check the mask value to determine if a bit instruction can achieve the same functionality. unsigned int foo; foo |= 0x40; will produce the instruction: BSF _foo,6 To set or clear individual bits within integral type, the following macros could be used: #define bitset(var, bitno) #define bitclr(var, bitno) ((var) |= 1UL << (bitno)) ((var) &= ~(1UL << (bitno))) To perform the same operation on foo as above, the bitset macro could be employed as follows: bitset(foo,6); 2012-2013 Microchip Technology Inc. DS50002053D-page 143 MPLAB® XC8 C Compiler User’s Guide 5.3.9 Hardware Multiply Instructions The PIC18 instruction set includes several 8-by-8-bit hardware multiple instructions, and these are used by the compiler in some situations. Non-PIC18 targets always use an iterative library routine for multiplication operations. The hardware multiply instructions are used for all 8-bit (char) multiplications by the compiler. A special form of the int (16-bit) multiplication library routine (__wmul) has been employed to decompose the 16-bit operation into the addition of several 8-bit multiplication results. Thus, the benefits of hardware multiplication are also utilized for 16-bit multiplication expressions. A similar algorithm has not been implemented for long, 32-bit, multiplies (__lmul) as the decomposition overhead outweighs the advantage of the hardware multiply instruction. For all devices, when the hardware multiply instructions are not available, a library routine consisting of an iterative algorithm is called to perform the multiply. The smaller the first argument to these routines (i.e., the smaller the left operand to the * operator), the faster the routine will run. For example, for the code: x = 10; y = 200; result = x * y; result = y * x; // first multiply // second multiply the variable result will be assigned the same value in both statements, but the first multiplication expression will be performed faster than the second. 5.3.10 Baseline PIC MCU Special Instructions The Baseline devices have some registers which are not in the normal SFR space and cannot be accessed using an ordinary file instruction. These are the OPTION and TRIS registers. Both registers are write-only and cannot be used in expression that read their value. They can only be accessed using special instructions which the compiler will use automatically. The definition of the variables that map to these registers make use of the control qualifier. This qualifier informs the compiler that the registers are outside of the normal address space and that a different access method is required. You should not use this qualifiers for any other registers. When you write to either of these SFR variables, the compiler will use the appropriate instruction to load the value. So, for example, to load the TRIS register, the following code: TRIS = 0xFF; can be encoded by the compiler as: MOVLW 0ffh TRIS Those PIC devices which have more than one output port can have definitions for objects: TRISA, TRISB and TRISC, depending on the exact number of ports available. This objects are used in the same manner as described above. DS50002053D-page 144 2012-2013 Microchip Technology Inc. C Language Features 5.3.11 Oscillator Calibration Constants Some Baseline and Mid-range devices come with an oscillator calibration constant which is pre-programmed into the device’s program memory. This constant can be read from program memory and written to the OSCCAL register to calibrate the internal RC oscillator. On some Baseline PIC devices, the calibration constant is stored as a MOVLW instruction at the top of program memory, e.g., the PIC10F509 device. On Reset, the program counter is made to point to this instruction and it is executed first before the program counter wraps around to 0x0000, which is the effective Reset vector for the device. The default runtime startup routine (see Section 5.10.1 “Runtime Startup Code”) will automatically include code to load the OSCCAL register with the value contained in the W register after Reset on such devices. No other code is required. For other chips, such as PIC12F629 device, the oscillator constant is also stored at the top of program memory, but as a RETLW instruction. The compiler’s startup code will automatically generate code to retrieve this value and perform the configuration. At runtime, the calibration value stored as a RETLW instruction can be read using the ‘function’ __osccal_val(),as a label is assigned the RETLW instruction address. A prototype for the function is provided in . For example: calVal = __osccal_val(); Loading of the calibration value can be turned off via the osccal suboption to the --RUNTIME option (see Section 4.8.51 “--RUNTIME: Specify Runtime Environment”). At runtime, this calibration value can be read using the macro _READ_OSCCAL_DATA(). To be able to use this macro, make sure that is included into the relevant modules of your program. This macro returns the calibration constant which can then be stored into the OSCCAL register, as follows: OSCCAL = _READ_OSCCAL_DATA(); Note: The location which stores the calibration constant is never code protected and will be lost if you reprogram the device. Thus, if you are using a windowed or Flash device, the calibration constant must be saved from the last ROM location before it is erased. The constant must then be reprogrammed at the same location along with the new program and data. If you are using an in-circuit emulator (ICE), the location used by the calibration RETLW instruction cannot be programmed. Calling the _READ_OSCCAL_DATA() macro will not work and will almost certainly not return correctly. If you wish to test code that includes this macro on an ICE, you will have to program a RETLW instruction at the appropriate location in program memory. Remember to remove this instruction when programming the actual part so you do not destroy the calibration value. 2012-2013 Microchip Technology Inc. DS50002053D-page 145 MPLAB® XC8 C Compiler User’s Guide 5.3.12 REAL ICE Support The compiler supports log and trace functions (instrumented trace) when using the Microchip REAL ICE debugger. See the REAL ICE documentation for more information on the instrumented trace features. Only native trace is currently supported by the compiler. Not all devices support instrumented trace, and the IDE you are using also needs to have instrumented trace support for your target device, as well. The log and trace macro calls need to be added by hand to your source code in MPLAB IDE. They have the following form. __TRACE(id); __LOG(id, expression); MPLAB X IDE will automatically substitute an appropriate value for id when you compile; however, you can specify these by hand if required. The trace id should be a constant in the range of 0x40 to 0x7F, and the log id is a constant in the range of 0x0 to 0x7F. Each macro should be given a unique number so that it can be properly identified. The same valid number can be used for both trace and log macros. The expression can be any integer or 32-bit floating point expression. Typically, this expression is simply a variable name so the variable’s contents are logged. Macros should be placed in the C source code at the desired locations. They will trigger information to be sent to the debugger and IDE when they are executed. Adding trace and log macros will increase the size of your code as they contribute to the program image that is downloaded to the device. Here is an example of these macros that you might add. inpStatus = readUser(); if(inpStatus == 0) { __TRACE(id); recovery(); } __LOG(id, inpStatus); 5.3.13 Function profiling The compiler can generate function registration code for the MPLAB REAL ICE debugger to provide function profiling. The flp suboption to the --RUNTIME option (see Section 4.8.51 “--RUNTIME: Specify Runtime Environment”) enables this feature. To obtain profiling results, you must also use a Power Monitor Board and MPLAB X IDE and power monitor plugin that support code profiling for the MPLAB XC8 C Compiler. When enabled, the compiler inserts assembly code into the prologue and epilogue of each function. This code communicates runtime information to the debugger to signal when a function is being entered and when it exits. This information, along with further measurements made by a Microchip Power Monitor Board, can determine how much energy each function is using. This feature is transparent, but note the following points when profiling is enabled: • The program will increase in size and run slower due to the profiling code • One extra level of hardware stack is used • Some additional RAM memory is consumed • Inlining of functions will not take place for any profiled function If a function cannot be profiled (due to hardware stack constraints) but is qualified inline, the compiler might inline the function. See Section 5.8.1.2 “Inline Specifier” for more information on inlining functions. DS50002053D-page 146 2012-2013 Microchip Technology Inc. C Language Features 5.4 SUPPORTED DATA TYPES AND VARIABLES 5.4.1 Identifiers A C variable identifier (the following is also true for function identifiers) is a sequence of letters and digits, where the underscore character “_” counts as a letter. Identifiers cannot start with a digit. Although they can start with an underscore, such identifiers are reserved for the compiler’s use and should not be defined by your programs. Such is not the case for assembly domain identifiers, which often begin with an underscore, see Section 5.12.3.1 “Equivalent Assembly Symbols”. Identifiers are case sensitive, so main is different to Main. Not every character is significant in an identifier. The maximum number of significant characters can be set using an option, see Section 4.8.8 “-N: Identifier Length”. If two identifiers differ only after the maximum number of significant characters, then the compiler will consider them to be the same symbol. 5.4.2 Integer Data Types The MPLAB XC8 compiler supports integer data types with 1, 2, 3 and 4 byte sizes as well as a single bit type. Table 5-1 shows the data types and their corresponding size and arithmetic type. The default type for each type is underlined. TABLE 5-1: INTEGER DATA TYPES Type Size (bits) Arithmetic Type bit 1 Unsigned integer signed char 8 Signed integer unsigned char 8 Unsigned integer signed short 16 Signed integer unsigned short 16 Unsigned integer signed int 16 Signed integer unsigned int 16 Unsigned integer signed short long 24 Signed integer unsigned short long 24 Unsigned integer signed long 32 Signed integer unsigned long 32 Unsigned integer signed long long 32 Signed integer unsigned long long 32 Unsigned integer The bit and short long types are non-standard types available in this implementation. The long long types are C99 Standard types, but this implementation limits their size to only 32 bits. All integer values are represented in little endian format with the Least Significant bit (LSb) at the lower address. If no signedness is specified in the type, then the type will be signed except for the char types which are always unsigned. The bit type is always unsigned and the concept of a signed bit is meaningless. Signed values are stored as a two’s complement integer value. The range of values capable of being held by these types is summarized in Table 5-2 The symbols in this table are preprocessor macros which are available after including in your source code. 2012-2013 Microchip Technology Inc. DS50002053D-page 147 MPLAB® XC8 C Compiler User’s Guide As the size of data types are not fully specified by the ANSI Standard, these macros allow for more portable code which can check the limits of the range of values held by the type on this implementation. The macros associated with the short long type are non-standard macros available in this implementation; those associated with the long long types are defined by the C99 Standard. TABLE 5-2: RANGES OF INTEGER TYPE VALUES Symbol Meaning Value CHAR_BIT bits per char 8 CHAR_MAX max. value of a char 127 CHAR_MIN min. value of a char -128 SCHAR_MAX max. value of a signed char 127 SCHAR_MIN min. value of a signed char -128 UCHAR_MAX max. value of an unsigned char 255 SHRT_MAX max. value of a short 32767 SHRT_MIN min. value of a short -32768 USHRT_MAX max. value of an unsigned short 65535 INT_MAX max. value of an int 32767 INT_MIN min. value of a int -32768 UINT_MAX max. value of an unsigned int 65535 SHRTLONG_MAX max. value of a short long 8388607 SHRTLONG_MIN min. value of a short long -8388608 USHRTLONG_MAX max. value of an unsigned short 16777215 long LONG_MAX max. value of a long 2147483647 LONG_MIN min. value of a long -2147483648 ULONG_MAX max. value of an unsigned long 4294967295 LLONG_MAX max. value of a long long 2147483647 LLONG_MIN min. value of a long long -2147483648 ULLONG_MAX max. value of an unsigned long 4294967295 long Macros are also available in which define values associated with fixed-width types. When specifying a signed or unsigned short int, short long int, long int or long long int type, the keyword int can be omitted. Thus a variable declared as short will contain a signed short int and a variable declared as unsigned short will contain an unsigned short int. It is a common misconception that the C char types are intended purely for ASCII character manipulation. However, the C language makes no guarantee that the default character representation is even ASCII. (This implementation does use ASCII as the character representation.) The char types are the smallest of the multi-bit integer sizes, and behave in all respects like integers. The reason for the name “char” is historical and does not mean that char can only be used to represent characters. It is possible to freely mix char values with values of other types in C expressions. With the MPLAB XC8 C Compiler, the char types are used for a number of purposes – as 8-bit integers, as storage for ASCII characters, and for access to I/O locations. DS50002053D-page 148 2012-2013 Microchip Technology Inc. C Language Features 5.4.2.1 BIT DATA TYPES AND VARIABLES The MPLAB XC8 C Compiler supports bit integral types which can hold the values 0 or 1. Single bit variables can be declared using the keyword bit (or __bit), for example: bit init_flag; You can also use the These variables cannot be auto or parameters to a function, but can be qualified static, allowing them to be defined locally within a function. For example: int func(void) { static bit flame_on; // ... } A function can return a bit object by using the bit keyword in the function’s prototype in the usual way. The 1 or 0 value will be returned in the carry flag in the STATUS register. The bit variables behave in most respects like normal unsigned char variables, but they can only contain the values 0 and 1, and therefore provide a convenient and efficient method of storing flags. Eight bit objects are packed into each byte of memory storage, so they don’t consume large amounts of internal RAM. Operations on bit objects are performed using the single bit instructions (bsf and bcf) wherever possible, thus the generated code to access bit objects is very efficient. It is not possible to declare a pointer to bit types or assign the address of a bit object to any pointer. Nor is it possible to statically initialize bit variables so they must be assigned any non-zero starting value (i.e., 1) in the code itself. Bit objects will be cleared on startup, unless the bit is qualified persistent. When assigning a larger integral type to a bit variable, only the LSb is used. For example, if the bit variable bitvar was assigned as in the following: int data = 0x54; bit bitvar; bitvar = data; it will be cleared by the assignment since the LSb of data is zero. This sets the bit type apart from the C99 Standard __Bool, which is a boolean type, not a 1-bit wide integer. The __Bool type is not supported on the MPLAB XC8 compiler. If you want to set a bit variable to be 0 or 1 depending on whether the larger integral type is zero (false) or non-zero (true), use the form: bitvar = (data != 0); The psects in which bit objects are allocated storage are declared using the bit PSECT directive flag, see Section 6.4.9.3 “PSECT”. All addresses assigned to bit objects and psects will be bit addresses. For absolute bit variables (see Section 5.5.4 “Absolute Variables”), the address specified in code must be a bit address. Take care when comparing these addresses to byte addresses used by all other variables. If the xc8 flag --STRICT is used, the bit keyword becomes unavailable, but you can use the __bit keyword. 2012-2013 Microchip Technology Inc. DS50002053D-page 149 MPLAB® XC8 C Compiler User’s Guide 5.4.3 Floating-Point Data Types The MPLAB XC8 compiler supports 24- and 32-bit floating-point types. Floating point is implemented using either a IEEE 754 32-bit format, or a modified (truncated) 24-bit form of this. Table 5-3 shows the data types and their corresponding size and arithmetic type. TABLE 5-3: FLOATING-POINT DATA TYPES Type Size (bits) Arithmetic Type float 24 or 32 Real double 24 or 32 Real long double same as double Real For both float and double values, the 24-bit format is the default. The options --FLOAT=24 and --DOUBLE=24 can also be used to specify this explicitly. The 32-bit format is used for double values if the --DOUBLE=32 option is used and for float values if --FLOAT=32 is used. Variables can be declared using the float and double keywords, respectively, to hold values of these types. Floating-point types are always signed and the unsigned keyword is illegal when specifying a floating-point type. Types declared as long double will use the same format as types declared as double. All floating-point values are represented in little endian format with the LSb at the lower address. This format is described in Table 5-4, where: • Sign is the sign bit which indicates if the number is positive or negative • The exponent is 8 bits which is stored as excess 127 (i.e., an exponent of 0 is stored as 127). • Mantissa is the mantissa, which is to the right of the radix point. There is an implied bit to the left of the radix point which is always 1 except for a zero value, where the implied bit is zero. A zero value is indicated by a zero exponent. The value of this number is (-1)sign x 2(exponent-127) x 1. mantissa. TABLE 5-4: FLOATING-POINT FORMATS Format Sign Biased exponent Mantissa IEEE 754 32-bit x xxxx xxxx xxx xxxx xxxx xxxx xxxx xxxx modified IEEE 754 24-bit x xxxx xxxx xxx xxxx xxxx xxxx Here are some examples of the IEEE 754 32-bit formats shown in Table 5-5. Note that the Most Significant Bit (MSb) of the mantissa column (i.e., the bit to the left of the radix point) is the implied bit, which is assumed to be 1 unless the exponent is zero (in which case the float is zero). TABLE 5-5: Format FLOATING-POINT FORMAT EXAMPLE IEEE 754 Number 32-bit 7DA6B69Bh 24-bit 42123Ah Biased exponent 11111011b 1.mantissa 1.0100110101101101 0011011b Decimal 2.77000e+37 (251) (1.302447676659) — 10000100b 1.001001000111010b 36.557 (132) (1.142395019531) — Use the following process to manually calculate the 32-bit example in Table 5-5. DS50002053D-page 150 2012-2013 Microchip Technology Inc. C Language Features The sign bit is zero; the biased exponent is 251, so the exponent is 251-127=124. Take the binary number to the right of the decimal point in the mantissa. Convert this to decimal and divide it by 223 where 23 is the number of bits taken up by the mantissa, to give 0.302447676659. Add 1 to this fraction. The floating-point number is then given by: -1021241.302447676659 which becomes: 12.126764793256e+371.302447676659 which is approximately equal to: 2.77000e+37 Binary floating-point values are sometimes misunderstood. It is important to remember that not every floating-point value can be represented by a finite sized floating-point number. The size of the exponent in the number dictates the range of values that the number can hold, and the size of the mantissa relates to the spacing of each value that can be represented exactly. Thus the 24-bit format allows for values with approximately the same range of values representable by the 32-bit format, but the values that can be exactly represented by this format are more widely spaced. So, for example, if you are using a 24-bit wide floating-point type, it can exactly store the value 95000.0. However, the next highest number it can represent is 95002.0 and it is impossible to represent any value in between these two in such a type as it will be rounded. This implies that C code which compares floating-point type cannot behave as expected. For example: volatile float myFloat; myFloat = 95002.0; if(myFloat == 95001.0) PORTA++; // value will be rounded // this line will be executed! in which the result of the if() expression will be true, even though it appears the two values being compared are different. Compare this to a 32-bit floating-point type, which has a higher precision. It also can exactly store 95000.0 as a value. The next highest value which can be represented is (approximately) 95000.00781. The characteristics of the floating-point formats are summarized in Table 5-6. The symbols in this table are preprocessor macros which are available after including in your source code. Two sets of macros are available for float and double types, where XXX represents FLT and DBL, respectively. So, for example, FLT_MAX represents the maximum floating-point value of the float type. It can have two values depending on whether float is a 24 or 32 bit wide format. DBL_MAX represents the same values for the double type. 2012-2013 Microchip Technology Inc. DS50002053D-page 151 MPLAB® XC8 C Compiler User’s Guide As the size and format of floating-point data types are not fully specified by the ANSI Standard, these macros allow for more portable code which can check the limits of the range of values held by the type on this implementation. TABLE 5-6: RANGES OF FLOATING-POINT TYPE VALUES Symbol Meaning 24-bit Value 32-bit Value XXX_RADIX Radix of exponent representation 2 2 XXX_ROUNDS Rounding mode for addition 0 XXX_MIN_EXP Min. n such that FLT_RADIXn-1 is -125 a normalized float value XXX_MIN_10_EXP Min. n such that 10n is a normalized float value 0 -125 -37 -37 128 128 38 38 XXX_MANT_DIG Number of FLT_RADIX mantissa 16 digits 24 XXX_EPSILON The smallest number which added to 1.0 does not yield 1.0 1.19209e-07 XXX_MAX_EXP Max. n such that FLT_RADIXn-1 is a normalized float value XXX_MAX_10_EXP Max. n such that 10n is a normalized float value 5.4.4 3.05176e-05 Structures and Unions MPLAB XC8 C Compiler supports struct and union types. Structures and unions only differ in the memory offset applied to each member. These types will be at least 1 byte wide. The members of structures and unions cannot be objects of type bit, but bit-fields are fully supported. Structures and unions can be passed freely as function arguments and function return values. Pointers to structures and unions are fully supported. 5.4.4.1 STRUCTURE AND UNION QUALIFIERS The compiler supports the use of type qualifiers on structures. When a qualifier is applied to a structure, all of its members will inherit this qualification. In the following example the structure is qualified const. const struct { int number; int *ptr; } record = { 0x55, &i }; In this case, the entire structure will be placed into the program space and each member will be read-only. Remember that all members are usually initialized if a structure is const as they cannot be initialized at runtime. If the members of the structure were individually qualified const, but the structure was not, then the structure would be positioned into RAM, but each member would be read-only. Compare the following structure with the above. struct { const int number; int * const ptr; } record = { 0x55, &i }; DS50002053D-page 152 2012-2013 Microchip Technology Inc. C Language Features 5.4.4.2 BIT-FIELDS IN STRUCTURES MPLAB XC8 C Compiler fully supports bit-fields in structures. Bit-fields are always allocated within 8-bit words, even though it is usual to use the type unsigned int in the definition. The first bit defined will be the LSb of the word in which it will be stored. When a bit-field is declared, it is allocated within the current 8-bit unit if it will fit; otherwise, a new byte is allocated within the structure. Bit-fields can never cross the boundary between 8-bit allocation units. For example, the declaration: struct { unsigned unsigned unsigned } foo; lo : 1; dummy : 6; hi : 1; will produce a structure occupying 1 byte. If foo was ultimately linked at address 10H, the field lo will be bit 0 of address 10H; hi will be bit 7 of address 10H. The LSb of dummy will be bit 1 of address 10H and the MSb of dummy will be bit 6 of address 10h. Note: Accessing bit-fields larger than a single bit can be very inefficient. If code size and execution speed are critical, consider using a char type or a char structure member, instead. Be aware that some SFRs are defined as bit-fields. Most are single bits, but some can be multi-bit objects. Unnamed bit-fields can be declared to pad out unused space between active bits in control registers. For example, if dummy is never referenced, the structure above could have been declared as: struct { unsigned unsigned unsigned } foo; lo : 1; : 6; hi : 1; A structure with bit-fields can be initialized by supplying a comma-separated list of initial values for each field. For example: struct { unsigned unsigned unsigned } foo = {1, 8, 0}; lo : 1; mid : 6; hi : 1; Structures with unnamed bit-fields can be initialized. No initial value should be supplied for the unnamed members, for example: struct { unsigned unsigned unsigned } foo = {1, 0}; lo hi : 1; : 6; : 1; will initialize the members lo and hi correctly. A bit-field that has a size of 0 is a special case. The Standard indicates that no further bit-field is to be packed into the allocation unit in which the previous bit-field, if any, was placed. 2012-2013 Microchip Technology Inc. DS50002053D-page 153 MPLAB® XC8 C Compiler User’s Guide 5.4.4.3 ANONYMOUS UNIONS The MPLAB XC8 compiler supports anonymous unions. These are unions with no identifier and whose members can be accessed without referencing the enclosing union. These unions can be used when placing inside structures. For example: struct { union { int x; double y; }; } aaa; void main(void) { aaa.x = 99; // ...} Here, the union is not named and its members accessed as if they are part of the structure. Anonymous unions are not part of the ISO C90 C Standard. Their use limits the portability of any code, and they are not recommended 5.4.5 Pointer Types There are two basic pointer types supported by the MPLAB XC8 C Compiler: data pointers and function pointers. Data pointers hold the addresses of variables which can be indirectly read, and possible indirectly written, by the program. Function pointers hold the address of an executable function which can be called indirectly via the pointer. To conserve memory requirements and reduce execution time, pointers are made different sizes and formats. The MPLAB XC8 C Compiler uses sophisticated algorithms to track the assignment of addresses to all pointers, and, as a result, non-standard qualifiers are not required when defining pointer variables. The standard qualifiers const and volatile can still be used and have their usual meaning. Despite this, the size of each pointer is optimal for its intended usage in the program. 5.4.5.1 COMBINING TYPE QUALIFIERS AND POINTERS It is helpful to first review the ANSI C standard conventions for definitions of pointer types. Pointers can be qualified like any other C object, but care must be taken when doing so as there are two quantities associated with pointers. The first is the actual pointer itself, which is treated like any ordinary C variable and has memory reserved for it. The second is the target, or targets, that the pointer references, or to which the pointer points. The general form of a pointer definition looks like the following: target_type_&_qualifiers * pointer’s_qualifiers pointer’s_name; Any qualifiers to the right of the * (i.e., next to the pointer’s name) relate to the pointer variable itself. The type and any qualifiers to the left of the * relate to the pointer’s targets. This makes sense since it is also the * operator that dereferences a pointer, which allows you to get from the pointer variable to its current target. Here are three examples of pointer definitions using the volatile qualifier. The fields in the definitions have been highlighted with spacing: volatile int * int * volatile volatile int * volatile DS50002053D-page 154 vip ; ivp ; vivp ; 2012-2013 Microchip Technology Inc. C Language Features The first example is a pointer called vip. It contains the address of int objects that are qualified volatile. The pointer itself — the variable that holds the address — is not volatile; however, the objects that are accessed when the pointer is dereferenced are treated as being volatile. In other words, the target objects accessible via the pointer can be externally modified. The second example is a pointer called ivp which also contains the address of int objects. In this example, the pointer itself is volatile, that is, the address the pointer contains can be externally modified; however, the objects that can be accessed when dereferencing the pointer are not volatile. The last example is of a pointer called vivp which is itself qualified volatile, and which also holds the address of volatile objects. Bear in mind that one pointer can be assigned the addresses of many objects; for example, a pointer that is a parameter to a function is assigned a new object address every time the function is called. The definition of the pointer must be valid for every target address assigned. Note: 5.4.5.2 Care must be taken when describing pointers. Is a “const pointer” a pointer that points to const objects, or a pointer that is const itself? You can talk about “pointers to const” and “const pointers” to help clarify the definition, but such terms cannot be universally understood. DATA POINTERS The MPLAB XC8 compiler monitors and records all assignments of addresses to each data pointer the program contains. This includes assignment of the addresses of objects to pointers; assignment of one pointer to another; initialization of pointers when they are defined; and takes into account when pointers are ordinary variables and function parameters, and when pointers are used to access basic objects, or structures or arrays. The size and format of the address held by each pointer is based on this information. When more than one address is assigned to a pointer at different places in the code, a set of all possible targets the pointer can address is maintained. This information is specific to each pointer defined in the program, thus two pointers with the same C type can hold addresses of different sizes and formats due to the way the pointers were used in the program. The compiler tracks the memory location of all targets, as well as the size of all targets to determine the size and scope of a pointer. The size of a target is important as well, particularly with arrays or structures. It must be possible to increment a pointer so it can access all the elements of an array, for example. There are several pointer classifications used with the MPLAB XC8 C Compiler, such as those indicated below. For baseline and mid-range devices: • An 8-bit pointer capable of accessing common memory and two consecutive banks, e.g., banks 0 and 1, or banks 7 and 8, etc. • A 16-bit pointer capable of accessing the entire data memory space • An 8-bit pointer capable of accessing up to 256 bytes of program space data • A 16-bit pointer capable of accessing up to 64 Kbytes of program space data • A 16-bit mixed target space pointer capable of accessing the entire data space memory and up to 64 Kbytes of program space data For PIC18 devices: • An 8-bit pointer capable of accessing the access bank • A 16-bit pointer capable of accessing the entire data memory space 2012-2013 Microchip Technology Inc. DS50002053D-page 155 MPLAB® XC8 C Compiler User’s Guide • • • • An 8-bit pointer capable of accessing up to 256 bytes of program space data A 16-bit pointer capable of accessing up to 64 Kbytes of program space data A 24-bit pointer capable of accessing the entire program space A 16-bit mixed target space pointer capable of accessing the entire data space memory and up to 64 Kbytes of program space data • A 24-bit mixed target space pointer capable of accessing the entire data space memory and the entire program space Each data pointer will be allocated one of the available classifications after preliminary scans of the source code. There is no mechanism by which the programmer can specify the style of pointer required (other than by the assignments to the pointer). The C code must convey the required information to the compiler. Information about the pointers and their targets are shown in the pointer reference graph, which is described in Section 6.6.5 “Pointer Reference Graph”. This graph is printed in the assembly list file, which is controlled by the option described in Section 4.8.16 “--ASMLIST: Generate Assembler List Files”. Consider the following mid-range device program in the early stages of development. It consists of the following code: int i, j; int getValue(const int * ip) { return *ip; } void main(void) { j = getValue(&i); // ... code that uses j } A pointer, ip, is a parameter to the function getValue(). The pointer target type uses the qualifier const because we do not want the pointer to be used to write to any objects whose addresses are passed to the function. The const qualification serves no other purpose and does not alter the format of the pointer variable. If the compiler allocates the variable i (defined in main()) to bank 0 data memory, it will also be noted that the pointer ip (parameter to getValue()) only points to one object that resides in bank 0 of the data memory. In this case, the pointer, ip, is made an 8-bit wide data pointer. The generated code that dereferences ip in getValue() will be generated assuming that the address can only be to an object in bank 0. As the program is developed, another variable, x, is defined and (unknown to the programmer) is allocated space in bank 2 data memory. The main() function now looks like: int i, j; int x; // allocated to bank 0 in this example // allocated to bank 2 in this example int getValue(const int * ip) { return *ip; } void main(void) { j = getValue(&i); // ... code that uses j j = getValue(&x); // ... code that uses j } DS50002053D-page 156 2012-2013 Microchip Technology Inc. C Language Features The pointer, ip, now has targets that are in bank 0 and in bank 2.To be able to accommodate this situation, the pointer is made 16 bits wide, and the code used to dereference the pointer will change accordingly. This takes place without any modification to the source code. One positive aspect of tracking pointer targets is less of a dependence on pointer qualifiers. The standard qualifiers const and volatile must still be used in pointer definitions to indicate a read-only or externally-modifiable target object, respectively. However, this is in strict accordance with the ANSI C standard. Non-standard qualifiers, like near and bank2, are not required to indicate pointer targets, have no effect, and should be avoided. Omitting these qualifiers will result in more portable and readable code, and reduce the chance of extraneous warnings being issued by the compiler. 5.4.5.2.1 Pointers to Both Memory Spaces When a pointer is assigned the address of one or more objects that have been allocated memory in the data space, and also assigned the address of one or more const objects, the pointer will fall into one of the mixed target space pointers listed in Section 5.4.5.2 “Data Pointers”, and the address will be encoded so that the target memory space can be determined at runtime. The encoding of these pointer types are as follows. For the Baseline/Mid-range 16-bit mixed target space pointer, the MSb of the address (i.e., bit number 15) indicates the memory space that the address references. If this bit is set, it indicates that the address is of something in program memory; clear indicates an object in the data memory. The remainder of this address represents the full address in the indicated memory space. For the PIC18 16-bit mixed target space pointer, any address above the highest data space address is that of an object in the program space memory; otherwise, the address is of a data space memory object. For the PIC18 24-bit mixed target space pointer, bit number 21 indicates the memory space that the address references. If this bit is set, it indicates that the address is of an object residing in data memory; if it is clear, it indicates an object in the program memory. The remainder of this address represents the full address in the indicated memory space. Note that for efficiency reasons, the meaning of the memory space bit is the opposite to that for baseline and mid-range devices. To extend the mid-range device example given in Section 5.4.5.2 “Data Pointers”, the code is now developed further. The function getValue() is now called with the address of an object that resides in the program memory, as shown. int i, j; // allocated to bank 0 in this example int x; // allocated to bank 2 in this example const int type = 0x3456; int getValue(const int * ip) { return *ip; } void main(void) { j = getValue(&i); // ... code that uses j j = getValue(&x); // ... code that uses j j = getValue(&type); // ... code that uses j } 2012-2013 Microchip Technology Inc. DS50002053D-page 157 MPLAB® XC8 C Compiler User’s Guide Again, the targets to the pointer, ip, are determined, and now the pointer is made of the class that can access both data and program memory. The generated code to dereference the pointer will be such that it can determine the required memory space from the address, and access either space accordingly. Again, this takes place without any change in the definition of the pointer. If assembly code references a C pointer, the compiler will force that pointer to become a 16-bit mixed target space pointer, in the case of baseline or mid-range programs, or a 24-bit mixed target space pointer, for PIC18 programs. These pointer types have unrestricted access to all memory areas and will operate correctly, even if assignments (of a correctly formatted address) are made to the pointer in the assembly code. 5.4.5.3 FUNCTION POINTERS The MPLAB XC8 compiler fully supports pointers to functions, which allows functions to be called indirectly. These are often used to call one of several function addresses stored in a user-defined C array, which acts like a lookup table. For baseline and mid-range devices, function pointers are always one byte in size and hold an offset into a jump table that is output by the compiler. This jump table contains jumps to the destination functions. For PIC18 devices, function pointers are either 16 or 24 bits wide. The pointer size is purely based on the amount of program memory available on the target device. As with data pointers, the target assigned to function pointers is tracked. This is an easier process to undertake compared to that associated with data pointers as all function instructions must reside in program memory. The pointer reference graph (described in Section 6.6.5 “Pointer Reference Graph”) will show function pointers, in addition to data pointers, as well as all their targets. The targets will be names of functions that could possibly be called via the pointer. One notable runtime feature for baseline and mid-range devices is that a function pointer which contains null (the value 0) and is used to call a function indirectly will cause the code to become stuck in a loop which branches to itself. This endless loop can be used to detect this erroneous situation. Typically calling a function via a null function would result in the code crashing or some other unexpected behavior. The label to which the endless loop will jump is called fpbase. 5.4.5.4 SPECIAL POINTER TARGETS Pointers and integers are not interchangeable. Assigning an integer constant to a pointer will generate a warning to this effect. For example: const char * cp = 0x123; // the compiler will flag this as bad code There is no information in the integer constant, 0x123, relating to the type, size or memory location of the destination. There is a very good chance of code failure if pointers are assigned integer addresses and dereferenced, particularly for PIC devices that have more than one memory space. Is 0x123 an address in data memory or program memory? How big is the object found at address 0x123? Always take the address of a C object when assigning an address to a pointer. If there is no C object defined at the destination address, then define or declare an object at this address which can be used for this purpose. Make sure the size of the object matches the range of the memory locations that are to be accessed by the pointer. DS50002053D-page 158 2012-2013 Microchip Technology Inc. C Language Features For example, a checksum for 1000 memory locations starting at address 0x900 in program memory is to be generated. A pointer is used to read this data. You can be tempted to write code such as: const char * cp; cp = 0x900; // what resides at 0x900??? and increment the pointer over the data. However, a much better solution is this: const char * cp; const char inputData[1000] @ 0x900; cp = &inputData; // cp is incremented over inputData and used to read values there In this case, the compiler can determine the size of the target and the memory space. The array size and type indicates the size of the pointer target, the const qualifier on the object (not the pointer) indicates the target is located in program memory space. Note that the const array does not need initial values to be specified in this instance, see Section 5.4.7.1 “Const Type Qualifier” and can reside over the top of other objects at these addresses. If the pointer has to access objects in data memory, you need to define a different object to act as a dummy target. For example, if the checksum was to be calculated over 10 bytes starting at address 0x90 in data memory, the following code could be used. const char * cp; extern char inputData[10] @ 0x90; cp = &inputData; // cp is incremented over inputData and used to read values there No memory is consumed by the extern declaration, and this can be mapped over the top of existing objects. User-defined absolute objects will not be cleared by the runtime startup code and can be placed over the top of other absolute variables. Take care when comparing (subtracting) pointers. For example: if(cp1 == cp2) ; // take appropriate action The ANSI C standard only allows pointer comparisons when the two pointer targets are the same object. One exception is that the address can extend to one element past the end of an array. Comparisons of pointers to integer constants are even more risky, for example: if(cp1 == 0x246) ; // take appropriate action Never compare pointers with integer constants. A null pointer is the one instance where a constant value can be assigned to a pointer and this is handled correctly by the compiler. A null pointer is numerically equal to 0 (zero), but this is a special case imposed by the ANSI C standard. Comparisons with the macro NULL are also allowed. If null is the only value assigned to a pointer, the pointer will be made as small as possible. 2012-2013 Microchip Technology Inc. DS50002053D-page 159 MPLAB® XC8 C Compiler User’s Guide 5.4.6 Constant Types and Formats A constant is used to represent an immediate value in the source code, as opposed to a variable that could hold the same value. For example 123 is a constant. Like any value, a constant must have a C type. In addition to a constant’s type, the actual value can be specified in one of several formats. 5.4.6.1 INTEGRAL CONSTANTS The format of integral constants specifies their radix. MPLAB XC8 supports the ANSI standard radix specifiers, as well as ones which enables binary constants to be specified in C code. The formats used to specify the radices are given in Table 5-7. The letters used to specify binary or hexadecimal radices are case insensitive, as are the letters used to specify the hexadecimal digits. TABLE 5-7: RADIX FORMATS Radix binary Format Example octal 0b number or 0B number 0 number 0763 0b10011010 decimal number 129 hexadecimal 0x number or 0X number 0x2F Any integral constant will have a type of int, long int or long long int, so that the type can hold the value without overflow. Constants specified in octal or hexadecimal can also be assigned a type of unsigned int, unsigned long int or unsigned long long int if the signed counterparts are too small to hold the value. The default types of constants can be changed by the addition of a suffix after the digits; e.g., 23U, where U is the suffix. Table 5-8 shows the possible combination of suffixes and the types that are considered when assigning a type. So, for example, if the suffix l is specified and the value is a decimal constant, the compiler will assign the type long int, if that type will hold the constant; otherwise, it will assigned long long int. If the constant was specified as an octal or hexadecimal constant, then unsigned types are also considered. TABLE 5-8: SUFFIXES AND ASSIGNED TYPES Suffix DS50002053D-page 160 Decimal Octal or Hexadecimal u or U unsigned int unsigned long int unsigned long long int unsigned int unsigned long int unsigned long long int l or L long int long long int long int unsigned long int long long int unsigned long long int u or U, and l or L unsigned long int unsigned long long int unsigned long int unsigned long long int ll or LL long long int long long int unsigned long long int u or U, and ll or LL unsigned long long int unsigned long long int 2012-2013 Microchip Technology Inc. C Language Features Here is an example of code that can fail because the default type assigned to a constant is not appropriate: unsigned long int result; unsigned char shifter; void main(void) { shifter = 20; result = 1 << shifter; // code that uses result } The constant 1 (one) will be assigned an int type, hence the result of the shift operation will be an int. Even though this result is assigned to the long variable, result, it can never become larger than the size of an int, regardless of how much the constant is shifted. In this case, the value 1 shifted left 20 bits will yield the result 0, not 0x100000. The following uses a suffix to change the type of the constant, hence ensure the shift result has an unsigned long type. result = 1UL << shifter; 5.4.6.2 FLOATING-POINT CONSTANT Floating-point constants have double type unless suffixed by f or F, in which case it is a float constant. The suffixes l or L specify a long double type which is considered an identical type to double by MPLAB XC8. 5.4.6.3 CHARACTER AND STRING CONSTANTS Character constants are enclosed by single quote characters, ’, for example ’a’. A character constant has int type, although this can be later optimized to a char type by the compiler. To comply with the ANSI C standard, the compiler does not support the extended character set in characters or character arrays. Instead, they need to be escaped using the backslash character, as in the following example. const char name[] = "Bj\370rk"; printf("%s's Resum\351", name); \\ prints "Bjørk's Resumé" Multi-byte character constants are not supported by this implementation. String constants, or string literals, are enclosed by double quote characters “, for example “hello world”. The type of string constants is const char * and the character that make up the string are stored in the program memory, as are all objects qualified const. A common warning relates to assigning a string literal to a pointer that does not specify a const target, for example: char * cp = "hello world\n"; The string characters cannot be modified, but this type of pointer allows writes to take place, hence the warning. To prevent yourself from trying to overwrite the string, qualifier the pointer target as follows. See also Section 5.4.5.1 “Combining Type Qualifiers and Pointers”. const char * cp = "hello world\n"; 2012-2013 Microchip Technology Inc. DS50002053D-page 161 MPLAB® XC8 C Compiler User’s Guide Defining and initializing an array (i.e., not a pointer) with a string is an exception. For example: char ca[]= "hello world\n"; will actually copy the string characters into the RAM array, rather than assign the address of the characters to a pointer, as in the previous examples. The string literal remains read-only, but the array is both readable and writable. The MPLAB XC8 compiler will use the same storage location and label for strings that have identical character sequences, except where the strings are used to initialize an array residing in the data space. For example, in the code snippet if(strncmp(scp, "hello", 6) == 0) fred = 0; if(strcmp(scp, "world") == 0) fred--; if(strcmp(scp, "hello world") == 0) fred++; the characters in the string “world” and the last 6 characters of the string “hello world” (the last character is the null terminator character) would be represented by the same characters in memory. The string “hello” would not overlap with the same characters in the string “hello world” as they differ in terms of the placement of the null character. Two adjacent string constants (i.e., two strings separated only by white space) are concatenated by the compiler. Thus: const char * cp = "hello" "world"; will assign the pointer with the address of the string “hello world “. DS50002053D-page 162 2012-2013 Microchip Technology Inc. C Language Features 5.4.7 Standard Type Qualifiers Type qualifiers provide additional information regarding how an object can be used. The MPLAB XC8 compiler supports both ANSI C qualifiers and additional special qualifiers which are useful for embedded applications and which take advantage of the 8-bit PIC MCU architecture. 5.4.7.1 CONST TYPE QUALIFIER MPLAB XC8 supports the use of the ANSI type qualifiers const and volatile. The const type qualifier is used to tell the compiler that an object is read only and will not be modified. If any attempt is made to modify an object declared const, the compiler will issue a warning or error. User-defined objects declared const are placed in a special psect linked into the program space. Objects qualified const can be absolute. The @ address construct is used to place the object at the specified address in program memory, as in the following example which places the object tableDef at address 0x100. const int tableDef[] @ 0x100 = { 0, 1, 2, 3, 4}; Usually a const object must be initialized when it is declared, as it cannot be assigned a value at any point at runtime. For example: const int version = 3; will define version as being an int variable that will be placed in the program memory, will always contain the value 3, and which can never be modified by the program. However, uninitialized const objects can be defined and are useful if you need to place an object in program memory over the top of other objects at a particular location. Usually uninitialized const objects will be defined as absolute, as in the following example. const char checksumRange[0x100] @ 0x800; will define the object checksumRange as a 0x100 byte array of characters located at address 0x800 in program memory. This definition will not place any data in the HEX file. 2012-2013 Microchip Technology Inc. DS50002053D-page 163 MPLAB® XC8 C Compiler User’s Guide 5.4.7.2 VOLATILE TYPE QUALIFIER The volatile type qualifier is used to tell the compiler that an object cannot be guaranteed to retain its value between successive accesses. This prevents the optimizer from eliminating apparently redundant references to objects declared volatile because it can alter the behavior of the program to do so. Any SFR which can be modified by hardware or which drives hardware is qualified as volatile, and any variables which can be modified by interrupt routines should use this qualifier as well. For example: volatile static unsigned int TACTL @ 0x160; The volatile qualifier does not guarantee that any access will be atomic, which is often not the case with the 8-bit PIC MCU architecture. All these devices can only access a maximum of 1 byte of data per instruction. The code produced by the compiler to access volatile objects can be different to that to access ordinary variables, and typically the code will be longer and slower for volatile objects, so only use this qualifier if it is necessary. However, failure to use this qualifier when it is required can lead to code failure. Another use of the volatile keyword is to prevent variables being removed if they are not used in the C source. If a non-volatile variable is never used, or used in a way that has no effect on the program’s function, then it can be removed before code is generated by the compiler. A C statement that consists only of a volatile variable’s name will produce code that reads the variable’s memory location and discards the result. For example the entire statement: PORTB; will produce assembly code the reads PORTB, but does nothing with this value. This is useful for some peripheral registers that require reading to reset the state of interrupt flags. Normally such a statement is not encoded as it has no effect. Some variables are treated as being volatile even though they cannot be qualified in the source code. See Section 5.12.3.4 “Undefined Symbols” if you have assembly code in your project. DS50002053D-page 164 2012-2013 Microchip Technology Inc. C Language Features 5.4.8 Special Type Qualifiers The MPLAB XC8 C Compiler supports special type qualifiers to allow the user to control placement of static and extern class variables into particular address spaces. 5.4.8.1 PERSISTENT TYPE QUALIFIER By default, any C variables that are not explicitly initialized are cleared on startup. This is consistent with the definition of the C language. However, there are occasions where it is desired for some data to be preserved across a Reset. The persistent type qualifier (or __persistent) is used to qualify variables that should not be cleared by the runtime startup code. In addition, any persistent variables will be stored in a different area of memory to other variables. Different psects are used to hold these objects. See 5.15.2 “Compiler-Generated Psects” for more information. This type qualifier cannot be used on variables of class auto; however, statically defined local variables can be qualified persistent. For example, you should write: void test(void) { static persistent int intvar; // ... } /* must be static */ If the xc8 option, --STRICT is used, you cannot use the persistent qualifier, but you can continue to use __persistent. 5.4.8.2 NEAR TYPE QUALIFIER Some of the 8-bit PIC architectures implement data memory which can be always accessed regardless of the currently selected bank. This common memory can be used to reduce code size and execution times as the bank selection instructions that are normally required to access data in banked memory are not required when accessing the common memory. PIC18 devices refer to this memory as the access bank memory. Mid-range and baseline devices have very small amounts of this memory, if it is present at all. PIC18 devices have substantially more common memory, but the amount differs between devices. See your device data sheet for more information. The near type qualifier (or __near) can be used to place global variables in common memory. This qualifier cannot be used with auto or static local objects. The compiler automatically uses the common memory for frequently accessed user-defined variables so this qualifier would only be needed for special memory placement of objects, for example if C variables are accessed in hand-written assembly code that assumes that they are located in this memory. This qualifier is controlled by the compiler option --ADDRQUAL, which determines its effect, see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”. Based on this option’s settings, this qualifier can be binding or ignored (which is the default operation). Qualifiers which are ignored will not produce an error or warning, but will have no effect. Here is an example of an unsigned char object qualified as near: near unsigned char fred; Note that the compiler can store some temporary objects in the common memory, so not all of this space can be available for user-defined variables. If the xc8 option, --STRICT is used, the near qualifier is no longer available, but you can continue to use __near. 2012-2013 Microchip Technology Inc. DS50002053D-page 165 MPLAB® XC8 C Compiler User’s Guide 5.4.8.3 FAR TYPE QUALIFIER The far type qualifier (or __far) is used to place global variables into the program memory space for those PIC18 devices that can support external memory. It will be ignored when compiling for PIC10/12/16 targets. This qualifier cannot be used with auto or static local objects. The compiler assumes that far variables will be located in RAM that is implemented in the external memory space. Access of far variables are less efficient than that of internal variables and will result in larger, slower code. This qualifier is controlled by the compiler option --ADDRQUAL, which determines its effect on PIC18 devices, see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”. Based on this option’s settings, this qualifier can be binding or ignored (which is the default operation). Qualifiers which are ignored will not produce an error or warning, but will have no effect. Here is an example of an unsigned int object placed into the device’s external program memory space: far unsigned int farvar; If the --STRICT is used, you can only use the __far form of the qualifier. Note that not all PIC18 devices support external memory in their program memory space and, thus, the far qualifier is not applicable to all PIC18 devices. On supported devices, the address range where the additional memory will be mapped must first be specified with the --RAM option, Section 4.8.49 “--RAM: Adjust RAM Ranges”. For example, to map additional data memory from 20000h to 2FFFFh use --RAM=default,+20000-2FFFF. 5.4.8.4 BANK0, BANK1, BANK2 AND BANK3 TYPE QUALIFIERS The bank0, bank1, bank2 and bank3 type qualifiers are recognized by the compiler and allow some degree of control of the placement of objects in the device’s data memory banks. When compiling for PIC18 targets, these qualifiers are only accepted for portability and have no effect on variable placement; on other devices they can be used to define C objects that are assumed to be located in certain memory banks by hand-written assembly code. The compiler automatically allocates variables to all data banks, so these qualifiers are not normally needed. Although a few devices implement more than 4 banks of data RAM, bank qualifiers to allow placement into these upper banks are not currently available. These qualifiers are controlled by the compiler option --ADDRQUAL, which determines their effect, see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”. Based on this option’s settings, these qualifiers can be binding or ignored (which is the default operation). Qualifiers which are ignored will not produce an error or warning, but will have no effect. Objects qualified with any of these qualifiers cannot be auto or parameters to a function, but can be qualified static, allowing them to be defined locally within a function, as in: void myFunc(void) { static bank1 unsigned char play_mode; If the xc8 option, --STRICT is used, these qualifiers are changed to 0, 1, 2 and 3. DS50002053D-page 166 2012-2013 Microchip Technology Inc. C Language Features 5.4.8.5 EEPROM TYPE QUALIFIER The eeprom type qualifier (or __eeprom) is recognized by the compiler for baseline and mid-range devices only and indicates that objects should be placed in the EEPROM memory. Not all devices implement EEPROM memory. Check your device data sheet for more information. Objects qualified with this qualifier cannot be auto or parameters to a function, but can be qualified static, allowing them to be defined locally within a function, as in: void myFunc(void) { static eeprom unsigned char inputData[3]; See Section 5.5.5 “Variables in EEPROM” for more information on these variables and other ways of accessing the EEPROM. If the --STRICT option is used, only the __eeprom form of this qualifier is available. 2012-2013 Microchip Technology Inc. DS50002053D-page 167 MPLAB® XC8 C Compiler User’s Guide 5.5 MEMORY ALLOCATION AND ACCESS There are two broad groups of RAM-based variables: auto/parameter variables, which are allocated to some form of stack, and global/static variables, which are positioned freely throughout the data memory space at static locations. The memory allocation of these two groups is discussed separately in the following sections. 5.5.1 Address Spaces All 8-bit PIC devices have a Harvard architecture, which has a separate data memory (RAM) and program memory space (often flash). Some devices also implement EEPROM. The data memory uses banking to increase the amount of available memory (referred to in the data sheets as the general purpose register file) without having to increase the assembly instruction width. One bank is “selected” by setting one or more bits in an SFR. (Consult your device data sheet for the exact operation of the device you are using.) Most instructions which access a data address use only the offset into the currently selected bank to access data. The exception is the PIC18 instruction MOVFF, which takes a full banked address and operates independently of the selected bank. Some devices only have one bank but many have more than one. Both the general purpose RAM and SFRs both share the same data space and can appear in all available memory banks. PIC18 devices have all SFRs in the one data bank, but mid-range and baseline devices have SFRs at the lower addresses of each bank. Due to the location of SFRs in these devices, the general purpose memory becomes fragmented and this limits the size of most C objects. The Enhanced mid-range devices overcome this fragmentation by allowing a linear addressing mode, which allows the general purpose memory to be accessed as one contiguous chunk. Thus, when compiling for these devices, the maximum allowable size of objects typically increases. Objects defined when using PIC18 devices can also typically use the entire data memory. See Section 5.5.2.2.2 “Software Stack Operation” and Section 5.5.2.1.2 “Non-Auto Variable Size Limits”. Many devices have several bytes which can be accessed regardless of which bank is currently selected. This memory is called common memory. The PIC18 data sheets refer to the bank in which this memory is stored as the access bank, and hence it is often referred to as the access bank memory. Since no code is required to select a bank before accessing these locations, access to objects in this memory is typically faster and produces smaller code. The compiler always tries to use this memory if possible. The program memory space is primarily for executable code, but data can also be located here. There are several ways the different device families locate and read data from this memory, but all objects located here will be read-only. DS50002053D-page 168 2012-2013 Microchip Technology Inc. C Language Features 5.5.2 Variables in Data Space Memory Most variables are ultimately positioned into the data space memory. The exceptions are non-auto variables which are qualified as const, which are placed in the program memory space, or eeprom qualified variables. Due to the fundamentally different way in which auto variables and non-auto variables are allocated memory, they are discussed separately. To use the C language terminology, these two groups of variables are those with automatic storage duration and those with permanent storage duration, respectively. Note: 5.5.2.1 The terms “local” and “global” are commonly used to describe variables, but are not ones defined by the language Standard. The term “local variable” is often taken to mean a variable which has scope inside a function, and “global variable” is one which has scope throughout the entire program. However, the C language has three common scopes: block, file (i.e., internal linkage) and program (i.e., external linkage), so using only two terms to describe these can be confusing. For example, a static variable defined outside a function has scope only in that file, so it is not globally accessible, but it can be accessed by more than one function inside that file, so it is not local to any one function, either. NON-AUTO VARIABLE ALLOCATION Non-auto variables (those with permanent storage duration) are located by the compiler into any of the available data banks. This is done in a two-stage process: placing each variable into an appropriate psect and later linking that psect into a predetermined bank. See Section 5.15.1 “Program Sections” for an introductory guide to psects. Thus, during compilation, the code generator can determine which bank will hold each variable and encode the output accordingly, but it will not know the exact location within that bank. The compiler will attempt to locate all variables in one bank (i.e., place all variables in the psect destined for this bank), but if this fills (i.e., if the compiler detects that the psect has become too large for the free space in a bank), variables will be located in other banks via different psects. Qualifiers are not required to have these variables placed in banks other than bank 0 but can be used if you want to force a variable to a particular bank. See “--RAM=default,+20000-2FFFF.” and Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers” for more information on how to do this. If common memory is available on the target device, this will also be considered for variables. This memory can be limited in size and can be reserved for special use, so only a few variables can be allocated to it. 2012-2013 Microchip Technology Inc. DS50002053D-page 169 MPLAB® XC8 C Compiler User’s Guide The compiler considers three categories of non-auto variables, which all relate to the value the variable should contain by the time the program begins. Each variable category has a corresponding psect which is used to hold the output code which reserves memory for each variable. The base name of each psect category is tabulated below. A full list of all psect names are in Section 5.15.2 “Compiler-Generated Psects”. nv These psects are used to store variables qualified persistent, whose values should not be altered by the runtime startup code. They are not cleared or otherwise modified at startup. bss These psects contain any uninitialized variables, which are not assigned a value when they are defined, or variables which should be cleared by the runtime startup code. data These psects contain the RAM image of any initialized variables, which are assigned a non-zero initial value when they are defined and which must have a value copied to them by the runtime startup code. As described in Section 5.10 “Main, Runtime Startup and Reset”, the base name of data space psects is always used in conjunction with a linker class name to indicate the RAM bank in which the psect will be positioned. This section also lists other variants of these psects and indicates where these psect must be linked. See also Section 5.15.2 “Compiler-Generated Psects” for more information on how initial values are assigned to the variables. Note that the data psect used to hold initialized variables is the psect that holds the RAM variables themselves. There is a corresponding psect (called idata) that is placed into program memory (so it is non-volatile) and which is used to hold the initial values that are copied to the RAM variables by the runtime startup code. All non-auto variables, except for static variables, discussed in Section 5.5.2.1.1 “Static Variables”, always use their lexical name with a leading underscore character as the assembly identifier used for this object. See Section 5.12.3.1 “Equivalent Assembly Symbols” for more information on the mapping between C- and assembly-domain symbols. 5.5.2.1.1 Static Variables All static variables have permanent storage duration, even those defined inside a function which are “local static” variables. Local static variables only have scope in the function or block in which they are defined, but unlike auto variables, their memory is reserved for the entire duration of the program. Thus they are allocated memory like other non-auto variables. Static variables can be accessed by other functions via pointers since they have permanent duration. Variables which are static are guaranteed to retain their value between calls to a function, unless explicitly modified via a pointer. Variables which are static and which are initialized only have their initial value assigned once during the program’s execution. Thus, they can be preferable over initialized auto objects which are assigned a value every time the block in they are defined begins execution. Any initialized static variables are initialized in the same way as other non-auto initialized objects by the runtime startup code, see Section 4.4.2 “Startup and Initialization”. The assembly symbols used to access static objects in assembly code are discussed in Section 5.12.3.1 “Equivalent Assembly Symbols”. DS50002053D-page 170 2012-2013 Microchip Technology Inc. C Language Features 5.5.2.1.2 Non-Auto Variable Size Limits Arrays of any type (including arrays of aggregate types) are fully supported by the compiler. So too are the structure and union aggregate types, see 5.4.4 “Structures and Unions”. These objects can often become large in size and can affect memory allocation. When compiling for enhanced mid-range PIC devices, the size of an object (array or aggregate object) is typically limited only by the total available data memory. Single objects that will not fit into any of the available general purpose RAM ranges will be allocated memory in several RAM banks and accessed using the device’s linear GPR (general purpose RAM). Note that the special function registers (which reside in the data memory space) or memory reservations in general purpose RAM can prevent objects from being allocated contiguous memory in the one bank. In this case objects that are smaller than the size of a RAM bank can also be allocated across multi-banks. The generated code to access multi-bank objects will always be slower and the associated code size will be larger than for objects fully contained within a single RAM bank. When compiling for PIC18 devices, the size of an object is also typically limited only by the data memory available. Objects can span several data banks. On baseline and other mid-range devices, arrays and structures are limited to the maximum size of the available GPR memory in each RAM bank, not the total amount of memory remaining. An error will result if an array is defined which is larger than this size. With any device, reserving memory in general purpose RAM (see Section 4.8.49 “--RAM: Adjust RAM Ranges”), or defining absolute variables in the middle of data banks (see Section 5.5.4 “Absolute Variables”), further restricts the contiguous memory in the data banks and can reduce the maximum size of objects you can define. 5.5.2.1.3 Changing the Default Non-Auto Variable Allocation There are several ways in which non-auto variables can be located in locations other than those chosen by the compiler. Variables can be placed in other memory spaces by the use of qualifiers. For example if you wish to place variables in the program memory space, then the const specifier should be used (see Section 5.4.7.1 “Const Type Qualifier”). The eeprom qualifier (see 5.4.8.5 “Eeprom Type Qualifier”) can be used to allocate variables to the EEPROM, if such memory exists on your target device. If you wish to prevent variables from using one or more data memory locations so that these locations can be used for some other purpose, you are best reserving the memory using the memory adjust options. See Section 4.8.49 “--RAM: Adjust RAM Ranges” for information on how to do this. If only a few non-auto variables are to be located at specific addresses in data space memory, then the variables can be made absolute. This allows individual variables to be explicitly positioned in memory at an absolute address. Absolute variables are described in Section 5.5.4 “Absolute Variables”. Once variables are made absolute, their address is hard coded in generated output code, they are no longer placed in a psect and do not follow the normal memory allocation procedure. 2012-2013 Microchip Technology Inc. DS50002053D-page 171 MPLAB® XC8 C Compiler User’s Guide The psects in which the different categories of non-auto variables (the nv, bss and data psects described in Section 5.5.2.1 “Non-Auto Variable Allocation”) can be shifted as a whole by changing the default linker options. So, for example, you could move all the persistent variables. However, typically these psects can only be moved within the data bank in which they were allocated by default. See Section 5.10 “Main, Runtime Startup and Reset”, for more information on changing the default linker options for psects. The code generator makes assumptions as to the location of these psects and if you move them to a location that breaks these assumptions, code can fail. Non-auto can also be placed at specific positions by using the psect pragma, see Section 5.14.4.8 “The #pragma psect Directive”. The decision whether variables should be positioned this way or using absolute variables should be based on the location requirements. 5.5.2.2 AUTO VARIABLE ALLOCATION AND ACCESS This section discusses allocation of auto variables (those with automatic storage duration) to a data stack. This also includes function parameter variables, which behave like auto variables, in terms of their storage duration and scope. Temporary variables defined by the compiler also fall into this group. They are identical to auto variables, except they are defined by the compiler and, hence, have no C name. Together, these objects are often called stack-based objects. The auto (short for automatic) variables are the default type of local variable. Unless explicitly declared to be static, a local variable will be made auto. The auto keyword can be used if desired. The auto variables, as their name suggests, automatically come into existence when a function is executed, then disappear once the function returns. Since they are not in existence for the entire duration of the program, there is the possibility to reclaim memory they use when the variables are not in existence and allocate it to other variables in the program. Typically such variables are stored on some sort of a dynamic data stack where memory can be easily allocated and deallocated by each function. This is not possible on all 8-bit devices supported by MPLAB XC8. Nor is it the most efficient means of storing objects. MPLAB XC8 has two methods of implementing data stacks for stack-based variables: a compiled stack and a software stack1. Section 5.3.4 “Stacks” describes all the stacks used by MPLAB XC8 and the 8-bit PIC devices. Each C function is compiled to use exactly one of these stacks. The stack used affects whether a function allows reentrancy. If a function is encoded to place its stack-based objects on the software stack, it is said to be using a reentrant function model. A function uses a non-reentrant function model if it places its stack-based objects on the compiled stack. This information is summarized in Table 5-9 along with the devices that support each model. The function model directly implies the stack used by a function. See subsections below for specific details on how the compiled stack and software stack operate. TABLE 5-9: FUNCTION MODELS IMPLEMENTATION Function Model Data stack used Supported device families Non-reentrant Compiled stack All devices Reentrant Software stack Enhanced mid-range and PIC18 devices 1. What is referred to as a software stack in this user’s guide is the typical dynamic stack arrangement employed by most computers and is ordinary data memory accessed by some sort of push and pop instructions, and a stack pointer register. DS50002053D-page 172 2012-2013 Microchip Technology Inc. C Language Features When compiling for those devices that do not support the reentrant function model, all functions are encoded to use the compiler stack, and these functions are non-reentrant. For the enhanced mid-range and PIC18 devices, by default the compiler will use the non-reentrant model for all functions. Alternatively the user can dictate which functions are to be compiled reentrantly (and those which are not) by using compiler options or function specifiers. There is also a hybrid stack mode which allows the compiler to choose which functions need to be compiled using a reentrant model and which can use the non-reentrant model. The hybrid mode allows the program to use recursion but still take advantage of the more efficient compiled stack. The --STACK option (see Section 4.8.56 “--STACK: Specify Data Stack Type For Entire Program”) can be used to change the compiler’s default behavior when assigning function models. Set the --STACK option to software so the compiler will always choose the reentrant model (software stack) for each function. If the --STACK option is set to compiled or this option is omitted, all functions are encoded to use the non-reentrant (compiled stack) function model. Set this option to hybrid for hybrid stack mode and to allow the compiler to decide how each function should be encoded. In hybrid mode the compiler will choose a function model based on how the function is called in the program. If the function is not reentrantly called, then it will be encoded to use the non-reentrant model and the compiled stack. If the function appears in more than one call graph (i.e., it is called from main-line and interrupt code), or it appears in a loop in a call graph (i.e., it is called recursively), then the compiler will use the reentrant model. The --STACK option’s software and compiled settings changes the function model for all functions. You can change the function model for individual functions by using function specifiers when you define the function. Use either the compiled or nonreentrant specifier (identical meanings) to indicate that the specified function must use the compiled stack, without affecting any other function. Alternatively, use either the software or reentrant specifier to indicate a function must be encoded to use the software stack. The function specifiers have precedence over the --STACK option setting. If, for example, the option --STACK=compiled has been used, but one function uses the software (or reentrant) specifier, then the specified function will use the software stack and all the remaining functions will use the compiled stack. These functions specifiers also override any choice made by the compiler in hybrid mode. If a function has been specified as compiled (or nonreentrant), or the --STACK=compiled option has been issued, and that function appears in more than one call graph in the program, then the usual function duplication feature automatically comes into effect. See Section 5.9.6 “Function Duplication”, for more information on how this is performed. Duplicating a non-reentrant function allows it to be called from multiple call graphs, but cannot be used if the function is called recursively. The auto variables defined in a function will not necessarily be allocated memory in the order declared, in contrast to parameters which are always allocated memory based on their lexical order. In fact, auto variables for one function can be allocated in many RAM banks. The standard qualifiers: const and volatile can both be used with auto variables and these do not affect how they are positioned in memory. This implies that a local const-qualified object is still an auto object and, as such, will be allocated memory in the compiled stack in the data space memory, not in the program memory like with non-auto const objects. 2012-2013 Microchip Technology Inc. DS50002053D-page 173 MPLAB® XC8 C Compiler User’s Guide The compiler will try to locate the stack in one data bank, but if this fills (i.e., if the compiler detects that the stack psect has become too large), it can build up the stack into several components (each with their own psect) and link each in a different bank. Each auto object is referenced in assembly code using a special symbol defined by the code generator. If accessing auto variables defined in C source code, you must use these symbols, which are discussed in Section 5.12.3 “Interaction between Assembly and C Code”. 5.5.2.2.1 Compiled Stack Operation A compiled stack consists of fixed memory areas that are usable by each function’s stack-based variables. When a compiled stack is used, functions are not re-entrant since stack-based variables in each function will use the same fixed area of memory every time the function is invoked. Fundamental to the generation of the compiled stack is the call graph, which defines a tree-like hierarchy of function calls, i.e it shows what functions can be called by each function. There will be one graph produced for each root function. A root function is typically not called, but which is executed via other means and contains a program entry point. The function main() is an example of a root function that will be in every project. Interrupt functions which are executed when a hardware interrupt occurs, are another example. FIGURE 5-1: FORMATION OF CALL GRAPH main { F1(…); F2(…); F3(…); } Call graph main F1 F4 F1 { F4(…); } isr { F5(…); F6(…); } code generator F2 F3 isr F5 F6 Analysis of program Figure 5-1 shows sections of a program being analyzed by the code generator to form a call graph. In the original source code, the function main() calls F1(), F2() and F3(). F1() calls F4(), but the other two functions make no calls. The call graph for main() indicates these calls. The symbols F1, F2 and F3 are all indented one level under main. F4 is indented one level under F1. DS50002053D-page 174 2012-2013 Microchip Technology Inc. C Language Features This is a static call graph which shows all possible calls. If the exact code for function F1() looked like: int F1(void) { if(PORTA == 44) return F4(); return 55; } the function F4() will always appear in the call graph, even though it is conditionally executed in the actual source code. Thus, the call graph indicates all functions that might be called. In the diagram, there is also an interrupt function, isr(), and it too has a separate graph generated. The term main-line code is often used, and refers to any code that is executed as a result of the main() function being executed. In the above figure, F1(), F2(), F3() and F4() are only ever called by main-line code. The term interrupt code refers to any code that is executed as a result of an interrupt being generated, in the above figure, F5() and F6() are called by interrupt code. Figure 5-2 graphically shows an example of how the compiled stack is formed. FIGURE 5-2: FORMATION OF THE COMPILED STACK 1 Formation of auto-parameter block (APB) for function F2 int aa , int int bb ) { F2(int ; char c a b c F2 2 Analysis of call graph } main F1 compiled stack main F4 F3 F 3 1 F1 F2 F 2 F4 4 isr F5 5 3 F2 F3 isr F5 F6 F6 F 6 Overlap of non-concurrently active APBs to form compiled stack Each function in the program is allocated a block of memory for its parameter, auto and temporary variables. Each block is referred to as an auto-parameter block (APB). The figure shows the APB being formed for function F2(), which has two parameters, a and b, and one auto variable, c. The parameters to the function are first grouped in an order strictly determined by the lexical order in which they appear in the source code. These are then followed by any auto objects; however, the auto objects can be placed in any order. So we see memory for a is followed by that for b and lastly c. 2012-2013 Microchip Technology Inc. DS50002053D-page 175 MPLAB® XC8 C Compiler User’s Guide Once these variables have been grouped, the exact location of each object is not important at this point and we can represent this memory by one block — the APB for this function. The APBs are formed for all functions in the program. Then, by analyzing the call graph, these blocks are assigned positions, or base values, in the compiled stack. Memory can be saved if the following point is observed: If two functions are never active at the same time, then their APBs can be overlapped. In the example shown in the figure, F4() and F1() are active at the same time, in fact F1() calls F4(). However, F2(), F3() and F1() are never active at the same time; F1() must return before F2() or F3() can be called by main(). The function main() will always be active and so its APB can never overlap with that of another function. In the compiled stack, you can see that the APB for main() is allocated unique memory. The blocks for F1(), F2() and F3() are all placed on top of each other and the same base value in the compiled stack; however, the memory taken up by the APBs for F1() and F4() are unique and do not overlap. Our example also has an interrupt function, isr(), and its call graph is used to assemble the APBs for any interrupt code in the same way. Being the root of a graph, isr() will always be allocated unique memory, and the APBs for interrupt functions will be allocated memory following. The end result is a block of memory which forms the compiled stack. This block can then be placed into the device’s memory by the linker. Once auto variables have been allocated a relative position in the compiled stack, the stack itself is then allocated memory in the data space. This is done is a similar fashion to the way other variables are assigned memory: a psect is used to hold the stack and this psect is placed into the available data memory by the linker. The psect base name used to hold the compiled stack is called cstack, and, like with other psects, the base name is always used in conjunction with a linker class name to indicate the RAM bank in which the psect will be positioned. See Section 5.15.2 “Compiler-Generated Psects” for the limitations associated with where this psect can be linked. For devices with more than one bank of data memory, the compiled stack can be built up into components, each located in a different memory bank. The compiler will try to allocate the compiled stack in one bank, but if this fills, it will consider other banks. The process of building these components of the stack is the same, but each function can have more than one APB and these will be allocated to one of the stack components based on the remaining memory in the component’s destination bank. Human readable symbols are defined by the code generator which can be used to access auto and parameter variables in the compiled stack from assembly code, if required. See Section 5.12.3 “Interaction between Assembly and C Code” for full information between C domain and assembly domain symbols. 5.5.2.2.2 Software Stack Operation Functions using a software stack (reentrant model) dynamically allocate memory for their stack-based variables in a region of memory specifically reserved for this software stack. Allocation starts at one end of this reserved area, and the stack memory grows as new function instances come into existence. When a function using the reentrant model exits, any stack memory it used is freed and made available for other functions. The stack grows up in memory, toward larger addresses. Main-line code and each interrupt routine use unique areas in the stack space. The maximum size allocated to each area can be specified using the --STACK option, see Section 4.8.56 “--STACK: Specify Data Stack Type For Entire Program”. DS50002053D-page 176 2012-2013 Microchip Technology Inc. C Language Features MPLAB XC8 designates a register, known as the stack pointer, which always holds the address of the next free location in the software stack. The register used by the stack pointer is FSR1 for both enhanced mid-range and PIC18 devices. The address held by the stack pointer is increased when variables are allocated (pushed) to the stack; it is decreased when a function returns and variables are removed (popped) from the stack. Note that if there are any functions in the program that are reentrantly encoded, the FSR1 register is reserved for the stack pointer for the entire program’s duration, even when executing code associated with non-reentrant functions. With this register unavailable for use with general statements, the code generated may be less efficient or "Can’t generate code" errors may result. The stack pointer is reloaded when an interrupt occurs so it accesses the interrupt function’s unique stack area. It is restored by the interrupt context switch code when the interrupt routine is complete. There is no register assigned to hold a frame pointer. All access of stack-based objects must use an address that is an offset from the stack pointer. When a function is called, any arguments to that function are pushed onto the stack by the calling function, in a reverse order to that in which the corresponding parameters appear in the function’s prototype. If required, the called function will increase the value stored in the stack pointer to allocate storage for any auto or temporary variables it needs to allocate. If the reentrant function returns a value on the stack (this might happen for return values larger than 4 bytes in size), the calling function will adjust the stack to remove the return value. Recall that a function’s return address is not stored on this stack. It is automatically stored on the hardware stack by the device, see Section 5.3.4.1 “Function Return Address Stack”. The compiler can detect if the software stack memory requirements for each function will exceed set limits. These limits are 256 bytes for PIC18 devices and typically 31 bytes for enhanced mid-range devices. Note that the compiler cannot detect for overflow of the memory reserved for the stack as a whole. There is no runtime check made for software stack overflows. If the software stack overflows, data corruption and code failure might result. 5.5.2.2.3 Size Limits of Auto Variables The compiled stack is built up as one contiguous block which can be placed into one of the available data banks. However, if the stack becomes too large for this space, it can be assembled into several blocks, with each block being positioned in a different bank of memory. Thus the total size of the stack is roughly limited only by the available memory on the device. Unlike with non-auto variables, it is not efficient to access auto variables within the compiled stack using the linear memory of enhanced mid-range devices. For all devices, including PIC18 and Enhanced mid-range PIC MCUs, each component of the compiled stack must fit entirely within one bank of data memory on the target device (however, you can have more than one component, each allocated to a different bank). This limits the size of objects within the stack to the maximum free space of the bank in which it is allocated. The more auto variables in the stack; the more restrictive the space is to large objects. Recall that SFRs on mid-range devices are usually present in each data bank, so the maximum amount of GPR available in each bank is typically less than the bank size for these devices. 2012-2013 Microchip Technology Inc. DS50002053D-page 177 MPLAB® XC8 C Compiler User’s Guide The software stack is always allocated one block of memory. This memory may cross bank boundaries. The size is typically limited by the amount of free data space remaining. An auto object placed on the software stack may be any size, providing it fits in the allocated stack space. It may be allocated memory that crosses a bank boundary and will always be accessed via the stack pointer, FSR1. If a program requires large objects that should not be accessible to the entire program, consider leaving them as local objects, but using the static specifier. Such variables are still local to a function, but are no longer auto and have fewer size limitations. They are allocated memory as described in Section 5.5.2.1 “Non-Auto Variable Allocation”. 5.5.2.2.4 Changing the Default Auto Variable Allocation As auto variables are placed in a stack, there is no means to move individual variables. They cannot be made absolute, nor can they be moved using the #pragma psect directive. The psects in which the auto variables reside can be shifted as a whole by changing the default linker options. However, these psects can only be moved within the data bank in which they were allocated by default. See Section 5.10 “Main, Runtime Startup and Reset” for more information on changing the default linker options for psects. The code generate makes assumptions as to the location of these psects and if you move them to a location that breaks these assumptions, code can fail. 5.5.3 Variables in Program Space The only variables that are placed into program memory are those that are not auto and which have been qualified const. Any auto variables qualified const are placed in the compiled stack along with other auto variables, and all components of the compiled stack will only ever be located in the data space memory. Any const-qualified (auto or non-auto) variable will always be read-only and any attempt to write to these in your source code will result in an error being issued by the compiler. On some 8-bit PIC devices, the program space is not directly readable by the device. For these devices, the compiler stores data in the program memory by means of RETLW instructions which can be called, and which will return a byte of data in the W register. The compiler will generate the code necessary to make it appear that program memory is being read directly. Enhanced mid-range PIC devices can directly read their program memory, although the compiler will still usually store data as RETLW instructions. This way the compiler can either produce code that can call these instructions to obtain the program memory data as with the ordinary mid-range devices, or directly read the operand to the instruction (the LSB of the RETLW instruction). The most efficient access method can be selected by the compiler when the data needs to be read. DS50002053D-page 178 2012-2013 Microchip Technology Inc. C Language Features Data can be stored as individual bytes in the program memory of PIC18 devices. This can be read using table read instructions. On all devices, accessing data located in program memory is much slower than accessing objects in the data memory. The code associated with the access is also larger. A const object is usually defined with initial values, as the program cannot write to these objects at runtime. However, this is not a requirement. An uninitialized const object can be defined to define a symbol, or label, but not make a contribution to the output file. Uninitialized const objects are often made absolute, see Section 5.5.4 “Absolute Variables”. Here are examples of const object definitions. const char IOtype = ’A’; const char buffer[10]; // initialized const object // I just define a label The data held by non-auto const variables is placed in one of several psects, based on the target device. See Section 5.15.2 “Compiler-Generated Psects” for the limitations associated with where these psects can be linked. See Section 5.12.3 “Interaction between Assembly and C Code” for the equivalent assembly symbols that are used to represent const-qualified variables in program memory. 5.5.3.1 SIZE LIMITATIONS OF CONST VARIABLES Arrays of any type (including arrays of aggregate types) can be qualified const and placed in the program memory. So too can structure and union aggregate types, see 5.4.4 “Structures and Unions”. These objects can often become large in size and can affect memory allocation. For baseline PIC devices, the maximum size of a single const object is 255 bytes. However, you can define as many const objects as required provided the total size does not exceed the available program memory size of the device. Note that as well as other program code, there is also code required to be able to access const-qualified data in the program memory space. Thus, you can need additional program memory space over the size of the object itself. This additional code to access the const data is only included once, regardless of the amount or number of const-qualified objects. For all other 8-bit devices, the maximum size of a const-qualified object is limited only by the available program memory. These devices also use additional code that accesses the const data. PIC18 devices need additional code each time an object is accessed, but this is typically small. The mid-range devices include a larger routine, but this code is also only included once, regardless of the amount or number of const-qualified objects. 2012-2013 Microchip Technology Inc. DS50002053D-page 179 MPLAB® XC8 C Compiler User’s Guide 5.5.3.2 CHANGING THE DEFAULT ALLOCATION If you only intend to prevent all variables from using one or more program memory locations so that you can use those locations for some other purpose, you are best reserving the memory using the memory adjust options. See Section 4.8.50 “--ROM: Adjust ROM Ranges” for information on how to do this. If only a few non-auto const variables are to be located at specific addresses in program space memory, then the variables can be made absolute. This allows individual variables to be explicitly positioned in memory at an absolute address. Absolute variables are described in Section 5.5.4 “Absolute Variables”. Once variables are made absolute, their address is hard coded in generated output code, they are no longer placed in a psect and do not follow the normal memory allocation procedure. The psects in which the different categories of non-auto const variables can be shifted as a whole by changing the default linker options. However, there are limitations in where these psects can be moved to. See Section 5.10 “Main, Runtime Startup and Reset” for more information on changing the default linker options for these psects. Variables in program memory can also be placed at specific positions by using the psect pragma, see Section 5.14.4.8 “The #pragma psect Directive”. The decision whether variables should be positioned this way or using absolute variables should be based on the location requirements. DS50002053D-page 180 2012-2013 Microchip Technology Inc. C Language Features 5.5.4 Absolute Variables Most variables can be located at an absolute address by following its declaration with the construct @ address, where address is the location in memory where the variable is to be positioned. Such a variable is known as an absolute variable. 5.5.4.1 ABSOLUTE VARIABLES IN DATA MEMORY Absolute variables are primarily intended for equating the address of a C identifier with a special function register, but can be used to place ordinary variables at an absolute address in data memory. For example: volatile unsigned char Portvar @ 0x06; will declare a variable called Portvar located at 06h in the data memory. The compiler will reserve storage for this object (if the address falls into general-purpose RAM) and will equate the variable’s identifier to that address. The auto variables cannot be made absolute as they are located in a stack. Nor can you make static local objects absolute. The compiler does not make any checks for overlap of absolute variables with other absolute variables, so this must be considered when choosing the variable locations. There is no harm in defining more than one absolute variable to live at the same address if this is what you require. The compiler will not locate ordinary variables over the top of absolutes, so there is no overlap between these objects. Note: Defining absolute objects can fragment memory and can make it impossible for the linker to position other objects. Avoid absolute objects if at all possible. If absolute objects must be defined, try to place them at either end of a memory bank or page so that the remaining free memory is not fragmented into smaller chunks. When defining absolute bit variables (see Section 5.4.2.1 “Bit Data Types and Variables”), the address specified must be a bit address. A bit address is obtained by multiplying the desired byte address by 8, then adding the bit offset within that bit. So, for example, to place a bit variable called mode at bit position #2 at byte address 0x50, use the following: bit mode @ 0x282; If you wish to place a bit variable over an existing object (typically this will be an SFR variable or another absolute variable) then you can use the symbol of that object, as in the following example which places flag at bit position #3 in the char variable MOT_STATUS: bit flag @ ((unsigned) &MOT_STATUS)*8 + 3; When compiling for an enhanced mid-range PIC device, the memory allocated for some objects can be spread over multiple RAM banks. Such objects will only ever be accessed indirectly in assembly code, and will use the linear GPR memory implemented on these devices. A linear address (which can be mapped back to the ordinary banked address) will be used with these objects internally by the compiler. 2012-2013 Microchip Technology Inc. DS50002053D-page 181 MPLAB® XC8 C Compiler User’s Guide The address specified for absolute objects on these devices can either be the traditional banked memory address or the linear address. As the linear addresses start above the largest banked address, it is clear which address is intended. In the following example: int inputBuffer[100] @ 0x2000; it is clear that inputBuffer should placed at address 0x2000 in the linear address space, which is address 0x20 in bank 0 RAM in the traditional banked address space. See the device data sheet for exact details regarding your selected device. Absolute variables in RAM cannot be initialized when they are defined. Define the absolute variables, then assign them a value at a suitable point in your main-line code. 5.5.4.2 ABSOLUTE OBJECTS IN PROGRAM MEMORY Non-auto objects qualified const can also be made absolute in the same way, however, the address will indicate an address in program memory. For example: const int settings[] @ 0x200 = { 1, 5, 10, 50, 100 }; will place the array settings at address 0x200 in the program memory. Both initialized and uninitialized const objects can be made absolute. That latter is useful when you only need to define a label in program memory without making a contribution to the output file. Variables can also be placed at specific positions by using the psect pragma, see Section 5.14.4.8 “The #pragma psect Directive”. The decision whether variables should be positioned this way or using absolute variables should be based on the location requirements. Using absolute variables is the easiest method, but only allows placement at an address which must be known prior to compilation. The psect pragma is more complex, but offers all the flexibility of the linker to position the new psect into memory. You can, for example, specify that variables reside at a fixed address, or that they be placed after other psects, or that the they be placed anywhere in a compiler-defined or user-defined range of address. DS50002053D-page 182 2012-2013 Microchip Technology Inc. C Language Features 5.5.5 Variables in EEPROM For devices with on-chip EEPROM, the compiler offers several methods of accessing this memory. You can defined named variables in this memory space, or use block-access routines to read or write EEPROM. The EEPROM access methods are described in the following sections. 5.5.5.1 EEPROM VARIABLES When compiling for baseline and mid-range parts, the eeprom qualifier allows you to create named C variables that reside in the EEPROM space. See Section 5.4.8.5 “Eeprom Type Qualifier”. Variables qualified eeprom are cleared or initialized, just like ordinary RAM-based variables; however, the initialization process is not carried out by the runtime startup code. Initial values are placed into the HEX file and are burnt into the EEPROM when you program the device. Thus, if you modify the EEPROM during program execution and then reset the device, these variables will not contain the initial values specified in your code at startup up. The following example defines two arrays in EEPROM. eeprom char regNumber[10] = "A93213"; eeprom int lastValues[3]; For both these objects, initial values will be placed into psects and will appear in the HEX file. Zeros will be used as the initial values for lastValues. The generated code to access eeprom-qualified variables will be much longer and slower than code to access RAM-based variables. You should avoid using eeprom-qualified variables in complicated expressions. Consider copying values from the EEPROM to regular RAM-based variables and using these in your code. 5.5.5.2 EEPROM INITIALIZATION For those devices that support external programming of their EEPROM data area, the __EEPROM_DATA() macro can be used to place initial values into the HEX file ready for programming. The macro is used as follows. #include __EEPROM_DATA(0, 1, 2, 3, 4, 5, 6, 7); The macro has eight parameters, representing eight data values. Each value should be a byte in size. Unused values should be specified with zero. The __EEPROM_DATA() macro expands into in-line assembly code. If expressions are used to evaluate the macro arguments, ensure that any operators or tokens in these expressions are written in assembly code (see Section 6.4 “MPLAB XC8 Assembly Language”). The macro can be called multiple times to define the required amount of EEPROM data. It is recommended that the macro be placed outside any function definition. This macro cannot used to write to EEPROM locations during runtime; it is used for pre-loading EEPROM contents at program time only. The values defined by this macro share the EEPROM space with any eeprom-qualified variables. The macro cannot be used to initialize eeprom-qualified variables. The psect used by this macro to hold the data values is different to those used by eeprom-qualified variables. The link order of these psects can be adjusted, if required (see Section 4.8.6 “-L-: Adjust Linker Options Directly”). For convenience, the macro _EEPROMSIZE represents the number of bytes of EEPROM available on the target device. 2012-2013 Microchip Technology Inc. DS50002053D-page 183 MPLAB® XC8 C Compiler User’s Guide 5.5.5.3 EEPROM ACCESS FUNCTIONS The library functions eeprom_read() and eeprom_write(), can be called to read from, and write to, the EEPROM during program execution. On PIC18 devices, these functions are derived from the peripheral library. The prototypes for these functions are as below. #include unsigned char eeprom_read(unsigned char address); void eeprom_write(unsigned char address, unsigned char value); These functions test and wait for any concurrent writes to EEPROM to conclude before performing the required operation. The eeprom_write() function will initiate the process of writing to EEPROM and this process will not have completed by the time that eeprom_write() returns. The new data written to EEPROM will become valid at a later time. See your device data sheet for exact information about EEPROM on your target device. It can also be convenient to use the preprocessor symbol, _EEPROMSIZE, in conjunction with some of these access methods. This symbol defines the number of EEPROM bytes available for the selected chip. 5.5.5.4 EEPROM ACCESS MACROS Macro version of the EEPROM functions are also provided. The PIC18 version of these macros purely call the function equivalents. Those for other 8-bit PIC devices perform similar operations to their function counterparts, with the exception of some timing issues described below. Use the macro forms of these routines for faster execution and to save a level of stack, but note that their repeated use will increase code size. The usage of these macros for all devices is as follows. EEPROM_READ(address) EEPROM_WRITE(address, value) The EEPROM_READ macro returns the byte read. In the case of the baseline and mid-range macro EEPROM_READ(), there is another very important difference from the function version to note. Unlike eeprom_read(), this macro does not wait for any concurrent EEPROM writes to complete before proceeding to select and read EEPROM. If it cannot be guaranteed that all writes to EEPROM have completed at the time of calling EEPROM_READ(), the appropriate flag should be polled prior to executing EEPROM_READ(). For example: xc.h // wait for end-of-write before EEPROM_READ while(WR) continue; // read from EEPROM at address value = EEPROM_READ(address); DS50002053D-page 184 2012-2013 Microchip Technology Inc. C Language Features 5.5.6 Variables in Registers Allocating variables to registers, rather than to a memory location, can make code more efficient. With MPLAB XC8, there is no direct control of placement of variables in registers. The register keyword (which can only be used with auto variables) is silently ignored and has no effect on memory allocation of variables. There are very few registers available for caching of variables on PIC baseline and mid-range devices, and as these registers must be frequently used by generated code for other purposes, there is little advantage in using them. The cost involved in loading variables into registers would far outweigh any advantage of accessing the register. At present, code compiled for PIC18 devices also does not utilize registers other than that described below. Some arguments are passed to functions in the W register rather than in a memory location; however, these values will typically be stored back to memory by code inside the function so that W can be used by code associated with that function. See Section 5.8.5 “Function Size Limits” for more information as to which parameter variables can use registers. 5.5.7 Dynamic Memory Allocation Dynamic memory allocation, (heap-based allocation using malloc, etc.) is not supported on any 8-bit device. This is due to the limited amount of data memory, and that this memory is banked. The wasteful nature of dynamic memory allocation does not suit itself to the 8-bit PIC device architectures. 5.5.8 Memory Models MPLAB XC8 C Compiler does not use fixed memory models to alter allocation of variables to memory. Memory allocation is fully automatic and there are no memory model controls. 2012-2013 Microchip Technology Inc. DS50002053D-page 185 MPLAB® XC8 C Compiler User’s Guide 5.6 OPERATORS AND STATEMENTS The MPLAB XC8 C Compiler supports all the ANSI operators. The exact results of some of these are implementation defined. Implementation-defined behavior is fully documented in Appendix D. “Implementation-Defined Behavior”. The following sections illustrate code operations that are often misunderstood as well as additional operations that the compiler is capable of performing. 5.6.1 Integral Promotion When there is more than one operand to an operator, they typically must be of exactly the same type. The compiler will automatically convert the operands, if necessary, so they do have the same type. The conversion is to a “larger” type so there is no loss of information; however, the change in type can cause different code behavior to what is sometimes expected. These form the standard type conversions. Prior to these type conversions, some operands are unconditionally converted to a larger type, even if both operands to an operator have the same type. This conversion is called integral promotion and is part of Standard C behavior. The compiler performs these integral promotions where required, and there are no options that can control or disable this operation. If you are not aware that the type has changed, the results of some expressions are not what would normally be expected. Integral promotion is the implicit conversion of enumerated types, signed or unsigned varieties of char, short int or bit-field types to either signed int or unsigned int. If the result of the conversion can be represented by an signed int, then that is the destination type, otherwise the conversion is to unsigned int. Consider the following example. unsigned char count, a=0, b=50; if(a - b < 10) count++; The unsigned char result of a - b is 206 (which is not less than 10), but both a and b are converted to signed int via integral promotion before the subtraction takes place. The result of the subtraction with these data types is -50 (which is less than 10) and hence the body of the if() statement is executed. If the result of the subtraction is to be an unsigned quantity, then apply a cast. For example: if((unsigned int)(a - b) < 10) count++; The comparison is then done using unsigned int, in this case, and the body of the if() would not be executed. Another problem that frequently occurs is with the bitwise complement operator, ~. This operator toggles each bit within a value. Consider the following code. unsigned char count, c; c = 0x55; if( ~c == 0xAA) count++; If c contains the value 0x55, it often assumed that ~c will produce 0xAA; however, the result is 0xFFAA and so the comparison in the above example would fail. The compiler can be able to issue a mismatched comparison error to this effect in some circumstances. Again, a cast could be used to change this behavior. DS50002053D-page 186 2012-2013 Microchip Technology Inc. C Language Features The consequence of integral promotion as illustrated above is that operations are not performed with char -type operands, but with int -type operands. However, there are circumstances when the result of an operation is identical regardless of whether the operands are of type char or int. In these cases, the compiler will not perform the integral promotion so as to increase the code efficiency. Consider this example. unsigned char a, b, c; a = b + c; Strictly speaking, this statement requires that the values of b and c should be promoted to unsigned int, the addition performed, the result of the addition cast to the type of a, and then the assignment can take place. Even if the result of the unsigned int addition of the promoted values of b and c was different to the result of the unsigned char addition of these values without promotion, after the unsigned int result was converted back to unsigned char, the final result would be the same. If an 8-bit addition is more efficient than a 16-bit addition, the compiler will encode the former. If, in the above example, the type of a was unsigned int, then integral promotion would have to be performed to comply with the ANSI C standard. 5.6.2 Rotation The C language does not specify a rotate operator; however, it does allow shifts. The compiler will detect expressions that implement rotate operations using shift and logical operators and compile them efficiently. For the following code: c = (c << 1) | (c >> 7); if c is unsigned and non-volatile, the compiler will detect that the intended operation is a rotate left of 1 bit and will encode the output using the PIC MCU rotate instructions. A rotate left of 2 bits would be implemented with code like: c = (c << 2) | (c >> 6); This code optimization will also work for integral types larger than a char. If the optimization cannot be applied, or this code is ported to another compiler, the rotate will be implemented, but typically with shifts and a bitwise OR operation. 5.6.3 Switch Statements The compiler can encode switch statements using one of several strategies. By default, the compiler chooses a strategy based on the case values that are used inside the switch statement. Each switch statement is assigned its strategy independently. The type of strategy can be indicated by using the #pragma switch directive. See Section 5.14.4.10 “The #pragma switch Directive”, which also lists the available strategy types. There can be more than one strategy associated with each type. There is information printed in the assembly list file for each switch statement detailing the value being switched and the case values listed. See Section 6.6.4 “Switch Statement Information”. 2012-2013 Microchip Technology Inc. DS50002053D-page 187 MPLAB® XC8 C Compiler User’s Guide 5.7 REGISTER USAGE The assembly generated from C source code by the compiler will use certain registers in the PIC MCU register set. Most importantly, the compiler assumes that nothing other than code it generates can alter the contents of these registers. So, if compiler-generated assembly code loads a register with a value and no subsequent code requires this register, the compiler will assume that the contents of the register are still valid later in the output sequence. If any of the applicable registers listed are used by interrupt code, they will be saved and restored when an interrupt occurs, either in hardware or software. See Section 5.9.4 “Context Switching”. The registers that are special and which are used by the compiler are listed in Table 5-10 TABLE 5-10: REGISTERS USED BY THE COMPILER Applicable devices Register name All 8-bit devices W All 8-bit devices STATUS All mid-range devices PCLATH All PIC18 devices PCLATH, PCLATU Enhanced mid-range and PIC18 devices BSR Non-enhanced mid-range devices FSR Enhanced mid-range and PIC18 devices FSR0L, FSR0H, FSR1L, FSR1H All PIC18 devices FSR2L, FSR2H All PIC18 devices TBLPTRL, TBLPTRH, TBLPTRU, TABLAT All PIC18 devices PRODL, PRODH Enhanced mid-range and PIC18 devices btemp, wtemp, ttemp, ltemp The xtemp registers are variables that the compiler treats as registers, These are saved like any other register if they are used in interrupt code. The state of these registers must never be changed directly by C code, or by any assembly code in-line with C code. The following example shows a C statement and in-line assembly that violates these rules and changes the ZERO bit in the STATUS register. #include void getInput(void) { ZERO = 0x1; // do not write using C code c = read(); #asm bcf ZERO ; do not write using inline assembly code #endasm process(c); } MPLAB XC8 is unable to interpret the register usage of in-line assembly code that is encountered in C code. Nor does it associate a variable mapped over an SFR to the actual register itself. Writing to an SFR register using either of these two methods will not flag the register as having changed and can lead to code failure. DS50002053D-page 188 2012-2013 Microchip Technology Inc. C Language Features 5.8 FUNCTIONS Functions are written in the usual way, in accordance with C language. Implementation and special features associated with functions are discussed in following sections. 5.8.1 Function Specifiers Functions can, in the usual way, use the standard specifier static. A function defined using the static specifier only affects the scope of the function; i.e., limits the places in the source code where the function can be called. Functions that are static can only be directly called from code in the file in which the function is defined. The equivalent symbol used in assembly code to represent the function can change if the function is static, see 5.12.3 “Interaction between Assembly and C Code”. This specifier does not change the way the function is encoded. Non-standard qualifiers are discussed below. 5.8.1.1 INTERRUPT SPECIFIER The interrupt specifier indicates that the function is an interrupt service routine and that it is to be encoded specially to suit this task. Interrupt functions are described in detail in 5.9.1 “Writing an Interrupt Service Routine”. 5.8.1.2 INLINE SPECIFIER The inline function specifier is a recommendation that calls to the specified function be as fast as possible. The compiler can be able to inline the body of the function specified if certain conditions are met. The following is an example of a function which has been made a candidate for inlining. inline int combine(int x, int y) { return 2*x-y; } All function calls to any function that was inlined by the compiler will be encoded as if the call was replaced with the body of the called function. This is performed at the assembly code level. Inlining will only take place if the assembly optimizers are enabled. The function itself can still be encoded by the compiler even if it is inlined. If inlining takes place, this will increase the program’s execution speed, since the call and return sequences associated with the call will be eliminated. It will also reduce the hardware stack usage as no call instruction is actually executed. Any stack reduction is not reflected in the call graphs shown in the assembly list file as this file is generated before inlining takes place. If inlining takes place, code size can be reduced if the assembly code associated with the body of the inlined function is very small and the function itself is not output. Code size will increase if the body of the inlined function is larger than the call/return sequence it replaces and that function is called more than once. You should only consider this specifier for functions which generate small amounts of assembly code. Note that the amount of C code in the body of a function is not a good indicator of the size of the assembly code which it generates (see Section 3.6.13 “How Can I Tell How Big a Function Is?”). A function cannot be inlined if it itself contains in-line assembly. If the assembly for the function contains certain assembly sequences, this can also prevent inlining of the function. A warning will be generated if the function references static objects, to comply with the ANSI Standard. A warning is also issued if it is not inlined successfully. Your code should not make any assumption about whether inlining was successful and which assembly code associated with the function is being executed. This specifier performs the same task as the #pragma inline directive, see Section 5.14.4.4 “The #pragma Intrinsic Directive”. 2012-2013 Microchip Technology Inc. DS50002053D-page 189 MPLAB® XC8 C Compiler User’s Guide 5.8.1.3 REENTRANT AND NONREENTRANT SPECIFIERS The reentrant and nonreentrant function specifiers indicate the function model (stack) that should be used for that function’s stack-based variables (auto, parameter, and temporary variables), as shown in Table 5-11. The aliases software and compiled, respectively, can also be used. If the --STRICT option has been enabled (see Section 4.8.57 “--STRICT: Strict ANSI Conformance”) these specifiers must use two leading underscore characters, (e.g., __reentrant). TABLE 5-11: STACK RELATED FUNCTION SPECIFIERS Specifier Allocation for Stack-based variables compiled, nonreentrant Always use the compiled stack; functions are non-reentrant. software, reentrant Use the software stack, if available; functions are reentrant. You would only use these specifiers if the default allocation of a function’s stack-based variables is unacceptable. These specifiers override any setting indicated using the --STACK option, see Section 4.8.56 “--STACK: Specify Data Stack Type For Entire Program”. If no specifier or --STACK option has been used, all functions are encoded as non-reentrant and use the compiled stack. The following shows an example of a function that will always be encoded as reentrant. reentrant int setWriteMode(int mode) { if(mode != 3) mode = 0; return mode; } The reentrant specifier only has an effect if the target device supports a software stack. In addition, not all functions allow reentrancy. Interrupt functions and main() must always use the compiled stack, but functions they call may use the software stack. Functions encoded for baseline and mid-range devices always use the non-reentrant model and the compiled stack. Repeated use of the software (reentrant) specifier will increase substantially the size of the software stack leading to possible overflow. The size of the software stack is not accurately known at compile time, so the compiler cannot issue a warning if it is likely to overwrite memory used for some other purpose. The stack may overwrite other sections of the program in data memory, or memory used by something outside the program, such as hardware or another independently-compiled application. See Section 5.3.4.2 “Data Stacks” for device specific information relating to the data stacks available on each device. DS50002053D-page 190 2012-2013 Microchip Technology Inc. C Language Features 5.8.2 External Functions If a call to a function that is defined outside the program C source code is required (it can be part of code compiled separately, e.g., the bootloader, or in assembly code), you will need to provide a declaration of the function so that the compiler knows how to encode the call. If this function takes arguments or returns a value, the compiler can use a symbol to represent the memory locations used to store these values, see Section 5.8.6 “Function Parameters” and Section 5.8.6.2 “Software Stack Parameters” to determine if a register or memory locations are used in this transfer. Usually, the compiler defines this symbol when it encodes the C function, but if the function is external and not encoded by the compiler, then the symbol value must be manually defined. If an argument or return value is used and this will be stored in memory, the corresponding symbol must be defined by your code and assigned the value of the appropriate memory location. The value can be determined from the map file of the external build, which compiled the function, or from the assembly code. If the function was written in C, look for the symbol ?_funcName, where funcName is the name of the function. It can be defined in the program which makes the call via a simple EQU directive in assembler. For example, the following snippet of code could be placed in the C source: #asm GLOBAL ?_extReadFn ?_extReadFn EQU 0x20 #endasm Alternatively, the assembly code could be contained directly in an assembly module. This defines the base address of the parameter area for an extern function extReadFn to be 0x20. If this symbol is not defined, the compiler will issue an undefined symbol error. This error can be used to verify the name being used by the compiler to encode the call, if required. It is not recommended to call the function indirectly by casting an integer to a function pointer, but in such a circumstance, the compiler will use the value of the constant in the symbol name, for example calling a function at address 0x200 will require the definition of the symbol ?0x200 to be the location of the parameter/return value location for the function. For example: #asm GLOBAL ?0x200 ?0x200 EQU 0x55 #endasm Note that the return value of a function (if used) shares the same locations assigned to any parameters to that function and both use the same symbol. 2012-2013 Microchip Technology Inc. DS50002053D-page 191 MPLAB® XC8 C Compiler User’s Guide 5.8.3 Allocation of Executable Code Code associated with functions is always placed in the program memory of the target device. On baseline and mid-range devices, the program memory is paged (compare: banking used in the data memory space). This memory is still sequential (addresses are contiguous across a page boundary), but the paging means that any call or jump from code in one page to a label in another must use a longer sequence of instructions to accomplish this. See your device data sheet for more information on the program memory and instruction set. PIC18 devices do not implement any program memory paging. The CALL and GOTO instruction are two-word instructions and their destinations are not limited. The relative branch instructions have a limited range, but this is not based on any paging boundaries. The generated code associated with each function is initially placed in its own psect by the compiler, see Section 5.15.1 “Program Sections”. These psects have names such as textn, where n is a number, e.g., text98. However, psects can be merged later in the compilation process so that more than one function can contribute to a psect. When the program memory is paged, functions within the same psect can use a shorter form of call and jump to labels so it is advantageous to merge the code for as many functions into the same psect. These text psects are linked anywhere in the program memory (see 5.10 “Main, Runtime Startup and Reset”). If the size of a psect that holds the code associated with a function exceeds the size of a page, it can be split by the assembler optimizer. A split psect will have a name of the form textn_split_s. So, for example, if the text102 psect exceeds the size of a page, it can be split into a text102_split_1 and a text102_split_2 psect. This process is fully automatic, but you should be aware that if the code associated with a function does become larger than one page in size, the efficiency of that code can drop fractionally due to any longer jump and call instruction sequences being used to transfer control to code in other pages. The base name of each psect category is tabulated below. A full list of all program-memory psects psect names are listed in Section 5.15.2.1 “Program Space Psects”. maintext The generated code associated with the special function, main, is placed in this psect. Some optimizations and features are not applied to this psect. textn These psects (where n is a decimal number) contain all other executable code that does not require a special link location. DS50002053D-page 192 2012-2013 Microchip Technology Inc. C Language Features 5.8.4 Changing the Default Function Allocation You can change the default memory allocation of functions several ways. If you intend only to prevent functions from using one or more program memory locations so that you can use those locations for some other purpose, you are best reserving the memory using the memory adjust options. See Section 4.8.50 “--ROM: Adjust ROM Ranges”, for information on how to do this. The assembly code associated with a C function can be placed at an absolute address. This can be accomplished by using an @ address construct in a similar fashion to that used with absolute variables. Such functions are called absolute functions. The following example of an absolute function will place the function at address 400h: int mach_status(int mode) @ 0x400 { /* function body */ } If you check the assembly list file you will see the function label and the first assembly instruction associated with the function located at 0x400. You can use either the assembly list file (see 6.5 “Assembly-Level Optimizations”) or the map file (see 7.4 “Map Files”) to confirm that the function was moved as you expect. If this construct is used with interrupt functions it will only affect the position of the code associated with the interrupt function body. The interrupt context switch code that precedes the function code will not be relocated, as it must be linked to the interrupt vector. See also Section 4.8.20 “--CLIST: Generate C Listing File”, for information on how to move Reset and interrupt vector locations – which can be useful for designing applications such as bootloaders. Unlike absolute variables, the generated code associated with absolute functions is still placed in a psect, but the psect is dedicated to that function only. The psect name has the form below. A full list of all psect names are listed in Section 5.10 “Main, Runtime Startup and Reset”. xxx_text Defines the psect for a function that has been made absolute; i.e., placed at an address. xxx will be the assembly symbol associated with the function. For example if the function rv() is made absolute, code associated with it will appear in the psect called _rv_text. Functions can be allocated to a user-defined psect using the __section() specifier (see Section 5.15.4 “Changing and Linking the Allocated Section”). This new psect can then be linked at the required location. As with absolute functions, this specifier will only affect the position of the code associated with the interrupt function body. The interrupt entry code will still be located in the default psect and be linked at the usual interrupt vector address. Functions can also be placed at specific positions by using the psect pragma, see Section 5.14.4.8 “The #pragma psect Directive”, although using the __section() directive is an easier option. The decision whether functions should be positioned by using the pragma, specifier, or by making them absolute should be based on the location requirements. Using absolute functions is the easiest method, but only allows placement at an address that must be known prior to compilation. The __section() specifier offers all the flexibility of the linker to position the new psect into memory. For example, you can specify that functions reside at a fixed address, or that they be placed after other psects, or that the they be placed anywhere in a compiler-defined or user-defined range of addresses. 2012-2013 Microchip Technology Inc. DS50002053D-page 193 MPLAB® XC8 C Compiler User’s Guide 5.8.5 Function Size Limits For all devices, the code generated for a function is only limited only by the available program memory. Functions can become larger than one page in size on paged devices; however, these functions cannot be as efficient due to longer call sequences to jump to and call destinations in other pages. See 5.8.3 “Allocation of Executable Code” for more details. 5.8.6 Function Parameters MPLAB XC8 uses a fixed convention to pass arguments to a function. The method used to pass the arguments depends on the size and number of arguments involved, and on which stack model is used with the function. Note: 5.8.6.1 The names “argument” and “parameter” are often used interchangeably, but typically an argument is the actual value that is passed to the function and a parameter is the variable defined by the function to store the argument. COMPILED STACK PARAMETERS For functions using the non-reentrant model, the compiler will either pass arguments in the W register, or in the called function’s parameter memory. If the first parameter is one byte in size, it is passed in the W register. All other parameters are passed in the parameter memory. This applies to basic types and to aggregate types, like structures. The parameters are grouped along with the function’s auto variables in the parameter memory and are placed in the compiled stack. See Section 5.5.2.2.1 “Compiled Stack Operation”, for detailed information on the compiled stack. The parameter variables will be referenced as an offset from the symbol ?_function, where function is the name of the function in which the parameter is defined (i.e., the function that is to be called). Unlike auto variables, parameter variables are allocated memory strictly in the order in which they appear in the function’s prototype. This means that the parameters will always be placed in the same memory bank. The auto variables for a function can be allocated across multiple banks and in any order. The parameters for variadic functions that take a variable argument list (defined using an ellipsis in the prototype and which are called non-prototyped parameters) are placed in the parameter memory, along with named parameters. Take, for example, the following ANSI-style function. void test(char a, int b); The function test() will receive the parameter b in its function auto-parameter block and a in the W register. A call to this function: test(xyz, 8); would generate code similar to: MOVLW MOVWF CLRF MOVF CALL 08h ?_test ?_test+1 _xyz,w (_test) ; ; ; ; move literal 0x8 into... the auto-parameter memory locations for the 16-bit parameter move xyz into the W register In this example, the parameter b is held in the memory locations ?_test (LSB) and ?_test+1 (MSB). DS50002053D-page 194 2012-2013 Microchip Technology Inc. C Language Features The exact code used to call a function, or the code used to access a parameters from within a function, can always be examined in the assembly list file. See Section 4.8.16 “--ASMLIST: Generate Assembler List Files” for the option that generates this file. This is useful if you are writing an assembly routine that must call a function with parameters, or accept arguments when it is called. The above example does not consider data memory banking or program memory paging, which can require additional instructions. 5.8.6.2 SOFTWARE STACK PARAMETERS When a function uses the reentrant model, most arguments to that function will be passed on the software stack. Parameters placed on the software stack are pushed in the reverse order to which they were defined in the called function’s prototype. This is unlike auto variables, which may be allocated memory in any order. The W register is sometimes used for the first function argument if it is byte-sized and the function uses the reentrant model. This will only take place for enhanced mid-range devices and provided the function is not variadic and returns a value in btemp registers (see Section 5.8.7.2 “Software Stack Return Values”Section 5.8.7.2 “Software Stack Return Values”). The W register is never used by reentrant function arguments when compiling for PIC18 devices. For variadic functions, which take a variable argument list (defined using an ellipsis in the prototype), the unprototyped parameters are placed on the software stack, before the named parameters. After all the function’s arguments have been pushed, the total size of the non-prototyped parameters is pushed on to the stack (except if this function has a return value which is returned on the stack). A maximum of 256 bytes of non-prototyped parameters are permitted per function. Recall that the function return address is not stored on this data stack. It is automatically stored on the hardware stack by the device, see Section 5.3.4.1 “Function Return Address Stack”. As there is no frame pointer, accessing function parameters, or other stack-based objects, is not recommended in hand-written assembly code. 2012-2013 Microchip Technology Inc. DS50002053D-page 195 MPLAB® XC8 C Compiler User’s Guide 5.8.7 Function Return Values Values returned from functions are loaded into a register or placed on the stack used by that function. The mechanism will depend on the function model used by the function. 5.8.7.1 COMPILED STACK RETURN VALUES For functions that use the non-reentrant model, return values are passed to the calling function using the W register, or the function’s parameter memory. Re-using the memory used by the parameters (which is no longer needed when the function is ready to return) can reduce the code and data requirements for functions. Eight-bit values are returned from a function in the W register. Values larger than a byte are returned in the function’s parameter memory area, with the least significant word (lsw) in the lowest memory location. For example, the function: int return_16(void) { return 0x1234; } will exit with the code similar to: MOVLW MOVWF MOVLW MOVWF RETURN 34h (?_return_16) 12h (?_return_16)+1 For PIC18 targets returning values greater than 4 bytes in size, the address of the parameter area is also placed in the FSR0 register. Functions that return a bit do so using the carry bit of the STATUS register. 5.8.7.2 SOFTWARE STACK RETURN VALUES Functions that use the reentrant model will pass values back to the calling function via btemp variables, provided the value is 4 bytes or less in size. The W register will be used to return byte-sized values for enhanced mid-range device functions that are not variadic. For objects larger than 4 bytes in size, they are returned on the stack. Reentrant PIC18 functions that return a bit do so using bit #0 in btemp0; other devices use the carry bit in the STATUS register. As there is no frame pointer, accessing the return value location, or other stack-based objects, is not recommended in hand-written assembly code. DS50002053D-page 196 2012-2013 Microchip Technology Inc. C Language Features 5.8.8 Calling Functions All 8-bit devices use a hardware stack for function return addresses. The depth of this stack varies from device to device. Typically, CALL assembly instructions are used to transfer control to a C function when it is called. Each function calls uses one level of stack. This stack level is freed after the called routine executes a RETURN instruction. The stack usage grows if a called function calls another before returning. If the hardware stack overflows, function return addresses will be destroyed and the code will eventually fail. The stackcall suboption to the --RUNTIME option controls how the compiler behaves when the compiler detects that the hardware stack is about to overflow due to too many nested calls. See Section 4.8.51 “--RUNTIME: Specify Runtime Environment” for details on this option. If this suboption is disabled (the default state), where the depth of the stack will be exceeded by a call, the compiler will issue a warning to indicate that this is the case. For PIC18 devices, this is the only way in which calls are made, but for other 8-bit devices, the compiler can swap to an alternate way of making calls, as detailed below. If the stackcall suboption is enabled, the compiler will, instead of issuing a warning, automatically swap to using a method that involves the use of a lookup table and which does not require use of the hardware stack. This feature is not available for PIC18 devices. When the lookup method is being employed, a function is reached by a jump (not a call) directly to its address. Before this is done the address of a special “return” instruction (implemented as a jump instruction) is stored in a temporary location inside the called function. This return instruction will be able to return control back to the calling function. This means of calling functions allows functions to be nested deeply without overflowing the limited stack available on baseline and mid-range devices; however, it does come at the expense of memory and program speed. 5.8.8.1 BANK SELECTION WITHIN FUNCTIONS A function can, and can, return with any RAM bank selected. See Section 5.5.1 “Address Spaces” for more information on RAM banks. The compiler tracks the bank selections made in the generated code associated with each function, even across function calls to other functions. If the bank that is selected when a function returns can be determined, the compiler will use this information to try to remove redundant bank selection instructions which might otherwise be inserted into the generated code. The compiler will not be able to track the bank selected by routines written in assembly, even if they are called from C code. The compiler will make no assumptions about the selected bank when such routines return. The “Tracked objects” section associated with each function and which is shown in the assembly list file relates to this bank tracking mechanism. See 6.5 “Assembly-Level Optimizations” for more information of the content of these files. 2012-2013 Microchip Technology Inc. DS50002053D-page 197 MPLAB® XC8 C Compiler User’s Guide 5.9 INTERRUPTS The MPLAB XC8 compiler incorporates features allowing interrupts to be fully handled from C code. Interrupt functions are often called Interrupt Service Routines, or ISRs. Note: Baseline devices do not utilize interrupts and so the following sections are only applicable for mid-range, Enhanced mid-range and PIC18 devices. There is only one interrupt vector on mid-range and Enhanced mid-range devices. Regardless of the source of the interrupt, the device will vector to one specific location in program memory and execution continues from that address. This address is a attribute of the device and cannot be changed. Each mid-range device interrupt source typically has a control flag in an SFR which can disable that interrupt source. In addition there is a global interrupt enable flag that can disable all interrupts sources and ensure that an interrupt can never occur. There is no priority of interrupt sources. Check your device data sheet for full information how your device handles interrupts. PIC18 devices have two separate interrupt vectors and a priority scheme to dictate which interrupt code is executed. The two interrupts are designated as low and high priority. Peripherals are associated one of the interrupt priorities (vectors) through settings in the peripheral’s SFRs. Interrupt functions always use the non-reentrant function model. These functions ignore any option or function specifier that might otherwise specify reentrancy. Individual interrupt sources can be disabled via a control flag in an SFR associated with that interrupt source. In addition to the global interrupt enable flag, there are other flags that can disable each interrupt priority. Interrupt code is the name given to any code that executes as a result of an interrupt occurring, including functions called from the ISR and library code. Interrupt code completes at the point where the corresponding return from interrupt instruction is executed. This contrasts with main-line code, which, for a freestanding application, is usually the main part of the program that executes after Reset. 5.9.1 Writing an Interrupt Service Routine The function qualifier interrupt (or __interrupt) can be applied to a C function definition so that it will be executed once the interrupt occurs. The compiler will process the interrupt function differently to any other functions, generating code to save and restore any registers used and return using a special instruction. If the xc8 option --STRICT is used, you must use the __interrupt form of the keyword. An interrupt function must be declared as type void interrupt and cannot have parameters. This is the only function prototype that makes sense for an interrupt function since they are never directly called in the source code. On PIC18 devices, interrupt functions default to being high priority. To create a low-priority interrupt function, use the qualifier low_priority in addition to interrupt in the function definition. Interrupt functions must not be called directly from C code (due to the different return instruction that is used), but they themselves can call other functions, both user-defined and library functions. There can be many sources of interrupt that share the same interrupt vector, but there is only ever one interrupt function associated with each vector. The interrupt function must then contain code to determine the source of the interrupt before proceeding. An error will result if there are more interrupt functions than interrupt vectors in a program. DS50002053D-page 198 2012-2013 Microchip Technology Inc. C Language Features An example of an interrupt function is shown here. int tick_count; void interrupt tc_int(void) { if (TMR0IE && TMR0IF) { TMR0IF=0; ++tick_count; return; } // process other interrupt sources here, if required } Code generated by the compiler will be placed at the interrupt vector address which will execute this function after any context switch that is required. Notice that the code in the interrupt function checks for the source of the interrupt, in this case a timer, by looking at the interrupt enable bit (TMR0IE) and the interrupt flag bit (TMR0IF). Checking the interrupt enable flag is required since interrupt flags associated with a peripheral can be asserted even if the peripheral is not configured to generate an interrupt. The following is an example of a low priority interrupt function that could be written for PIC18 devices. void interrupt low_priority tc_clr(void) { if (TMR1IE && TMR1IF) { TMR1IF=0; tick_count = 0; return; } // process any other low priority sources here } 5.9.2 Changing the Default Interrupt Function Allocation Moving the code associated with interrupt functions is more difficult than that for ordinary functions, as an interrupt routine has an entry point strictly defined by the device. If you require the interrupt functions and the code associated with their entry points to be moved up in memory, then use the --CODEOFFSET option, see Section 4.8.21 “--CODEOFFSET: Offset Program Code to Address”. You might do this, for example, if you are writing a bootloader or bootloader application that must be remapped in memory. You can use the __section() specifier (see Section 5.15.4 “Changing and Linking the Allocated Section”) if you want to move the bulk of the interrupt code, but leave the interrupt entry point at the default interrupt vector location. A jump instruction will be placed in the interrupt psect that will jump to the remainder of the interrupt code, which is placed in the user-defined psect indicated in this specifier. 5.9.3 Specifying the Interrupt Vector The interrupt function(s) cannot be changed at runtime. That is, you cannot have alternate interrupt functions and select which will be active during program execution. An error will result if there are more interrupt functions than interrupt vectors in a program. 2012-2013 Microchip Technology Inc. DS50002053D-page 199 MPLAB® XC8 C Compiler User’s Guide 5.9.4 Context Switching 5.9.4.1 CONTEXT SAVING ON INTERRUPTS Some registers are automatically saved by the hardware when an interrupt occurs. Any registers or compiler temporary objects used by the interrupt function, other than those saved by the hardware, must be saved in code generated by the compiler. This is the context save, or context switch code. See Section 5.7 “Register Usage” for the registers that must be saved and restored either by hardware or software when an interrupt occurs. Enhanced mid-range PIC devices save the W, STATUS, BSR and FSRx registers in hardware (using special shadow registers) and hence these registers do not need to be saved by software. The only register that can need to be saved is BTEMP11, a compiler temporary location that acts like a pseudo register. This makes interrupt functions on Enhanced mid-range PIC devices very fast and efficient. Other mid-range PIC processors only save the entire PC (excluding the PCLATH register) when an interrupt occurs. The W, STATUS, FSR and PCLATH registers and the BTEMP1 pseudo register must be saved by code produced by the compiler, if required. By default, the PIC18 high-priority interrupt function will utilize its internal shadow register to save the W, STATUS and BSR registers. All other used registers are saved in software. Note that for some older devices, the compiler will not use the shadow registers if compiling for the MPLAB ICD debugger, as the debugger itself utilizes these shadow registers. Some errata workarounds also prevent the use of the shadow registers see Section 4.8.26 “--ERRATA: Specify Errata Workarounds”. For the low priority PIC18 interrupts, or when the shadow registers cannot be used, all registers that has been used by the interrupt code will be saved by software. The compiler determines exactly which registers and objects are used by an interrupt function, or any of the functions that it calls (based on the call graph generated by the compiler), and saves these appropriately. Assembly code placed in-line within the interrupt function is not scanned for register usage. Thus, if you include in-line assembly code into an interrupt function, you can have to add extra assembly code to save and restore any registers or locations used. The same is true for any assembly routines called by the interrupt code. If the W register is to be saved by the compiler, it can be stored to memory reserved in the common RAM. If the device for which the code is written does not have common memory, a byte is reserved in all RAM banks for the storage location for W register. Most registers to be saved are allocated memory in the interrupt function’s auto area. They can be treated like any other auto variable and use the same assembly symbols. On mid-range devices, the W register is stored in BTEMP0, a pseudo register, see Section 5.7 “Register Usage”. If the software stack is in use, the context switch code will also initialize the stack pointer register so it is accessing the area of the stack reserved for the interrupt. See Section 5.5.2.2.2 “Software Stack Operation”, for more information on the software stack. 1. The BTEMP register is a memory location allocated by the compiler, but it is treated like a register for code generation purposes. It is not used by all devices. DS50002053D-page 200 2012-2013 Microchip Technology Inc. C Language Features 5.9.4.2 CONTEXT RESTORATION Any objects saved by software are automatically restored by software before the interrupt function returns. The order of restoration is the reverse to that used when context is saved. If the software stack is in use, the context restoration code will also restore the stack pointer register so that it is accessing the area of the stack used before the interrupt occurred. See Section 5.5.2.2.2 “Software Stack Operation”, for more information on the software stack. 5.9.5 Enabling Interrupts Two macros are available, once you have included , which control the masking of all available interrupts. These macros are ei(), which enable or unmask all interrupts, and di(), which disable or mask all interrupts. On all devices, they affect the GIE bit in the INTCON register. These macros should be used once the appropriate interrupt enable bits for the interrupts that are required in a program have been enabled. For example: ADIE = 1; PEIE = 1; ei(); // ... di(); Note: // A/D interrupts will be used // all peripheral interrupts are enabled // enable all interrupts // disable all interrupts Never re-enable interrupts inside the interrupt function itself. Interrupts are automatically re-enabled by hardware on execution of the RETFIE instruction. Re-enabling interrupts inside an interrupt function can result in code failure. 2012-2013 Microchip Technology Inc. DS50002053D-page 201 MPLAB® XC8 C Compiler User’s Guide 5.9.6 Function Duplication It is assumed by the compiler that an interrupt can occur at any time. Functions encoded to use the compiled stack are not reentrant (see Section 5.5.2.2.1 “Compiled Stack Operation”), so, if such a function is called by an interrupt function and by main-line code, this could lead to code failure. MPLAB XC8 has a feature which will duplicate the output associated with any function that uses the non-reentrant function model and which is called from more than one call graph. There is one call graph associated with main-line code, and one for each interrupt function, if defined. This allows reentrancy, but recursion is still not possible even if the function is duplicated. Although the compiler can compile functions using a reentrant model, this feature is not available with all devices; it can also be disabled using the --STACK option or the nonreentrant specifier. See Section 5.5.2.2 “Auto Variable Allocation and access”, for information on which function model is chosen for a function. Main-line code will call the original function’s output, and the interrupt will call the duplicated function’s output. The duplication takes place only in the called function’s output; there is no duplication of the C source code itself. The duplicated code and data uses different symbols and are allocated different memory, so are fully independent. This is similar to the process you would need to undertake if this feature was not implemented in the compiler: the C function could be duplicated by hand, given different names and one called from main-line code; the other from the interrupt function. However, you would have to maintain both functions, and the code would need to be reverted if it was ported to a compiler which did support reentrancy. The compiler-generated duplicate will have unique identifiers for the assembly symbols used within it. The identifiers consists of the same name used in the original output prefixed with i1. Duplicated PIC18 functions use the prefixes i1 and i2 for the low- and high-priority interrupts, respectively. The output of the function called from main-line code will not use any prefixes and the assembly names will be those normally used. To illustrate, in a program the function main calls a function called input. This function is also called by an interrupt function. Examination of the assembly list file will show assembly code for both the original and duplicate function outputs. The output corresponding to the C function input() will use the assembly label _input. The corresponding label used by the duplicate function will be i1_input. If the original function makes reference to a temporary variable, the generated output will use the symbol ??_input, compared to ??i1_input for the duplicate output. Even local labels within the function output will be duplicated in the same way. The call graph, in the assembly list file, will show the calls made to both of these functions as if they were independently written. These symbols will also be seen in the map file symbol table. This feature allows the programmer to use the same source code with compilers that use either reentrant or non-reentrant models. It does not handle cases where functions are called recursively. Code associated with library functions are duplicated in the same way. This also applies to implicitly called library routines, such as those that perform division or floating-point operations associated with C operators. 5.9.6.1 DISABLING DUPLICATION The automatic duplication of the function can be inhibited by the use of a special pragma. DS50002053D-page 202 2012-2013 Microchip Technology Inc. C Language Features This should only be done if the source code guarantees that an interrupt cannot occur while the function is being called from any main-line code. Typically this would be achieved by disabling interrupts before calling the function. It is not sufficient to disable the interrupts inside the function after it has been called; if an interrupt occurs when executing the function, the code can fail. See Section 5.9.5 “Enabling Interrupts” for more information on how interrupts can be disabled. The pragma is: #pragma interrupt_level 1 The pragma should be placed before the definition of the function that is not to be duplicated. The pragma will only affect the first function whose definition follows. For example, if the function read is only ever called from main-line code when the interrupts are disabled, then duplication of the function can be prevented if it is also called from an interrupt function as follows. #pragma interrupt_level 1 int read(char device) { // ... } In main-line code, this function would typically be called as follows: di(); // turn off interrupts read(IN_CH1); ei(); // re-enable interrupts The level value specified indicates for which interrupt the function will not be duplicated. For mid-range devices, the level should always be 1; for PIC18 devices it can be 1 or 2 for the low- or high-priority interrupt functions, respectively. To disable duplication for both interrupt priorities, use the pragma twice to specify both levels 1 and 2. The following function will not be duplicated if it is also called from the low- and high-priority interrupt functions. #pragma interrupt_level 1 #pragma interrupt_level 2 int timestwo(int a) { return a * 2; } 2012-2013 Microchip Technology Inc. DS50002053D-page 203 MPLAB® XC8 C Compiler User’s Guide 5.10 MAIN, RUNTIME STARTUP AND RESET The identifier main is special. You must always have one and only one function called main() in your programs. This is the first function to execute in your program. Code associated with main(); however, is not the first code to execute after Reset. Additional code provided by the compiler, and known as the runtime startup code, is executed first and is responsible for transferring control to the main() function. The actions and control of this code is described in the following sections. The compiler inserts special code at the end of main() which is executed if this function ends, i.e., a return statement inside main() is executed, or code execution reaches the main()’s terminating right brace. This special code causes execution to jump to address 0, the Reset vector for all 8-bit PIC devices. This essentially performs a software Reset. Note that the state of registers after a software Reset can be different to that after a hardware Reset. It is recommended that the main() function does not end. Add a loop construct (such as a while(1)) that will never terminate either around your code in main() or at the end of your code, so that execution of the function will never terminate. For example, void main(void) { // your code goes here // finished that, now just wait for interrupts while(1) continue; } 5.10.1 Runtime Startup Code A C program requires certain objects to be initialized and the device to be in a particular state before it can begin execution of its function main(). It is the job of the runtime startup code to perform these tasks, specifically (and in no particular order): • Initialization of global variables assigned a value when defined • Clearing of non-initialized global variables • General setup of registers or device state Rather than the traditional method of linking in a generic, precompiled routine, MPLAB XC8 uses a more efficient method which actually determines what runtime startup code is required from the user’s program. Details of the files used and how the process can be controlled are described in Section 4.4.2 “Startup and Initialization”. The following sections detail exactly what the runtime startup code actually does. The runtime startup code is executed before main(), but If you require any special initialization to be performed immediately after Reset, you should use powerup feature described later in Section 5.10.2 “The Powerup Routine”. The following table lists the significant assembly labels used by the startup and powerup code. TABLE 5-12: SIGNIFICANT ASSEMBLY LABELS Label Location reset_vec At the Reset vector location (0x0) powerup The beginning of the powerup routine, if used start The beginning of the runtime startup code, in startup.as start_initialization DS50002053D-page 204 The beginning of the C initialization startup code, in the C output code. 2012-2013 Microchip Technology Inc. C Language Features 5.10.1.1 INITIALIZATION OF OBJECTS One task of the runtime startup code is to ensure that any initialized variables contain their initial value before the program begins execution. Initialized variables are those which are not auto objects and which are assigned an initial value in their definition, for example input in the following example. int input = 88; void main(void) { ... Such initialized objects have two components: their initial value (0x0088 in the above example) stored in program memory (i.e., placed in the HEX file), and space for the variable reserved in RAM it will reside and be accessed during program execution (runtime). The psects used for storing these components are described in Section 5.15.2 “Compiler-Generated Psects”. The runtime startup code will copy all the blocks of initial values from program memory to RAM so that the variables will contain the correct values before main() is executed. This action can be omitted by disabling the init suboption of --RUNTIME. For example: --RUNTIME=default,-init With this part of the runtime startup code absent, the contents of initialized variables will be unpredictable when the program begins execution. Code relying on variables containing their initial value will fail. Since auto objects are dynamically created, they require code to be positioned in the function in which they are defined to perform their initialization. It is possible that the initial value of an auto object can change on each instance of the function and so the initial values cannot be stored in program memory and copied. As a result, initialized auto objects are not considered by the runtime startup code but are instead initialized by assembly code in each function output. Note: Initialized auto variables can impact on code performance, particularly if the objects are large in size. Consider using global or static objects instead. Variables whose contents should be preserved over a Reset, or even power off, should be qualified with the persistent qualifier, see Section 5.4.8.1 “Persistent Type Qualifier”. Such variables are linked at a different area of memory and are not altered by the runtime startup code in any way. If objects are initialized, the runtime startup code which performs this will destroy the contents of the STATUS register. With some devices, the TO and PD bits in this register are required to determine the cause of Reset. You can choose to have a copy of this register taken so that it can later be examined. See Section 5.10.1.4 “STATUS Register Preservation”, for more information. 2012-2013 Microchip Technology Inc. DS50002053D-page 205 MPLAB® XC8 C Compiler User’s Guide 5.10.1.2 CLEARING OBJECTS Those non-auto objects which are not initialized must be cleared before execution of the program begins. This task is also performed by the runtime startup code. Uninitialized variables are those which are not auto objects and which are not assigned a value in their definition, for example output in the following example. int output; void main(void) {... Such uninitialized objects will only require space to be reserved in RAM where they will reside and be accessed during program execution (runtime). The psects used for storing these components are described in Section 5.15.2 “Compiler-Generated Psects” and typically have a name based on the initialism “bss” (Block Started by Symbol). The runtime startup code will clear all the memory location occupied by uninitialized variables so they will contain zero before main() is executed. Variables whose contents should be preserved over a Reset should be qualified with persistent. See Section 5.4.8.1 “Persistent Type Qualifier” for more information. Such variables are linked at a different area of memory and are not altered by the runtime startup code in any way. If objects are initialized, the runtime startup code that performs this will destroy the contents of the STATUS register. With some devices, the TO and PD bits in this register are required to determine the cause of Reset. You can choose to have a copy of this register taken so that it can later be examined. See Section “If the software stack is being used by the program, the stack pointer (FSR1) is also initialized by the runtime startup code. See Section 5.5.2.2.2 “Software Stack Operation”.” for more information. 5.10.1.3 SETUP OF DEVICE STATE Some PIC devices come with an oscillator calibration constant which is pre-programmed into the device’s program memory. This constant can be written to the OSCCAL register to calibrate the internal RC oscillator, if required. Code is automatically placed in the runtime startup code to load this calibration value, see Section 5.3.11 “Oscillator Calibration Constants”. If the software stack is being used by the program, the stack pointer (FSR1) is also initialized by the runtime startup code. See Section 5.5.2.2.2 “Software Stack Operation”. DS50002053D-page 206 2012-2013 Microchip Technology Inc. C Language Features 5.10.1.4 STATUS REGISTER PRESERVATION The resetbits suboption of the --RUNTIME option (see 4.8.51 “--RUNTIME: Specify Runtime Environment”) preserves some of the bits in the STATUS register before being clobbered by the remainder of the runtime startup code. The state of these bits can be examined after recovering from a Reset condition to determine the cause of the Reset. The entire STATUS register is saved to an assembly variable ___resetbits. This variable can be accessed from C code using the declaration: extern unsigned char __resetbits; The compiler defines the assembly symbols ___powerdown and ___timeout to represent the bit address of the Power-down and Time-out bits within the STATUS register and can be used if required. These can be accessed from C code using the declarations: extern bit __powerdown; extern bit __timeout; In the above symbols, note that the C variables use two leading underscore characters, and the assembly equivalent symbols use three. See Section 5.12.3.1 “Equivalent Assembly Symbols” for more details of the mapping. See Section 4.9 “MPLAB X Option Equivalents” for use of this option in MPLAB IDE. 5.10.2 The Powerup Routine Some hardware configurations require special initialization, often within the first few instruction cycles after Reset. To achieve this there is a hook to the Reset vector provided via the powerup routine. This routine can be supplied in a user-defined assembler module that will be executed immediately after Reset. A template powerup routine is provided in the file powerup.as which is located in the sources directory of your compiler distribution. Refer to comments in this file for more details. The file should be copied to your working directory, modified and included into your project as a source file. No special linker options or other code is required. The compiler will detect if you have defined a powerup routine and will automatically use it, provided the code in this routine is contained in a psect called powerup. For correct operation (when using the default compiler-generated runtime startup code), the code must end with a GOTO instruction to the label called start. As with all user-defined assembly code, any code inside this file must take into consideration program memory paging and/or data memory banking, as well as any applicable errata issues for the device you are using. The program’s entry point is already defined by the runtime startup code, so this should not be specified in the powerup routine with the END directive (if used). See Section 6.4.9.2 “END” for more information on this assembler directive. 2012-2013 Microchip Technology Inc. DS50002053D-page 207 MPLAB® XC8 C Compiler User’s Guide 5.11 LIBRARY ROUTINES 5.11.0.1 USING LIBRARY ROUTINES Library functions (and any associated variables) will be automatically linked into a program once they have been referenced in your source code. The use of a function from one library file will not include any other functions from that library. Only used library functions will be linked into the program output and consume memory. Your program will require declarations for any functions or symbols used from libraries. These are contained in the standard C header (.h) files. Header files are not library files and the two files types should not be confused. Library files contain precompiled code, typically functions and variable definitions; the header files provide declarations (as opposed to definitions) for functions, variables and types in the library files, as well as other preprocessor macros. In the following example, the definition for sqrt is not contained in source code, so the compiler searches the libraries to find a definition there. Once found, it links in the function for sqrt into your program. #include // declare function prototype for sqrt void main(void) { double i; // sqrt referenced; sqrt will be linked in from library file i = sqrt(23.5); } 5.11.1 The printf Routine The code associated with the printf function is not precompiled into the library files. The printf() function is generated from a special C template file that is customized after analysis of the user’s C code. See “PRINTF, VPRINTF” for more information on using the printf library function. The template file is found in the lib directory of the compiler distribution and is called doprnt.c. It contains a minimal implementation of the printf() function, but with the more advanced features included as conditional code which can be utilized via preprocessor macros that are defined when it (along with your code) is compiled. The parser and code generator analyze the C source code, searching for calls to the printf function. For all calls, the placeholders that were specified in the printf() format strings are collated to produce a list of the desired functionality of the final function. The doprnt.c file is then preprocessed with the those macros specified by the preliminary analysis, thus creating a custom printf() function for the project being compiled. After parsing, the p-code output derived from doprnt.c is then combined with the remainder of the C program in the final code generation step. For example, if a program contains one call to printf(), which looks like: printf(”input is: %d”); The compiler will note that only the %d placeholder is used and the doprnt.c module that is linked into the program will only contain code that handles printing of decimal integers. Consider now that the code is changed and another call to printf() is added. The new call looks like: printf(”output is %6d”); Now the compiler will detect that additional code to handle printing decimal integers to a specific width must be enabled as well. DS50002053D-page 208 2012-2013 Microchip Technology Inc. C Language Features As more features of printf() are detected, the size of the code generated for the printf() function will increase. If the format string in a call to printf() is not a string literal as above, but is rather a pointer to a string, then the compiler will not be able to reliably predict the printf() usage, and so it forces a more complete version of printf() to be generated. However, even without being able to scan printf() placeholders, the compiler can still make certain assumptions regarding the usage of the function. In particular, the compiler can look at the number and type of the additional arguments to printf() (those following the format string expression) to determine which placeholders could be valid. This enables the size and complexity of the generated printf() routine to be kept to a minimum even in this case. For example, if printf() was called as follows: printf(myFormatString, 4, 6); the compiler could determine that, for example, no floating-point placeholders are required and omit these from being included in the printf() function output. As the arguments after the format string are non-prototyped parameters, their type must match that of the placeholders. No aspect of this operation is user-controllable (other than by adjusting the calls to printf() ); however, the actual printf() code used by a program can be observed. If compiling a program using printf(), the driver will leave behind the pre-processed version of doprnt.c. This module, called doprnt.pre in your working directory, will show the C code that will actually be contained in the printf routine. As this code has been pre-processed, indentation and comments will have been stripped out as part of the normal actions taken by the C pre-processor. 2012-2013 Microchip Technology Inc. DS50002053D-page 209 MPLAB® XC8 C Compiler User’s Guide 5.12 MIXING C AND ASSEMBLY CODE Assembly language code can be mixed with C code using two different techniques: writing assembly code and placing it into a separate assembler module, or including it as in-line assembly in a C module. Note: 5.12.1 The more assembly code a project contains, the more difficult and time consuming will be its maintenance. As the project is developed, the compiler can perform different optimizations as these are based on the entire program. Assembly code can need revision if the compiler is updated due to differences in the way the updated compiler can work. These factors do not affect code written in C. If assembly must be added, it is preferable to write this as a self-contained routine in a separate assembly module, rather than in-lining it in C code. Integrating Assembly Language Modules Entire functions can be coded in assembly language as separate .as or .asm source files included into your project. They will be assembled and combined into the output image using the linker. By default, such modules are not optimized by the assembler optimizer. Optimization can be enabled by using the --OPT option, see Section 4.8.42 “--OPT: Invoke Compiler Optimizations”. The following are guidelines that must be adhered to when writing a C-callable assembly routine. • Select, or define, a suitable psect for the executable assembly code (See Section 5.15.1 “Program Sections” for an introductory guide to these.) • Select a name (label) for the routine using a leading underscore character • Ensure that the routine’s label is globally accessible from other modules • Select an appropriate C-equivalent prototype for the routine on which argument passing can be modeled • Limit arguments and return values to single byte-sized objects (Assembly routines cannot define variables that reside in the compiled stack. Use global variables for additional arguments.) • Optionally, use a signature value to enable type checking when the function is called • Use bank selection instructions and mask addresses of any variable symbols The following example goes through these steps for a mid-range device. The process is the same for other devices. A mapping is performed on the names of all C functions and non-static global variables. See Section 5.12.3 “Interaction between Assembly and C Code” for a complete description of mappings between C and assembly identifiers. An assembly routine is required which can add an 8-bit quantity passed to the routine with the contents of PORTB and return this as an 8-bit quantity. Most compiler-generated executable code is placed in psects called textn, where n is a number. (see Section 5.15.2 “Compiler-Generated Psects”). We will create our own text psect based on the psect the compiler uses. Check the assembly list file to see how the text psects normally appear for assembly generated from C code. You can see a psect, such as the following, generated by the code generator when compiling for baseline or mid-range devices. PSECT text0,local,class=CODE,delta=2 DS50002053D-page 210 2012-2013 Microchip Technology Inc. C Language Features See Section 6.4.9.3 “PSECT” for detailed information on the flags used with the PSECT assembler directive. This psect is called text0. It is flagged local, which means that it is distinct from other psects with the same name. This flag is not important in this example and can be omitted, if required. It lives in the CODE class. This flag is important as it means it will be automatically placed in the area of memory set aside for code. With this flag in place, you do not need to adjust the default linker options to have the psect correctly placed in memory. The last option, the delta value, is also very important. This indicates that the memory space in which the psect will be placed is word addressable (value of 2). The PIC10/12/16 program memory space is word addressable; the data space is byte addressable. For PIC18 devices, program memory is byte addressable, but instructions must be word-aligned, so you will see code such as the following. PSECT text0,local,class=CODE,reloc=2 In this case, the delta value is 1 (which is the default setting), but the reloc (alignment) flag is set to 2, to ensure that the section starts on a word-aligned address. We simply need to choose a different name, so we might choose the name mytext, as the psect name in which we will place out routine, so we have for our mid-range example: PSECT mytext,local,class=CODE,delta=2 Let’s assume we would like to call this routine add in the C domain. In assembly domain we must choose the name _add as this then maps to the C identifier add. If we had chosen add as the assembly routine, then it could never be called from C code. The name of the assembly routine is the label that we will place at the beginning of the assembly code. The label we would use would look like this. _add: We need to be able to call this from other modules, so make this label globally accessible, by using the GLOBAL assembler directive (Section 6.4.9.1 “GLOBAL”). GLOBAL _add By compiling a dummy C function with a similar prototype to this assembly routine, we can determine the signature value. The C-equivalent prototype to this routine would look like: unsigned char add(unsigned char); Check the assembly list file for the signature value of such a function. You will need to turn the assembler optimizer off for this step, as the optimizer removes these values from the assembly list file. Signature values are not mandatory, but allow for additional type checking to be made by the linker. We determine that the following SIGNAT directive (Section 6.4.9.21 “SIGNAT”) can be used. SIGNAT _add,4217 The W register will be used for passing in the argument. See Section 5.8.6 “Function Parameters”, for the convention used to pass parameters. 2012-2013 Microchip Technology Inc. DS50002053D-page 211 MPLAB® XC8 C Compiler User’s Guide Here is an example of the complete routine for a mid-range device which could be placed into an assembly file and added to your project. The GLOBAL and SIGNAT directives do not generator code, and hence do not need to be inside the mytext psect, although you can place them there if you prefer. The BANKSEL directive and BANKMASK macro have been used to ensure that the correct bank was selected and that all addresses are masked to the appropriate size. #include GLOBAL _add SIGNAT _add,4217 ; make _add globally accessible ; tell the linker how it should be called ; everything following will be placed into the mytext psect PSECT mytext,local,class=CODE,delta=2 ; our routine to add to ints and return the result _add: ; W is loaded by the calling function; BANKSEL (PORTB) ; select the bank of this object ADDWF BANKMASK(PORTB),w ; add parameter to port ; the result is already in the required location (W)so we can ; just return immediately RETURN To compile this, the assembly file must be preprocessed as we have used the C preprocessor #include directive. See Section 4.8.10 “-P: Preprocess Assembly Files”. To call an assembly routine from C code, a declaration for the routine must be provided. This ensures that the compiler knows how to encode the function call in terms of parameters and return values. Here is a C code snippet that declares the operation of the assembler routine, then calls the routine. // declare the assembly routine so it can be correctly called extern unsigned char add(unsigned char a); void main(void) { volatile unsigned char result; result = add(5); // call the assembly routine } 5.12.2 #asm, #endasm and asm() Assembly instructions can also be directly embedded in-line into C code using the directives #asm, #endasm or the statement asm();. The #asm and #endasm directives are used to start and end a block of assembly instructions which are to be embedded into the assembly output of the code generator. The #asm block is not syntactically part of the C program, and thus it does not obey normal C flow-of-control rules. This means that you should not use this form of in-line assembly inside or near C constructs like if(), while() , and for() statements. However, this is the easiest means of adding multiple assembly instructions. The #asm and #endasm directives should appear on lines separate from the assembly code and each other. DS50002053D-page 212 2012-2013 Microchip Technology Inc. C Language Features The asm(); statement is used to embed assembler instructions in-line with C code. This form looks and behaves like a C statement. The instructions are placed in a string inside what look like function call brackets, although no call takes place. Typically one instruction is placed in the string, but you can specify more than one assembly instruction by separating the instructions with a \n character, (e.g., asm(“MOVLW 55\nMOVWF _x”);) Code will be more readable if you one place one instruction in each statement and use multiple statements. You can use the asm() form of in-line assembly at any point in the C source code as it will correctly interact with all C flow-of-control structures. The following example shows both methods used: unsigned int var; void main(void) { var = 1; #asm // like this... BCF 0,3 BANKSEL(_var) RLF (_var)&07fh RLF (_var+1)&07fh #endasm // do it again the other way... asm(“BCF 0,3”); asm(“BANKSEL _var”); asm(“RLF (_var)&07fh”); asm(“RLF (_var+1)&07fh”); } In-line assembly code is never optimized by the assembler optimizer. When using in-line assembler code, great care must be taken to avoid interacting with compiler-generated code. The code generator cannot scan the assembler code for register usage and so will remain unaware if registers are clobbered or used by the assembly code. The registers used by the compiler are explained in Section 5.7 “Register Usage”. If you are in doubt as to which registers are being used in surrounding code, compile your program with the --ASMLIST option (see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”) and examine the assembler code generated by the compiler. Remember that as the rest of the program changes, the registers and code strategy used by the compiler will change as well. If a C function is called from main-line and interrupt code, it can be duplicated, see Section 5.9.6 “Function Duplication”. Although a special prefix is used to ensure that labels generated by the compiler are not duplicated, this does not apply to labels defined in hand-written, in-line assembly code in C functions. Thus, you should not define assembly labels in in-line assembly if the containing function might be duplicated. 2012-2013 Microchip Technology Inc. DS50002053D-page 213 MPLAB® XC8 C Compiler User’s Guide 5.12.3 Interaction between Assembly and C Code MPLAB XC8 C Compiler incorporates several features designed to allow C code to obey requirements of user-defined assembly code. There are also precautions that must be followed to ensure that assembly code does not interfere with the assembly generated from C code. The command-line driver ensures that all user-defined assembly files have been processed first, before compilation of C source files begin. The driver is able to read and analyze certain information in the relocatable object files and pass this information to the code generator. This information is used to ensure the code generator takes into account requirement of the assembly code. See Section 4.3.4 “Compilation of Assembly Source” for further information on the compile sequence. 5.12.3.1 EQUIVALENT ASSEMBLY SYMBOLS Most C symbols map to an corresponding assembly equivalent. This mapping is such that an “ordinary” symbol defined in the assembly domain cannot interfere with an “ordinary” symbol in the C domain. So for example, if the symbol main is defined in the assembly domain, it is quite distinct to the main symbol used in C code and they refer to different locations. The name of a C function maps to an assembly label that will have the same name, but with an underscore prepended. So the function main() will define an assembly label _main. Baseline PIC devices can use alternate assembly domain symbols for functions. The destinations of call instructions on these devices are limited to the first half of a program memory page. The compiler, thus, encodes functions in two parts, as illustrated in the following example of a C function, add(), compiled for a baseline device. entry__add: LJMP _add The label entry__add is the function’s entry point and will always be located in a special psect linked in the first half of a program memory page. The code associated with this label is simply a long jump (see Section 6.4.1.7 “Long Jumps and Calls”) to the actual function body located elsewhere and identified by the label _add. If you plan to call routines from assembly code, you must be aware of this limitation in the device and the way the compiler works around it for C functions. Hand-written assembly code should always call the entry__funcName label rather than the usual assembly-equivalent function label. If a C function is qualified static, and there is more than one static function in the program with exactly the same name, the name of the first function will map to the usual assembly symbol and the subsequent functions will map to a special symbol with the form: fileName@functionName, where fileName is the name of the file that contains the function, and functionName is the name of the function. For example, a program contains the definition for two static functions, both called add. One lives in the file main.c and the other in lcd.c. The first function will generate an assembly label _add. The second will generate the label lcd@add. The name of a non-auto C variable also maps to an assembler label that will have the same name, but with an underscore prepended. So the variable result will define an assembly label: _result. DS50002053D-page 214 2012-2013 Microchip Technology Inc. C Language Features If the C variable is qualified static, there, again, is a chance that there could be more than one variable in the program with exactly the same C name. The same rules apply to non-local static variables as to static functions. The name of the first variable will map to a symbol prepended with an underscore; the subsequent symbols will have the form: fileName@variableName, where fileName is the name of the file that contains the variable, and variableName is the name of the variable. For example a program contains the definition for two static variables, both called result. One lives in the file main.c and the other in lcd.c. The first function will generate an assembly label _result. The second will generate the label lcd@result. If there is more than one local static variable (i.e., it is defined inside a function definition) then all the variables will have an assembly name of the form: functionName@variableName. So, if there is a static variable called output in the function read, and another static variable with the same name defined in the function update, then in assembly the symbols can be accessed using the symbols read@output and update@output, respectively. If there is more than one static function with the same name, and they contain definitions for static variables of the same name, then the assembly symbol used for these variables will be of the form: fileName@functionName@variableName. Having two static variables or functions with the same name is legal, but not recommended as is easy to write code that accesses the wrong variable or calls the wrong function. Functions that use the reentrant model do not define any symbols that allow you to access auto and parameter variables. You should not attempt to access these in assembly code. Special symbols for auto and parameter variables are defined, however, by functions that use the non-reentrant model. These symbols are described in the following paragraphs. To allow easy access to parameter and auto variables on the compiled stack, special equates are defined which map a unique symbol to each variable. The symbol has the form: functionName@variableName. Thus, if the function main defines an auto variable called foobar, the symbol main@foobar can be used in assembly code to access this C variable. Function parameters use the same symbol mapping as auto variables. If a function called read has a parameter called channel, then the assembly symbol for that parameter is read@channel. Function return values have no C identifier associated with them. The return value for a function shares the same memory as that function’s parameter variables, if they are present. The assembly symbol used for return values has the form ?_funcName, where funcName is the name of the function returning the value. Thus, if a function, getPort returns a value, it will be located the address held by the assembly symbol ?_getPort. If this return value is more than one byte in size, then an offset is added to the symbol to access each byte, e.g., ?_getPort+1. If the compiler creates temporary variables to hold intermediate results, these will behave like auto variables. As there is no corresponding C variable, the assembly symbol is based on the symbol that represents the auto block for the function plus an offset. That symbol is ??_funcName, where funcName is the function in which the symbol is being used. So for example, if the function main uses temporary variables, they will be accessed as an offset from the symbol ??_main. 2012-2013 Microchip Technology Inc. DS50002053D-page 215 MPLAB® XC8 C Compiler User’s Guide 5.12.3.2 ACCESSING REGISTERS FROM ASSEMBLY CODE If writing separate assembly modules, SFR definitions will not automatically be accessible. The assembly header file can be used to gain access to these register definitions. Do not use this file for assembly in-line with C code as it will clash with definitions in . Include the file using the assembler’s INCLUDE directive, (see Section 6.4.10.4 “INCLUDE”) or use the C preprocessor’s #include directive. If you are using the latter method, make sure you compile with the -P driver option to preprocess assembly files, see Section 4.8.10 “-P: Preprocess Assembly Files”. The symbols for registers in this header file look similar to the identifiers used in the C domain when including , e.g., PORTA, EECON1, etc. They are different symbols in different domains, but will map to the same memory location. Bits within registers are defined as the registerName,bitNumber. So for example, RA0 is defined as PORTA,0. Here is an example of a mid-range assembly module that uses SFRs. #include GLOBAL _setports PSECT text,class=CODE,local,delta=2 _setports: MOVLW 0xAA BANKSEL (PORTA) MOVWF BANKMASK(PORTA) BANKSEL (PORTB) BSF RB1 If you wish to access register definitions from assembly that is in-line with C code, ensure that the header is included into the C module. Information included by this header will define in-line assembly symbols as well as the usual symbols accessible from C code. The symbols used for register names will be the same as those defined by . So for example, the example given previously could be rewritten as in-line assembly as follows. #asm MOVLW BANKSEL MOVWF BANKSEL BSF #endasm 0xAA (PORTA) BANKMASK(PORTA) (PORTB) RB1 Care must be taken to ensure that you do not destroy the contents of registers that are holding intermediate values of calculations. Some registers are used by the compiler and writing to these registers directly can result in code failure. The code generator does not detect when SFRs have changed as a result of assembly code that writes to them. The list of registers used by the compiler and further information can be found in Section 5.7 “Register Usage”. DS50002053D-page 216 2012-2013 Microchip Technology Inc. C Language Features 5.12.3.3 ABSOLUTE PSECTS Some of the information that is extracted from the initial compilation of assembly code, see Section 4.3.4 “Compilation of Assembly Source”, relates to absolute psects, specifically psects defined using the abs and ovrld, PSECT flags, see Section 6.4.9.3 “PSECT” for information on this directive. MPLAB XC8 is able to determine the address bounds of absolute psects and uses this information to ensure that the code produced from C source by the code generator does not use memory required by the assembly code. The code generator will reserve any memory used by the assembly code prior to compiling C source. Here is an example of how this works. An assembly code files defines a table that must be located at address 0x110 in the data space. The assembly file contains: PSECT lkuptbl,class=RAM,space=1,abs,ovrld ORG 110h lookup: DS 20h An absolute psect always starts at address 0. For such psects, you can specify a non-zero starting address by using the ORG directive. See Section 6.4.9.4 “ORG” for important information on this directive. When the project is compiled, this file is assembled and the resulting relocatable object file scanned for absolute psects. As this psect is flagged as being abs and ovrld, the bounds and space of the psect will be noted — in this case, a memory range from address 0x110 to 0x12F in memory space 1 is noted as being used. This information is passed to the code generator to ensure that this address range is not used by the assembly generated from the C code. The linker handles all of the allocation into program memory, and so for hand-written assembly, only the psects located in data memory need be defined in this way. 2012-2013 Microchip Technology Inc. DS50002053D-page 217 MPLAB® XC8 C Compiler User’s Guide 5.12.3.4 UNDEFINED SYMBOLS If a variable needs to be accessible from both assembly and C source code, it can be defined in assembly code, if required, but it is easier to do so in C source code. A problem could occur if there is a variable defined in C source, but is only ever referenced in the assembly code. In this case, the code generator would remove the variable believing it is unused. The linker would be unable to resolve the symbol referenced by the assembly code and an error will result. To work around this issue, MPLAB XC8 also searches assembly-derived object files for symbols which are undefined. see Section 4.3.4 “Compilation of Assembly Source”. These will be symbols that are used, but not defined, in assembly code. The code generator is informed of these symbols, and if they are encountered in the C code, the variable is automatically marked as being volatile. This action has the same effect as qualifying the variable volatile in the source code, see Section 5.4.7.2 “Volatile Type Qualifier”. Variables qualified as volatile will never be removed by the code generator, even if they appear to be unused throughout the program. For example, if a C program defines a global variable as follows: int input; but this variable is only ever used in assembly code. The assembly module(s) can simply declare this symbol using the GLOBAL assembler directive, and then use it. The following PIC18 example illustrates the assembly code accessing this variable. GLOBAL _input, _raster PSECT text,local,class=CODE,reloc=2 _raster: MOVF _input,w The compiler knows of the mapping between the C symbol input, and the corresponding assembly symbol _input (see Section 5.12.3 “Interaction between Assembly and C Code”). In this instance the C variable input will not be removed and be treated as if it was qualified volatile. DS50002053D-page 218 2012-2013 Microchip Technology Inc. C Language Features 5.13 OPTIMIZATIONS The optimizations in the HI-TECH C compiler can be broadly grouped into C-level optimizations performed on the source code before conversion into assembly, and assembly-level optimizations performed on the assembly code generated by the compiler. The C-level optimizations are performed early during the code generation phase and so have flow-on benefits: performing one optimizations can mean that another can then be applied. As these optimizations are applied before the debug information has been produced, there is typically little impact on source-level debugging of programs. Some of these optimizations are integral to the code generation process and so cannot be disabled via an option. Suggestions as to how specific optimizations can be defeated are given in the sections below. In Standard mode, and particularly Free mode, some of these optimizations are disabled. (Hence if you want to disable as many optimizations as possible, run the compiler in the Free operating mode.) Even if they are enabled, optimizations can only be applied if very specific conditions are met. As a result, you might find that some lines of code are optimized, but others are not. The compiler operating mode determines the available optimizations, which are listed in Table 5-13. TABLE 5-13: Mode OPERATING MODE OPTIMIZATION SETS Optimization sets available Free • Basic code generator optimizations STD • Basic code generator optimizations • Whole program assembly optimizations PRO • • • • Basic code generator optimizations Whole program assembly optimizations Procedural abstraction (assembly optimization) OCG C-level optimizations Assembly-level optimizations are described in Section 6.5 “Assembly-Level Optimizations”. 2012-2013 Microchip Technology Inc. DS50002053D-page 219 MPLAB® XC8 C Compiler User’s Guide The basic code generator optimizations consist of the following. Whole-program analysis for object allocation into data banks without having to use non-standard keywords or compiler directives. Simplification and folding of constant expressions to simplify expressions. Expression tree optimizations to ensure efficient assembly generation. The main OCG C-level optimizations, which simplify C expressions or code produced from C expressions, are listed below. These are applied across the entire program, not just on a module-by-module basis. Tracking of the current data bank is performed by the compiler as it generated assembly code. This allows the compiler to reduce the number of bank-selection instructions generated. Strength reductions and expression transformations are applied to all expression trees before code is generated. This involves replacing expressions with equivalent but less costly operations. Unused variables in a program are removed. This applies to local as well as global variables. Variables removed will not have memory reserved for them, will not appear in any list or map file, and will not be present in debug information, and so will not be observable in the debugger. A warning is produced if an unused variable is encountered. Objects qualified volatile will never be removed, see Section 5.4.7.2 “Volatile Type Qualifier”. Taking the address of a variable or referencing its assembly-domain symbol in hand-written assembly code also constitutes use of the variable. Redundant assignments to variables not subsequently used are removed, unless the variable is volatile. The assignment statement is completely removed, as if it was never present in the original source code. No code will be produced for it, and you will not be able to set a breakpoint on that line in the debugger. Unused functions in a program are removed. A function is considered unused if it is not called, directly or indirectly, nor has had its address taken. The entire function is removed, as if it was never present in the original source code. No code will be produced for it and you will not be able to set a breakpoint on any line in the function in the debugger. Referencing a function’s assembly-domain symbol in a separate hand-written assembly module will prevent it being removed. The assembly code need only use the symbol in the GLOBAL directive. Unused return expressions in a function are removed. The return value is considered unused if the result of all calls to that function discard the return value. The code associated with calculation of the return value will be removed, and the function will be encoded as if its return type was void. Propagation of constants is performed where the numerical contents of a variable can be determined. Variables which are not volatile and whose value can be exactly determined are replaced with the numerical value. Uninitialized global variables are assumed to contain zero prior to any assignment to them. DS50002053D-page 220 2012-2013 Microchip Technology Inc. C Language Features Variables assigned a value before being read are not cleared or initialized by the runtime startup code. Only non-auto variables are considered and if they are assigned a value before other code can read their value, they are treated as being persistent, see Section 5.4.8.1 “Persistent Type Qualifier”. All persistent objects are not cleared by the runtime startup code, so this optimization will speed execution of the program startup. Pointer sizes are optimized to suit the target objects they can access. The compiler tracks all assignments to pointer variables and keeps a list of targets each pointer can access. As the memory space of each target is known, the size and dereference method used can be customized for each pointer. Dereferencing pointers with only target can be replaced with direct access of the target object. This applies to data and function pointers. Unreachable code is removed. C Statements that cannot be reached are removed before they generate assembly code. This allows subsequent optimizations to be applied at the C level. As C-level optimizations are performed before debug information is produced, they tend to have less impact on debugging information. However, if a variable is located in a register, IDEs, such as MPLAB IDE, can indicate incorrect values in their Watch view. Try to use the ELF/DWARF debug file format to minimize such occurrences. Check the assembly list file to see if registers are used in the routine being debugged. The assembly-level optimizations are described in Section 6.5 “Assembly-Level Optimizations”. 2012-2013 Microchip Technology Inc. DS50002053D-page 221 MPLAB® XC8 C Compiler User’s Guide 5.14 PREPROCESSING All C source files are preprocessed before compilation. The preprocessed file is not deleted after compilation. It will have a .pre extension and the same base name as the source file from which it is derived. The --PRE option can be used to preprocess and then stop the compilation. See Section 4.8.47 “--PRE: Produce Preprocessed Source Code”. Assembler files can also be preprocessed if the -P driver option is issued. See Section 4.8.10 “-P: Preprocess Assembly Files”. 5.14.1 C Language Comments The MPLAB XC8 C compiler supports standard C comments, as well as C++ style comments. Both types are illustrated in the following table. Comment Syntax /* */ // 5.14.2 Description Example Standard C code comment. Used for one or more lines. /* This is line 1 This is line 2 */ C++ code comment. Used for one line only. // This is line 1 // This is line 2 Preprocessor Directives HI-TECH C accepts several specialized preprocessor directives, in addition to the standard directives. All of these are listed in Table 5-14 on the next page. Macro expansion using arguments can use the # character to convert an argument to a string, and the ## sequence to concatenate arguments. If two expressions are being concatenated, consider using two macros in case either expression requires substitution itself; so, for example #define __paste1(a,b) #define __paste(a,b) a##b __paste1(a,b) lets you use the paste macro to concatenate two expressions that themselves can require further expansion. Remember also that once a macro identifier has been expanded, it will not be expanded again if it appears after concatenation. DS50002053D-page 222 2012-2013 Microchip Technology Inc. C Language Features TABLE 5-14: PREPROCESSOR DIRECTIVES Directive Meaning Example # preprocessor null directive, do nothing # #advisory generate an advisory message #advisory TODO: I need to finish this #assert generate error if condition false #assert SIZE > 10 #asm signifies the beginning of in-line assembly #asm MOVLW FFh #endasm #define define preprocessor macro #define SIZE 5 #define FLAG #define add(a,b) ((a)+(b)) #elif short for #else #if see #ifdef #else conditionally include source lines see #if #endasm terminate in-line assembly see #asm #endif terminate conditional source inclusion see #if #error generate an error message #error Size too big #if include source lines if constant expression true #if SIZE < 10 c = process(10) #else skip(); #endif #ifdef include source lines if preprocessor symbol defined #ifdef FLAG do_loop(); #elif SIZE == 5 skip_loop(); #endif #ifndef include source lines if preprocessor symbol not defined #ifndef FLAG jump(); #endif #include include text file into source #include #include “project.h” #info alias for #advisory (see above) #info I wrote this bit #line specify line number and filename for listing #line 3 final #nn (where nn is a number) short for #line nn #20 #pragma compiler specific options Refer to Section “Each symbol, if defined, is equated to 1 (unless otherwise stated).” #undef undefines preprocessor symbol #undef FLAG #warning generate a warning message #warning Length not set The type and conversion of numeric values in the preprocessor domain is the same as in the C domain. Preprocessor values do not have a type, but acquire one as soon as they are converted by the preprocessor. Expressions can overflow their allocated type in the same way that C expressions can overflow. Overflow can be avoided by using a constant suffix. For example, an L after the number indicates it should be interpreted as a long once converted. So, for example #define MAX 1000*1000 and #define MAX 1000*1000L 2012-2013 Microchip Technology Inc. DS50002053D-page 223 MPLAB® XC8 C Compiler User’s Guide will define the values 0x4240 and 0xF4240, respectively. 5.14.3 Predefined Macros The compiler drivers define certain symbols to the preprocessor, allowing conditional compilation based on chip type, etc. The symbols listed in Table 5-15 show the more common symbols defined by the drivers. Each symbol, if defined, is equated to 1 (unless otherwise stated). TABLE 5-15: PREDEFINED MACROS Symbol When set Usage __CHIPNAME and __CHIPNAME__ when chip selected to indicate the specific chip type selected, e.g., __16F877 __DATABANK if eeprom or flash memory implemented identifies which bank the EEDATA/PMDATA register is found __DATE__ always to indicate the current date, e.g., can 21 2004 __FILE__ always to indicate this source file being preprocessed. __FLASHTYPE if flash memory is implemented to indicate the type of flash memory employed by the target device, see _PROGMEM below. __LINE__ always to indicate this source line number. __J_PART if PIC® 18 J device indicates device a ‘J’ series part __MPLAB_ICDX__ if compiling for MPLAB® ICD or (where X is empty, 2 or 3 MPLAB ICD 2/3 debugger Assigned 1 to indicate that the code is generated for use with the Microchip MPLAB ICD,ICD 2 or ICD 3. __MPLAB_PICKITX__ if compiling for MPLAB PICkit™ 2/3 Assigned 1 to indicate that the code is generated for use with the Microchip MPLAB PICKIT 2 or PICKIT 3. __MPLAB_REALICE__ if compiling for MPLAB REAL ICE™ Assigned 1 to indicate that the code is generated for use with the Microchip MPLAB REAL ICE. __PICCPRO__ and __PICC__ if any non-PIC18 device to indicate the target device is any PIC10/12/14/16 __PICC18__ if not in C18 compatibility mode to indicate non-C18 compatibility mode operation __RESETBITS_ADDR if --RUNTIME option request STATUS register save indicates the address at which the STATUS register will be saved __TIME__ always to indicate the current time, e.g., 08:06:31. __XC always indicates MPLAB XC compiler for Microchip in use __XC8 always indicates MPLAB XC compiler for Microchip 8-bit devices in use __XC8_VERSION always to indicate the compiler’s version number multiplied by 1000, e.g., v1.00 will be represented by 1000 _CHIPNAME when chip selected to indicate the specific chip type selected, e.g., _16F877 _BANKBITS_ always assigned 0, 1 or 2 to indicate 1, 2 or 4 available banks or RAM _BANKCOUNT_ always to indicate the number of banks of data memory implemented _COMMON_ if common RAM present to indicate whether device has common RAM area _EEPROMSIZE always to indicate how many bytes of EEPROM are available _ERRATA_TYPES always indicates the errata workarounds being applied, see --ERRATA option Section 4.8.26 _FAMILY_FAMILY_ (1) if PIC18 device indicates PIC18 family DS50002053D-page 224 2012-2013 Microchip Technology Inc. C Language Features TABLE 5-15: PREDEFINED MACROS (CONTINUED) Symbol When set Usage _FLASH_ERASE_SIZE always size of flash erase block _FLASH_WRITE_SIZE Always size of flash write block _GPRBITS_ always assigned 0, 1 or 2 to indicate 1, 2 or 4 available banks or general purpose RAM. _HAS_OSCVAL_ if the target device has an oscil- to indicate target device can require oscillator lator calibration register calibration _HTC_EDITION_ always indicates which of PRO, Standard or Free compiler is in use Values of 2, 1 or 0 are assigned, respectively. _HTC_VER_MAJOR_ _HTC_VER_MINOR_ always to indicate the whole or decimal component, respectively, of the compiler’s version number _HTC_VER_PATCH_ _HTC_VER_PLVL_ always to indicate the patch level of the compiler’s version number _MPC_ always indicates compiling for Microchip PIC MCU family _OMNI_CODE_ always indicates compiling using an OCG compiler _PIC12 if baseline (12-bit instruction) device to indicate selected device is a baseline PIC devices _PIC12E if enhanced baseline (12-bit instruction) device to indicate selected device is an enhanced baseline PIC devices _PIC14 if mid-range (14-bit instruction) device to indicate selected device is a mid-range PIC devices _PIC14E if Enhanced mid-range (14-bit instruction) device to indicate selected device is an Enhanced mid-range PIC devices _PIC18 if PIC18 (16-bit instruction) device to indicate selected device is an PIC18 devices _PROGMEM_ if compiling for mid-range device with flash memory to indicate the type of flash memory employed by the target device values 0xFF (unknown), 0xF0 (none), 0 (read-only), 1 (word write with auto erase), 2 (block write with auto erase), 3 (block write with manual erase) _RAMSIZE if PIC18 device to indicate how many bytes of data memory are available _ROMSIZE always to indicate how much program memory is available (byte units for PIC18 devices; words for other devices) EEPROMSIZE always ERRATA_4000_BOUNDARY if the ERRATA_4000 applies to indicate how many bytes of EEPROM are available to indicate that the 4000 word boundary errata is applied HI_TECH_C always to indicate that the C language variety is HI-TECH C compatible __TRADITIONAL18__ if PIC18 device to indicate the non-extended instruction set is selected __STRICT if the --STRICT option is enabled to indicate that strict ANSI compliance of keywords is in force MPLAB_ICD if compiling for MPLAB ICD or MPLAB ICD 2/3 debugger Assigned 1 to indicate that the code is generated for use with the Microchip MPLAB ICD 1. Assigned 2 for MPLAB ICD 2; 3 for MPLAB ICD 3. Note 1: To determine the family macro relevant to your device, look for the FAMILY field in the picc-18.ini file in the compiler's DAT directory. 2012-2013 Microchip Technology Inc. DS50002053D-page 225 MPLAB® XC8 C Compiler User’s Guide 5.14.4 Pragma Directives There are certain compile-time directives that can be used to modify the behavior of the compiler. These are implemented through the use of the ANSI standard #pragma facility. The format of a pragma is: #pragma keyword options where keyword is one of a set of keywords, some of which are followed by certain options. A list of the keywords is given in Table 5-16. Those keywords not discussed elsewhere are detailed below. TABLE 5-16: PRAGMA DIRECTIVES Directive Meaning Example addrqual specify action of qualifiers #pragma addrqual=require config specify configuration bits #pragma config WDT=ON inline inline function if possible #pragma inline(getPort) intrinsic specify function is inline #pragma intrinsic(_delay) interrupt_level allow call from interrupt and #pragma interrupt_level 1 main-line code pack specify structure packing #pragma pack 1 printf_check enable printf-style format string checking #pragma printf_check(printf) const psect rename compiler-generated psect #pragma psect nvBANK0=my_nvram regsused specify registers used by function #pragma regsused myFunc wreg,fsr switch specify code generation for #pragma switch direct switch statements warning control messaging parameters 5.14.4.1 #pragma warning disable 299,407 THE #PRAGMA ADDRQUAL DIRECTIVE This directive allows you to control the compiler’s response to non-standard memory qualifiers. This pragma is an in-code equivalent to the --ADDRQUAL option and both use the same arguments, see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”. The pragma has effect over the entire C program and should be issued once, if required. If the pragma is issued more than once, the last pragma determines the compiler’s response. For example: #pragma addrqual=require bank2 int foobar; 5.14.4.2 THE #PRAGMA CONFIG DIRECTIVE This directive allows the device Configuration bits to be specified for PIC18 target devices. See Section 5.3.5 “Configuration Bit Access” for full details. DS50002053D-page 226 2012-2013 Microchip Technology Inc. C Language Features 5.14.4.3 THE #PRAGMA INLINE DIRECTIVE The #pragma inline directive indicates to the compiler that calls to the specified function should be as fast as possible. This pragma has the same effect as using the inline function specifier. 5.14.4.4 THE #PRAGMA INTRINSIC DIRECTIVE The #pragma intrinsic directive is used to indicate to the compiler that a function will be inlined intrinsically by the compiler. The directive is only usable with special functions that the code generator will expand internally, e.g the _delay function. Such functions do not have corresponding source code and are handled specially by the compiler. Note: 5.14.4.5 Use of this pragma with a user-defined function does not mean that function will be inlined and an error will result. See the inline function specifier for that operation, in Section 5.8.1.2 “Inline Specifier”. THE #PRAGMA INTERRUPT_LEVEL DIRECTIVE The #pragma interrupt_level directive can be used to prevent function duplication of functions called from main-line and interrupt code. See Section 5.9.6.1 “Disabling Duplication” for more information. 5.14.4.6 THE #PRAGMA PACK DIRECTIVE All 8-bit PIC devices can only perform byte accesses to memory and so do not require any alignment of memory objects within structures. This pragma will have no effect when used. 5.14.4.7 THE #PRAGMA PRINTF_CHECK DIRECTIVE Certain library functions accept a format string followed by a variable number of arguments in the manner of printf(). Although the format string is interpreted at runtime, it can be compile-time checked for consistency with the remaining arguments. This directive enables this checking for the named function, for example the system header file includes the directive: #pragma printf_check(printf) const to enable this checking for printf(). You can also use this for any user-defined function that accepts printf -style format strings. The qualifier following the function name is to allow automatic conversion of pointers in variable argument lists. The above example would cast any pointers to strings in RAM to be pointers of the type (const char *) Note that the warning level must be set to -1 or below for this option to have any visible effect. See Section 4.8.61 “--WARN: Set Warning Level”. 2012-2013 Microchip Technology Inc. DS50002053D-page 227 MPLAB® XC8 C Compiler User’s Guide 5.14.4.8 THE #PRAGMA PSECT DIRECTIVE The #pragma psect can be used to redirect objects and functions to a new psect. It is recommended you consider instead the __section() specifier (see Section 5.15.4 “Changing and Linking the Allocated Section”), which cannot only perform the same task, but which allows a greater flexibility in where the new sections can be linked. The main difference between the specifier and the pragma is that the pragma preserves flags associated with the standard psect, whereas the specifier does not. Normally the object code generated by the compiler is broken into the standard psects. This is described in 5.15.2 “Compiler-Generated Psects”. For an introductory guide to psects, see Section 5.15.1 “Program Sections”. This is fine for most applications, but sometimes it is necessary to redirect variables or code into different psects when a special memory configuration is desired. Some compiler-generated psects that contain code and data can be redirected using a #pragma psect directive. The general form of this pragma looks like: #pragma psect standardPsect=newPsect and instructs the code generator that anything that would normally appear in the standard psect standardPsect, will now appear in a new psect called newPsect. This psect will be identical to standardPsect in terms of its flags and attributes; however, it will have a unique name. Thus, you can explicitly position this new psect without affecting the placement of anything in the original psect. If the name of the standard psect that is being redirected contains a counter (e.g., text0, text1, text2, etc.), the placeholder %%u should be used in the name of the psect at the position of the counter, e.g., text%%u. This will match any psect, regardless of the counter value. For example, to remap a C function, you could use: #pragma psect text%%u=lookupfunc int lookup(char ind) { ... Standard psects that make reference to a bank number are not using a counter and do not need the placeholder to match. For example, to redirect an uninitialized variable from bank 1 memory, use: #pragma psect bssBANK1=sharedObj int foobar; This pragma should not be used for any of the data psects (data or idata) that hold initialized variables. These psects must be assembled in a particular order and the use of this pragma to redirect some of their content will destroy this order. Use of this pragma with RAM-based psects that are intended to be linked into a particular RAM bank is acceptable, but the new psect must be linked into the same bank. Linking the new psect to a different bank can lead to code failure. This pragma affects the entire module in which it is located, regardless of the position of the pragma in the file. Any given psect should only be redirected once in a particular module. That is, you cannot redirect the standard psect for some of the module, then swap back to using the standard psect for the remainder of the source code. The pragma should typically be placed at the top of the source file. It is recommended that the code or variables to be separated be placed in a source file all to themselves so they are easily distinguished. DS50002053D-page 228 2012-2013 Microchip Technology Inc. C Language Features To determine the psect in which the function or object is normally located, define the function or object in the usual way and without this pragma. Now check the assembly list file (see 6.5 “Assembly-Level Optimizations”) to determine the psect in which the function or object is normally positioned. Note that the location of objects and functions can vary with the target device and compiler options selected. Check either the assembly list file or the map file with the pragma in place to ensure that the mapping has worked as expected and that the function or variable has been linked at the address specified. Consider this crude program that consists of several functions. #include void init(void) { } int getInput(void) { return PORTA; } int masher(int val) { return val; } void main(void) { int input; init(); while(1) { input = getInput(); masher(input); } } 2012-2013 Microchip Technology Inc. DS50002053D-page 229 MPLAB® XC8 C Compiler User’s Guide The programmer decides that the function masher() is to be linked at a particular address and they do not want to make the function absolute. The programmer first compiles they code and checked either the assembly list file or map file to find the psect in which masher is normally found. For example, the map file shows the symbol associated with the function and the psect in which it is located. Symbol Table _getInput _init _main _masher ... text13 text12 maintext text14 07E5 07E4 07ED 07E8 So it is allocated to a textn psect. The source code for the masher() function is removed from this source file and placed in is own file. The psect pragma is then applied so that this new file contains the following. #pragma psect text%%u=myMasherPsect int masher(int val) { return val; } After recompiling, the function is now located in its own psect. The updated map file now shows the following. Note that its address has not changed. Symbol Table _getInput _init _main _masher ... text13 text12 maintext myMasherPsect 07E5 07E4 07ED 07E8 The programmer now uses the driver option -L-pmyMasherPsect=1000h to locate this psect at the desired location. After recompiling, the map file now shows the following. Note that the addresses of other functions have not been affected by the pragma in this example. Symbol Table _getInput _init _main _masher ... text13 text12 maintext myMasherPsect 07E5 07E4 07ED 1000 Variables can also be placed at specific positions by making them absolute, see Section 5.5.4 “Absolute Variables”. The same is also true for functions. See 5.8.4 “Changing the Default Function Allocation”. The decision whether functions or variables should be positioned using absolutes or via the psect pragma should be based on the location requirements. Using absolute functions and variables is the easiest method, but only allows placement at an address which must be known prior to compilation. The psect pragma is more complex, but offers all the flexibility of the linker to position the new psect into memory. For example, you can specify that functions or variables reside at a fixed address, or that they be placed after other psects, or that the psect be placed anywhere in a compiler-defined or user-defined range of address. See Chapter 7. “Linker” for the features and options available when linking. See also 4.8.6 “-L-: Adjust Linker Options Directly” for information on controlling the linker from the driver or in MPLAB IDE. DS50002053D-page 230 2012-2013 Microchip Technology Inc. C Language Features 5.14.4.9 THE #PRAGMA REGSUSED DIRECTIVE The #pragma regsused directive allows the programmer to indicate register usage for functions that will not be “seen” by the code generator, for example if they were written in assembly code. It has no effect when used with functions defined in C code, but in these cases the register usage of these functions can be accurately determined by the compiler and the pragma is not required. The compiler will determine only those registers and objects which need to be saved for an interrupt function defined and use of this pragma allows the code generator to also determine register usage for routines written in assembly code. The general form of the pragma is: #pragma regsused routineName registerList where routineName is the C equivalent name of the function or routine whose register usage is being defined, and registerList is a space-separated list of registers names, as shown in Table 5-10. Those registers not listed are assumed to be unused by the function or routine. The code generator can use any of these registers to hold values across a function call. Hence, if the routine does in fact use these registers, unreliable program execution can eventuate. The register names are not case sensitive and a warning will be produced if the register name is not recognized. A blank list indicates that the specified function or routine uses no registers. If this pragma is not used, the compiler will assume that the external function uses all registers. For example, a routine called _search is written in PIC18 assembly code. In the C source, we can write: extern void search(void); #pragma regsused search wreg status fsr0 to indicate that this routine used the W register, STATUS and FSR0. Here, FSR0 expands to both FSR0L and FSR0H. These could be listed individually, if required. 5.14.4.10 THE #PRAGMA SWITCH DIRECTIVE Normally, the compiler chooses how switch statements will be encoded to produce the smallest possible code size. The #pragma switch directive can be used to force the compiler to use a different coding strategy. The general form of the switch pragma is: #pragma switch switchType where switch_type is one of the available switch types (the only switch type currently implemented for PIC18 devices is space) listed in Table 5-17. TABLE 5-17: SWITCH TYPES Switch Type Description speed use the fastest switch method space use the smallest code size method time use a fixed delay switch method auto use smallest code size method (default) direct (deprecated) use a fixed delay switch method simple (deprecated) sequential xor method 2012-2013 Microchip Technology Inc. DS50002053D-page 231 MPLAB® XC8 C Compiler User’s Guide Specifying the time option to the #pragma switch directive forces the compiler to generate a table look-up style switch method. The time taken to execute each case is the same, so this is useful where timing is an issue, e.g., state machines. This pragma affects all subsequent code. The auto option can be used to revert to the default behavior. There is information printed in the assembly list file for each switch statement showing the chosen strategy, see Section 6.6.4 “Switch Statement Information”. 5.14.4.11 THE #PRAGMA WARNING DIRECTIVE This pragma allows control over some of the compiler’s messages, such as warnings and errors. For full information on the massaging system employed by the compiler, see Section 4.6 “Compiler Messages”. 5.14.4.11.1 The Warning Disable Pragma Some warning messages can be disabled by using the warning disable pragma. This pragma will only affect warnings that are produced by the parser or the code generator; i.e., errors directly associated with C code. The position of the pragma is only significant for the parser; i.e., a parser warning number can be disabled for one section of the code to target specific instances of the warning. Specific instances of a warning produced by the code generator cannot be individually controlled and the pragma will remain in force during compilation of the entire module. The state of those warnings which have been disabled can preserved and recalled using the warning push and warning pop pragmas. Pushes and pops can be nested to allow a large degree of control over the message behavior. The following example shows the warning associated with assigning the address of a const object to a pointer to non-const objects. Such code normally produces warning number 359. int readp(int * ip) { return *ip; } const int i = 'd'; void main(void) { unsigned char c; #pragma warning disable 359 readp(&i); #pragma warning enable 359 } This same affect would be observed using the following code. #pragma warning push #pragma warning disable 359 readp(&i); #pragma warning pop Here the state of the messaging system is saved by the warning push pragma. Warning 359 is disabled, then after the source code which triggers the warning, the state of the messaging system is retrieved by using the warning pop pragma. DS50002053D-page 232 2012-2013 Microchip Technology Inc. C Language Features 5.14.4.11.2 The Warning Error/Warning Pragma It is also possible to change the type of some messages. This is only possible by the use of the warning pragma and only affects messages generated by the parser or code generator. The position of the pragma is only significant for the parser; i.e., a parser message number can have its type changed for one section of the code to target specific instances of the message. Specific instances of a message produced by the code generator cannot be individually controlled and the pragma will remain in force during compilation of the entire module. The following example shows the warning produced in the previous example being converted to an error for the instance in the function main(). void main(void) { unsigned char c; #pragma warning error 359 readp(&i); } Compilation of this code would result in an error, not the usual warning. The error will force compilation to cease after the current module has concluded, or immediately if the maximum error count has been reached. 2012-2013 Microchip Technology Inc. DS50002053D-page 233 MPLAB® XC8 C Compiler User’s Guide 5.15 LINKING PROGRAMS The compiler will automatically invoke the linker unless the compiler has been requested to stop after producing an intermediate file. The linker will run with options that are obtained from the command-line driver. These options specify the memory of the device and how the psects should be placed in the memory. No linker scripts are used. The linker options passed to the linker can be adjusted by the user, but this is only required in special circumstances. See Section 4.8.6 “-L-: Adjust Linker Options Directly” for more information.) The linker creates a map file which details the memory assigned to psects and some objects within the code. The map file is the best place to look for memory information. See Section 7.4 “Map Files” for a detailed explanation of the detailed information in this file. 5.15.1 Program Sections There is a lot of confusion as to what psects (program sections) actually are and even more confusion as to how they are placed in memory. The following aside takes the form of an analogy and examples, and serves as an introduction to how compilers must generate code and have it allocated into memory. Such an understanding is vital for assembly programmers and understanding “Can’t find space” error messages issued by the linker. Like all analogies, it can be misleading and can only be taken so far, but it relates the main principles of code generation, the linker and sections back to something that you should understand. By the end of this section, you should have a better understanding of: • Why assembly code has to be packed and manipulated in sections • Why the linker packs sections into classes rather than the device memory • Why a “Can’t find space” error message can be issued even though there is plenty of space left in a device’s memory 5.15.1.1 AN ANALOGY Our analogy is based around a company which sells components. Customers throughout the world place orders for these components. The consignments are sent from a central warehouse in shipping containers to a regional office and then delivered to the customer. In the warehouse, a robot assembles each order. The required components are collected and placed in a box of sufficient size. The box is labeled then placed on a conveyor belt. The label on the box indicates the destination city and country, as well as any special packing instructions. At the other end of the conveyor belt, a packing machine reads the labels on the boxes and sorts them by destination city. Thus, all the boxes destined for the same city are collated into one holding bay. Once the day’s order are all processed, the collated boxes in each holding bay are first wrapped in plastic to keep them together. These bundles are then placed into a shipping container bound for that country. As there might be more than one destination city in the same country, there could be many bundles placed in the same container. And so ends another productive day in the warehouse. DS50002053D-page 234 2012-2013 Microchip Technology Inc. C Language Features 5.15.1.2 THE COMPILER EQUIVALENT Let’s now look at the similarities and differences between this situation and the compilation process. Both these processes are compared in Figure 5-3. FIGURE 5-3: THE WAREHOUSE AND COMPILER SIDE BY SIDE ORDER 1 3 x TSC235ORDER 2 2 x RC123a 1 x TSD87 2 x XAM99 10 x TD66 P R ‧‧‧ if(in > 4) scale = 0; ‧‧‧ int result; float scale; PSECT cstackBANK0,class=BANK0,space=1 global __pcstackBANK0 __pcstackBANK0: global ?___ftge ?___ftge: ;@ 0x0 global ?___ftmul ?___ftmul: ;@ 0x0 global ?_frexp ?_frexp: ;@ 0x0 global ___ftmul@f1 ___ftmul@f1: ; 3 bytes @ 0x0 PSECT text,class=CODE fcall _srv movlw (?_barrr2)&0ffh movwf fsr0 movlw (?_srv)&0ffh movwf (??_main+0+0) movf fsr0,w movwf (??_main+0+0+1) movlw 12 movwf (??_main+0+0+2) u60: movf (??_main+0+0),w movwf fsr0 PSECT bssBANK0,class=BANK0,space=1 global __pbssCOMMON __pbssCOMMON: _c: ds 1 _more: ds 8 code generator PSECT text,class=CODE movlw (?_barrr2)&0ffh movwf fsr0 movlw (?_srv)&0ffh movwf (??_main+0+0) movf fsr0,w movwf (??_main+0+0+1) movlw 12 movwf (??_main+0+0+2) u60: movf (??_main+0+0),w int getDat(void) { ... linker In this analogy, the warehouse is likened to the compiler. The robot is akin to the compiler’s code generator application, which turns the C code into assembly code, and the packing machine, the linker, which is responsible for arranging how everything will appear in memory. The packing machine is not actually delivering the bundles of boxes; it is just putting them into containers in the warehouse. In the same way, the linker is not actually placing sections in the device’s memory; it is arranging them in conceptual containers. This analogy is not concerned with what happens outside the warehouse, nor after the compilation process – that is another story. The following sections detail points relevant at different stages of the process. 5.15.1.2.1 Orders and Source Code Both the warehouse and compiler take descriptions of something and use this to produce the final product: The warehouse receives orders, and assembles the components to fulfill that order; the compiler reads (variable and function) definitions in a C program and generates the assembly code to implement each definition. An order from a customer can be for any number of components. In the same way, each C definition can require few or many assembly instructions or directives to be produced. Orders arrive at the warehouse randomly, but the components are always shipped to one of several countries. Source code (variables and functions) can also be written in any order, but the compiled code is always allocated to one of several memory areas. 2012-2013 Microchip Technology Inc. DS50002053D-page 235 MPLAB® XC8 C Compiler User’s Guide 5.15.1.2.2 Boxes, Labels and Sections In our analogy, the components for each order are placed in a box. In the same way, the assembly output generated is also placed into boxes, called program sections (or psects, for short). There are several reasons why code is placed in a section. • • • • • The generated assembly code is more manageable. The ordering of the code sequence within the section is preserved. The ordering of sections with the same name are preserved. Code is easily sorted based on where it needs to reside in memory. Only one command is required to locate an entire section into memory. Any code sequence that must be contiguous is placed in the one section. The output of the compiler will typically appear in many different sections, but all sections with the same name will collate in the order in which they are produced. The compiler has a list of section names and chooses a section based on the code it is generating, see 5.15.2 “Compiler-Generated Psects”. A section is not a physical box, but rather a special directive is used in the assembly code to define the start of each section. The directive signifies the end of the previous box and the start of a new box. A section can hold any amount of assembly code and can even be empty. Both the warehouse boxes and compiler sections are labeled and in both instances, the label indicates a destination rather than the contents. In Figure 5-3 color is used to identify the destination city of a box, or the name of a section. Figure 5-4 shows what a typical box label might look like. The packing machine in the warehouse is only concerned with packing boxes in containers and so, other than the city and country, the customer’s actual address is not important here. FIGURE 5-4: A TYPICAL BOX LABEL BOX paris,country=FRANCE,fragile Figure 5-5 shows an example of the assembly directive that is used to start a new section. The city name is now a section name and the destination country a linker class. FIGURE 5-5: THE SECTION DIRECTIVE AS A BOX LABEL the name of this box special packing instructions for this box PSECT bss0,class=BANK0,space=1,reloc=4 close the previous box, start a new box DS50002053D-page 236 the container this box must be packed in 2012-2013 Microchip Technology Inc. C Language Features 5.15.1.2.3 Down the Conveyor Belt and Object files Once the robot has assembled an order and placed the components in a box, the contents of the box are no longer relevant, and the remaining activities in the warehouse only deal with boxes. It is a similar situation in the compiler: Once assembly code has been placed into a section, the instructions are no longer relevant during the link process. The linker only handles sections and is oblivious to each section’s contents.1 5.15.1.2.4 Sorting Boxes and Sections In the warehouse, all the boxes are sorted by destination city. After the day’s orders are processed, all these similar boxes are wrapped so they can be easily kept together. The concept is the same in the compiler domain: Sections are sorted based on the section’s name, and are merged into one larger section by the linker. The order in which the sections come together strictly follows the order in which they are produced. 5.15.1.2.5 Loading the Containers and Classes In the warehouse, the bundled boxes are loaded into the shipping containers. One or more shipping containers will be provided for each country. If there are orders for more than one city in the same country, then the bundled boxes for those cities can share the same container (as long as there is room). The linker does a similar thing: It arranges the collated sections in the classes. Typically, several sections are placed in each class. The classes represent the device memory, but might take into account addressing modes or banking/paging restrictions present in the target device. Classes are often smaller than the memory space in which they are theoretically located. Think of a large cargo ship. Although the ship can be very long, the biggest object it can hold is determined by the size of the shipping containers it transports. The shipping company cannot be able to carry an item due to size constraints even though the item is a fraction the size of the vessel. 5.15.1.3 SECTIONS AT WORK Now that we have a basic understanding of the concepts, let’s work through a simple example to help see how these work when compiling. An engineer has written a program. Exactly what it does or how it is written is not important. The target device has only one bank of RAM and one page of flash. In our very simplified example, the compiler will choose from the sections listed in Table 5-18. This table also shows the linker class associated with each section and the memory location in the device the class represents. TABLE 5-18: Section name SECTION NAMES FOR OUR SIMPLE EXAMPLE Contents Linker class Memory location text Executable code CODE Flash bss0 Variables that need to be cleared BANK0 RAM data0 Variables that need to be initialized BANK0 RAM idata Initialized variable’s values CODE Flash 1. Inside the compiler, there are actually a few steps between the code generator and linker. Assembly code is passed first to the assembler; however, the object files produced, which are passed to the linker, preserve all the sections and can be ignored in this exercise. The assembler optimizer can alter assembly instructions, but this is only after reading and interpreting the code – something the linker cannot do. Modified code is still functionally identical to the original. Other optimizations can further merge or split sections, but this is not important to a basic understanding of the concepts. 2012-2013 Microchip Technology Inc. DS50002053D-page 237 MPLAB® XC8 C Compiler User’s Guide The code generator starts to read the program description in C and produces the assembly code required to implement this on the target device. The first thing the code generator encounters is the C definition for a variable with no initial value. All variables like this will be kept together so that clearing them at startup will be easier and more efficient. The code generator outputs the directive that starts a new section. Its internal table of section names indicates that bss0 is the appropriate choice and that this section is placed in the BANK0 class. The directive looks like this. PSECT bss0,class=BANK0 Code follows this directive to define the variable: It merely consists of a label and another directive to reserve the required amount of memory. Altogether, this might look like: PSECT bss0,class=BANK0 myVariable: DS 2 ; reserve 2 bytes for this object The code generator continues and sees yet another uninitialized variable. As this variable will also use the same section, the code generator can keep adding to the current section and so immediately outputs another label and directive to reserve memory for the second variable. Now a function definition is encountered in the C source. The code generator sees that output now needs to be placed in the text section and outputs the following directive. PSECT text,class=CODE This directive ends the bss0 section and starts the text section. Any code following will be contained in the text section. Code associated with the function is generated and placed into this section. Moving on, the code generator encounters a variable which is initialized with a value. These variables will have their initial values stored in flash and copied as a block into the RAM space allocated to them. By using one section for the RAM image and another for the initial values, the compiler can ensure that all the initial values will align after linking, as the sections will be collated in order. The compiler first outputs the directive for the section with the name data0. It then outputs a label and directive to reserve memory for the variable in RAM. It next changes to the idata section which will hold all the initial values. Code is output that will place the initial value in this section. Notice that the code associated with initialized variables is output into two separate sections. PSECT data0,class=BANK0 _StartCount: DS 2 ; space in RAM for the variable PSECT idata,class=CODE DB 012h ;initial values places as bytes DB 034h ; which will be copied to RAM later The code generator reads yet another initialized variable. It selects the data0 section, outputs the label and memory reservation directive, then selects the idata section and stores the initial value. This process continues until all the program has been generated. The output is passed to the linker (via the assembler) which then looks for the sections it contains. The linker sorts each section as it is encountered. All text sections are collated, as well as the bss0, data0 and idata sections. The order in which they will be assembled will match the order in which they are passed to the linker, which in turn will be the order in which they were produced by the code generator. There is now one collated text, bss0, data0 and idata section. DS50002053D-page 238 2012-2013 Microchip Technology Inc. C Language Features These sections are now located in memory. The linker is passed options (from the compiler driver) which define the linker classes and the memory ranges they represent, see Section 7.2.1 “-Aclass =low-high,...”1. For our device, the linker options might look like this. -ACODE=0-7ffh -ABANK0=20h-6fh So for example, the BANK0 class covers memory from 20h to 6fh. You can see all the options passed to the linker in the map file, see Section 7.4 “Map Files”. The linker “fits” each section into the memory associated with the class it is associated with. It might, for example, place the text section at address 0 in the CODE class, then immediately follow it with the idata section at address 374h, for example. The bss0 and data0 sections will similarly be placed in the BANK0 class. All the sections are now allocated an address within a class. The addresses of symbols can now be determined and ultimately a hex file is produced. The compiler’s job is over. 5.15.1.4 MORE ADVANCED TOPICS Let’s look at more complex issues with linking now that we understand the basics. 5.15.1.4.1 Allocation at Specific Locations We have seen in the preceding analogy that the linker places sections in their corresponding class when it comes time to determine where they will live in memory. This is how most sections are linked, but there are some exceptions. Code to be executed on Reset or an interrupt are examples. They cannot just be placed anywhere in a class; they must be located at specific addresses. So how are these positioned by the linker? Any code that is to be linked at a particular address is placed in a section in the usual way. These sections will even have a class associated with them, but allocation anywhere in this class can be overridden by a special linker option which tells the linker to place the section at a specific address. In terms of our previous analogy, think of the special linker options as being explicit instructions given to the packing machine as to where in a container to place the box. We will see in the next example the linker options to place a section explicitly. -preset_vec=0h -pint_text=04h Note that if a section is linked in this way, the linker will follow these instructions strictly. It will warn if it is asked to place a section over the top of another, but since there is no container, which essentially represents a memory range, the linker cannot check for sections extending past the device’s memory limits. 5.15.1.4.2 Where Classes and Containers Differ Containers and linker classes differ in one important aspect: Linker classes are conceptual and merely represent the memory of a device; they are not physical storage. The compiler can, and often does, use more than one class to represent the same memory range. This is illustrated in Section 5.15.1.5 “More Advanced Sections at Work” where the example uses CODE and CONST classes for flash memory. Although classes can cover the same range, typically the size of the containers vary. This allows code with different restrictions and requirements to be accommodated. 1. Remember these are linker options and you cannot pass these straight to the compiler driver. You can, however, encapsulate these options in the driver’s -L- options, see Section 4.8.6 “-L-: Adjust Linker Options Directly”. 2012-2013 Microchip Technology Inc. DS50002053D-page 239 MPLAB® XC8 C Compiler User’s Guide When the memory ranges of classes overlap, allocating to one will also mark as being used memory from the other. In fact, when any memory is allocated by the linker by whatever means, it checks each class to see if it covers this memory and marks it as being used. This is quite a difference concept to physical containers. 5.15.1.4.3 Multi-bin containers Linker classes usually define one memory range, but there are instances where a class defines multiple memory ranges. You can think of this as several separate containers, but all with identical shipping destinations. Memory ranges in the class do not need to be contiguous. The compiler typically uses a multi-range class to represent program memory that is paged. The boundaries in the memory ranges coincide with the page boundaries. This prevents sections from crossing a page boundary. The compiler could use a similar class for banked RAM, but code can be considerably reduced in size if the destination bank of each variable is known by the code generator. You will usually see a separate class defined for each bank, and dedicated sections that are associated with these classes. The code generator will allocate a bank for each variable and choose a section destined for the class that represents that bank. 5.15.1.5 MORE ADVANCED SECTIONS AT WORK Let’s build on the previous example. Our target device now has two banks of RAM and two pages of flash, and Table 5-19 shows the extended list of sections the compiler now uses. These sections reference new classes, also shown in the table. TABLE 5-19: Section name SECTION NAMES FOR OUR EXTENDED EXAMPLE Contents Linker class Memory location textn Executable code CODE Flash bss0 Variables that need to be cleared BANK0 RAM bss1 Variables that need to be cleared BANK1 RAM data0 Variables that need to be initialized BANK0 RAM idata Initialized variable’s values CODE Flash reset_vec Code associated with Reset CODE Flash const Read-only variables CONST Flash init Runtime startup code CODE Flash int_text Interrupt function code CODE Flash The compiler operates as it did in the previous example, selecting and producing a section directive prior to generating the assembly code associated with the C source currently being compiled. DS50002053D-page 240 2012-2013 Microchip Technology Inc. C Language Features The assembly code associated with ordinary functions is still placed in a “text” section, but as there are now two pages of flash, we have to ensure that both pages can be used. If each function was placed into the same section, they will be merged by the linker and that section would grow until it is too large to fit into either page. To ensure that all the “text” sections do not merge, each function is placed in its own unique numbered section: text0, text1, text2, etc. As these sections do not have the same name, they will not be merged by the linker. The linker option to define the CODE class might now look like: -ACODE=0-7ffhx2 which tells the linker that CODE represents two pages of memory: one from 0 to 7ffh and another from 800h to fffh. This specification indicates that there is some sort of memory boundary that occurs at address 800h (the devices internal memory page) and is very different to a class definition that reads -ACODE=0-fffh, which covers the same memory range, but which does not have the boundary. The linker will try allocating each textx section to one page (class memory range); if it does not fit, it will try the other. If an interrupt function is encountered, the int_text section is selected for this code. As this is separate to the sections used to hold ordinary functions, it can be linked explicitly at the interrupt vector location. Assuming that the interrupt vector is at address 4, the linker option to locate this section might look like the following, see Section 7.2.19 “-Pspec”. -pint_text=4h For simplicity in this example, initialized variables are treated as they were in the previous example, even though there are now two RAM banks; i.e., they are always allocated in the first bank of RAM. In the previous example we ignored the code that would have copied the initial values from flash into RAM. This code is executed after Reset and before the function main, and forms part of the runtime startup code. It reads each initial value from flash and writes this to the corresponding variable in the RAM space. Provided the order in which the variables are allocated memory in RAM matches the order their initial values are allocated in flash, a single loop can be used to perform the copy. Even though the variables might be defined at many places in the source code, the order in memory of each variable and value will be preserved since the compiler uses sections to hold the code associated with each. This runtime startup code is output into a section called init. Code which jumps to the runtime startup codes is placed in the reset_vec section, which is linked to the Reset location. By linking these sections in the same page, the jump from one section to another will be shorter and faster. The linker options to make this happen might look like: -preset_vec=0 -pinit=int_text which says that the Reset vector code is linked to address 0 (which is the Reset vector location) and that the init section, which contains the runtime startup code, should be linked immediately after the interrupt code in the int_text section. If the int_text section is empty; i.e., there is no interrupt code defined in the program, then init will be placed at address 4. 2012-2013 Microchip Technology Inc. DS50002053D-page 241 MPLAB® XC8 C Compiler User’s Guide Previously all uninitialized variables were placed in the bss0 section. Now the code generator first checks that there will actually be enough room in bank 0 memory before doing so. If not, it chooses the bss1 section that will ultimately be linked into bank 1 memory. The code generator keeps track of any object allocated to RAM so it can maintain the amount of free memory in each RAM bank. For these variables, the linker allocates the sections to memory, but it is the code generator that decides which section will be used by each variable. Thus, both applications play a part in the memory allocation process. In this example, we also consider const variables which are stored in flash, not RAM. Each byte of data is encapsulated in a RETLW instruction that return the byte in the W register. Code is needed to access each byte of a variable or array. One way of doing this is a “computed goto” which involves loading the W register with an offset into the block of data and adding this to the PC.(The Microchip application note AN556 has examples of how this can be done for several devices.) A computed goto requires that the destination address (the result of adding W and PC) must not cross over a 256 word boundary (i.e., the addresses 100h, 200h, 300h, etc.). This requirement can be met using sections and a class. In this example a new class, called CONST, is created and defined as follows -ACONST=0-0ffhx16 which is to say that CONST is a container 100h long, but there are 16 of them one after the other in memory, so 0-ffh is one container, 100-1ffh is another, etc. We have the compiler place all the RETLW instructions and the computed goto code into the const section, which are linked into this class. The section can be located in any of the 16 containers, but must fit entirely within one. In this example, the compiler only allows one block of const data. It could be made to allow many by having each block of const data in a unique numbered section as we did for the text sections (e.g., const1, const2, etc.). Thus each sections could remain independent and be allocated to any memory bin of the CONST class. 5.15.1.6 EXPLAINING COMMON LINKER ERRORS AND PROBLEMS We can also use our knowledge to help explain some common linker problems and error messages. 5.15.1.6.1 Not Finding Space A common linker error is, “can’t find x words for psect ‘abc’ in class ‘XYZ’,” which we can now think of as, “can’t find 3 cubic feet for the boxes ‘paris’, in container ‘FRANCE’.” The most obvious reason for this error is that the containers have been steadily filling and have finally run out of free space, i.e., the total amount of code or data you need to store exceeds the amount of memory on the device. Another reason is that a box is larger than the container(s) in which it has to be placed. If this is the case, the section will never fit, even if the entire memory space is empty. This situation might occur when defining a very large C function, RAM or const array resulting in an abnormally large section. Other possible sources include large switch statements or even a switch with too many case labels. DS50002053D-page 242 2012-2013 Microchip Technology Inc. C Language Features The hold of a ship or aircraft might be a very large space, but freight is always packed into shipping containers and it is the size of the shipping container that dictates the maximum size of a object that can be transported. In the same way, the total amount of memory on a target device is irrelevant if sections must first be allocated to a class. Classes can seem restrictive, but without them, code will typically be less efficient or can simply fail to work altogether. The computed goto is a good example. If the table of instructions in a computed goto crosses a 100h boundary, it will fail due to the limited jump range of the instruction set. This space error can also occur if there are many free spaces remaining in containers, but none are large enough to hold the section. This can be confusing since the total amount of free space can be larger than the section to be placed. In the same way that boxes cannot be removed from a bundle or unpacked, if a section does not fit into any remaining space, it cannot be split by the linker and an error will result. The error message indicates that largest free space that is still available. In the same way that the label on a box can indicate special packing instructions, e.g., “fragile – pack at the top of the container”, or “this way up”, etc, a section can also indicate special memory allocation instructions. One of the most common requirements is that a section must start on an address boundary, see the reloc PSECT flag in Section 6.4.9.3.13 “Reloc”. If a section has to fit into a class, but also has to be aligned on an address, this makes it much more difficult to locate and increases the chance that this error is produced. This is also the case if other sections or objects have been placed in the middle of a container, as we saw in Section 5.15.1.4.1 “Allocation at Specific Locations”. 5.15.1.6.2 Not Being Shipped in the Right Container Clearly, boxes will not be delivered correctly if they are placed in the wrong container. So too, code cannot run if it is placed in the wrong class or address. The compiler will always ensure that code and data is allocated correctly, but it is possible to manually change linker options. The code generator assumes that sections will be located in certain memory areas and the code it generates can rely on this being the case. Typically, sections placed in RAM must remain in the bank in which they were originally destined. Sections containing executable code possibly more forgiving, but some have very specific requirements. Remember, the linker will allow any allocation you indicate; it has no way of checking if what you tell it is correct. Always take care if you override linker options and reposition a section. Section 5.15.2 “Compiler-Generated Psects” lists the common sections used by the compiler. it also indicates restrictions on how these sections can be linked. 5.15.1.6.3 Doing Things by Hand So far we have discussed assembly code produced by the code generator and how it is linked into memory; now, we consider hand-written assembly. Imagine that in the warehouse an order is not processed by the robot, but by a human worker. The components ordered are assembled by hand, packed in a box by hand and then placed on the conveyor belt along with those boxes packed by the robot. This is quite similar to when there is hand-written assembly code in a project: the linker is passed a number of sections, some created by the code generator and some by a human. Mistakes can be made by the warehouse worker or the assembly programmer. In the warehouse, the worker might not use a box and place the components loose on the conveyor belt, or a box might be used but it is not labeled, or it might be labeled incorrectly. In the compiler’s domain, assembly code can be written without a section, or it is in a section but with the wrong (or no) class, or the section can have incorrect allocation instructions. 2012-2013 Microchip Technology Inc. DS50002053D-page 243 MPLAB® XC8 C Compiler User’s Guide If assembly code is not in a section, the compiler will actually place it into a default section. But since there are no specific instructions as to what to do with this section, it could be linked anywhere. Such sections are like boxes labeled “ship to anywhere you want”. As a rule of thumb, put all assembly code inside a section, but some directives (e.g., GLOBAL or EQU) do not generate code and can be placed anywhere. The easiest way to write and locate hand-written assembly code is to associate the section you create with an existing linker class that represents a suitable memory area. This means the default linker options do not need to be altered. The association is made using the class flag of the PSECT directive, see Section 6.4.9.3.3 “Class”. If a section is to be placed at an explicit address rather than having it placed anywhere in a class, the class flag should still be used. A list of linker classes and the memory they represent is given in Section 5.15.3 “Default Linker Classes”. Even if you place your code into a section and use an appropriate class, other flags can be necessary to correctly link the code. The most important section flags are the delta, see Section 6.4.9.3.4 “Delta”, reloc, see Section 6.4.9.3.13 “Reloc” and space, see 6.4.9.3.15 “Space” flags. If these are incorrectly specified, the code not be positioned correctly and will almost certainly fail. 5.15.2 Compiler-Generated Psects The code generator places code and data into psects (program sections) with standard names, which are subsequent positioned by the default linker options. The linker does not treat these compiler-generated psects any differently to a psect that has been defined by yourself. A psect can be created in assembly code by using the PSECT assembler directive (see Section 6.4.9.3 “PSECT”). The code generator uses this directive to direct assembly code it produces into the appropriate psect. For an introductory guide to psects, see Section 5.15.1 “Program Sections”. Some psects, in particular the data memory psects, use special naming conventions. For example, take the bss psect. The name bss is historical. It holds uninitialized variables. However, there can be some uninitialized variables that will need to be located in bank 0 data memory; others can need to be located in bank 1 memory. As these two groups of variables will need to be placed into different memory banks, they will need to be in separate psects so they can be independently controlled by the linker. In addition, the uninitialized variables that are bit variables need to be treated specially so they need their own psect. So there are a number of different psects that all use the same base name, but which have prefixes and suffixes to make them unique. The general form of these psect names is: [bit]psectBaseNameCLASS[div] where psectBaseName is the base name of the psect, such as bss or data. The CLASS is a name derived from the linker class (see Section 7.2.1 “-Aclass =low-high,...”) in which the psect will be linked, e.g., BANK0. The prefix bit is used if the psect holds bit variables. So there can be psects like: bssBANK0, bssBANK1 and bitbssBANK0 defined by the compiler to hold the uninitialized variables. Note that eeprom-qualified variables can define psects called bssEEDATA or dataEEDATA, for example, in the same way. Any psect using the class suffix EEDATA is placed in the HEX file and is burnt into the EEPROM space when you program the device. DS50002053D-page 244 2012-2013 Microchip Technology Inc. C Language Features If locations in a bank are reserved or are taken up by absolute objects for example, a psect cannot be formed over the entire bank. Instead, a separate psect will be used to represent the free memory on either side of the used memory. The letters l (elle) and h are used as the div field in the psect name to indicate if it is the lower or higher division. Thus you might see bssBANK0l and bssBANK0h psects if a split took place. If you are unsure which psect holds an object or code in your project, check the assembly list file (see Section 6.6.1 “General Format”) The contents of these psects are described below, listed by psect base name. 5.15.2.1 PROGRAM SPACE PSECTS checksum – this is a psect that is used to mark the position of a checksum that has been requested using the --CHECKSUM option. See Section 4.8.16 “--ASMLIST: Generate Assembler List Files” for more information. The checksum value is added after the linker has executed so you will not see the contents of this psect in the assembly list file, nor specific information in the map file. Linking this psect at a non-default location will have no effect on where the checksum is stored, although the map file will indicate it located at the new address. Do not change the default linker options relating to this psect. cinit – used by the C initialization runtime startup code. Code in this psect is output by the code generator along with the generated code for the C program and does not appear in the runtime startup assembly module. This psect can be linked anywhere within a program memory page, provided it does not interfere with the requirements of other psects. config – used to store the Configuration Words. This psect must be stored in a special location in the HEX file. Do not change the default linker options relating to this psect. const – these PIC18-only psects hold objects that are declared const and string literals which are not modifiable. It is used when the total amount of const data in a program exceeds 64k. This psect can be linked anywhere within a program memory page, provided it does not interfere with the requirements of other psects. eeprom (PIC18: eeprom_data) – used to store initial values in the EEPROM memory. Do not change the default linker options relating to this psect. idata – these psects contain the ROM image of any initialized variables. These psects are copied into the data psects at startup. In this case, the class name is used to describe the class of the corresponding RAM-based data psect. These psects will be stored in program memory, not the data memory space. These psects are implicitly linked to a location that is anywhere within the CODE linker class. The linker options can be changed allowing this psect to be placed at any address within a program memory page, provided it does not interfere with the requirements of other psects. idloc – used to store the ID location words. This psect must be stored in a special location in the HEX file. Do not change the default linker options relating to this psect. 2012-2013 Microchip Technology Inc. DS50002053D-page 245 MPLAB® XC8 C Compiler User’s Guide init – used by assembly code in the runtime startup assembly module. The code in this and the cinit define the runtime startup code. PIC18 devices also use an end_init psect, which contains the code which transfers control to the main function. If no interrupt code is defined code from the Reset vector can “fall through” into this psect. It is recommended that the default linker options relating to this psect are not changed in case this situation is in effect. intcode, intcodelo – are the psects which contains the executable code for the high-priority (default) and low-priority interrupt service routines, respectively. These psects are linked to interrupt vector at address 0x8 and 0x18, respectively. Do not change the default linker options relating to these psects. See Section 4.8.20 “--CLIST: Generate C Listing File” if you want to move code when using a bootloader. intentry – contains the entry code for the interrupt service routine which is linked to the interrupt vector. This code saves the necessary registers and jumps to the main interrupt code in the case of mid-range devices; for enhanced mid-range devices this psect will contain the interrupt function body. (PIC18 devices use the intcode psects.) This psect must be linked at the interrupt vector. Do not change the default linker options relating to this psect. See the --CODEOFFSET option Section 4.8.20 “--CLIST: Generate C Listing File” if you want to move code when using a bootloader. jmp_tab – only used for the baseline processors, this is a psect used to store jump addresses and function return values. Do not change the default linker options relating to this psect. maintext – this psect will contain the assembly code for the main() function. The code for main() is segregated as it contains the program entry point. Do not change the default linker options relating to this psect as the runtime startup code can “fall through” into this psect which requires that it be linked immediately after this code. mediumconst – these PIC18-only psects hold objects that are declared const and string literals which are not modifiable. Used when the total amount of const data in a program exceeds 255 bytes, but does not exceed 64k. This psect can be linked anywhere in the lower 64k of program memory, provided it does not interfere with the requirements of other psects. For PIC18 devices, the location of the psect must be above the highest RAM address. powerup – contains executable code for a user-supplied powerup routine. Do not change the default linker options relating to this psect. reset_vec – this psect contains code associated with the Reset vector. Do not change the default linker options relating to this psect as it must be linked to the Reset vector location of the target device. See the --CODEOFFSET option Section 4.8.20 “--CLIST: Generate C Listing File”, if you want to move code when using a bootloader. DS50002053D-page 246 2012-2013 Microchip Technology Inc. C Language Features reset_wrap – for baseline PIC devices, this psect contains code which is executed after the device PC has wrapped around to address 0x0 from the oscillator calibration location at the top of program memory. Do not change the default linker options relating to this psect as it must be linked to the Reset vector location of the target device. smallconst – these psects hold objects that are declared const and string literals which are not modifiable. Used when the total amount of const data in a program is less than 255 bytes. This psect can be linked anywhere in the program memory, provided it does not cross a 0x100 boundary and it does not interfere with the requirements of other psects. For PIC18 devices, the location of the psect must be above the highest RAM address. strings – the strings psect is used for const objects. It also includes all unnamed string literals. This psect is linked into ROM, since the contents do not need to be modified. This psect can be linked anywhere in the program memory, provided it does not cross a 0x100 boundary or interfere with the requirements of other psects. stringtext – the stringtext psect is used for const objects when compiling for baseline devices. This psect is linked into ROM, since the contents do not need to be modified. This psect must be linked within the first half of each program memory page. textn – these psects (where n is a decimal number) contain all other executable code that does not require a special link location. These psects can be linked anywhere in the program memory, provided they does not interfere with the requirements of other psects. xxx_text – defines the psect for a function that has been made absolute; i.e., placed at an address. xxx will be the assembly symbol associated with the function. For example if the function rv() is made absolute, code associated with it will appear in the psect called _rv_text. As these psects are already placed at the address indicated in the C source code, the linker options that position them should not be changed. xxx_const – defines the psect for const object that has been made absolute; i.e., placed at an address. xxx will be the assembly symbol associated with the object. For example, if the array nba is made absolute, values stored in this array will appear in the psect called _nba_const. As these psects are already placed at the address indicated in the C source code, the linker options that position them should not be changed. 2012-2013 Microchip Technology Inc. DS50002053D-page 247 MPLAB® XC8 C Compiler User’s Guide 5.15.2.2 DS50002053D-page 248 DATA SPACE PSECTS nv – these psects are used to store variables qualified persistent. They are not cleared or otherwise modified at startup. These psects can be linked anywhere in their targeted memory bank and should not overlap any common (unbanked memory) that the device supports if it is a banked psect. bss – these psects contain any uninitialized variables. These psects can be linked anywhere in their targeted memory bank and should not overlap any common (unbanked memory) that the device supports if it is a banked psect. data – these psects contain the RAM image of any initialized variables. These psects can be linked anywhere in their targeted memory bank and should not overlap any common (unbanked memory) that the device supports if it is a banked psect. cstack – these psects contain the compiled stack. On the stack are auto and parameter variables for the entire program. See Section 5.5.2.2.1 “Compiled Stack Operation”, for information on the compiled stack. These psects can be linked anywhere in their targeted memory bank and should not overlap any common (unbanked memory) that the device supports if it is a banked psect. stack – this psect is used as a placeholder for the software stack. This stack is dynamic and its size is not known by the compiler. As described in 5.3.4.2 “Data Stacks” this psect is typically allocated the remainder of the free data space so that the stack may grow as large as possible. This psect may be linked anywhere in the data memory, but adjusting the default linker options for this psect may limit the size of the software stack. Any overflow of the software stack may cause code failure. 2012-2013 Microchip Technology Inc. C Language Features 5.15.3 Default Linker Classes The linker uses classes to represent memory ranges. For an introductory guide to psects and linker classes, see Section 5.15.1 “Program Sections”. The classes are defined by linker options, see Section 7.2.1 “-Aclass =low-high,...” passed to the linker by the compiler driver. Psects are typically allocated space in the class they are associated with. The association is made using the class flag of the PSECT directive, see Section 6.4.9.3.3 “Class”. Alternatively, a psect can be explicitly placed into a class using a linker option, see Section 7.2.19 “-Pspec”. Classes can represent a single memory range, or multiple ranges. Even if two ranges are contiguous, the address where one range ends and the other begins forms a boundary and psects placed in the class can never cross such boundaries. You will see classes that cover the same addresses, but will be divided into different ranges and have different boundaries. This is to accommodate psects whose contents were compiled under assumptions about where they would be located in memory. Memory allocated from one class will also be reserved from other classes that specify the same memory. To the linker, there is no significance to a class name or the memory it defines. Memory will be subtracted from these classes if using the --ROM or --RAM options, see Section 4.8.50 “--ROM: Adjust ROM Ranges” and Section 4.8.49 “--RAM: Adjust RAM Ranges”, to reserve memory. When specifying a debugger, such as an ICD, see Section 4.8.22 “--DEBUGGER: Select Debugger Type”, memory can also be removed from the ranges associated with some classes so that this memory is not used by your program. Although you can manually adjust the ranges associated with a class, this is not recommended. Never change or remove address boundaries specified by a class definition option. Below are the linker classes that can be defined by the compiler. Not all classes can be present for each device. 2012-2013 Microchip Technology Inc. DS50002053D-page 249 MPLAB® XC8 C Compiler User’s Guide 5.15.3.1 PROGRAM MEMORY CLASSES CODE — consists of ranges that map to the pages on the target device. Thus, it is typically used for psects containing executable code. On baseline devices, it can only be used by code that is accessed via a jump table. ENTRY — is used mainly by baseline devices for psects containing executable code that is accessed via a CALL instruction (calls can only be to the first half of a page). The class is defined in such a way that it is the size of a page, but psects it holds will be positioned so that they start in the first half of the page. This class is also used in mid-range devices and will consist of many ranges, each 0x100 words long and aligned on a 0x100 boundary. Thus, it is useful for psects whose contents cannot span a 0x100 word boundary. STRING — consists of ranges that are 0x100 words long and aligned on a 0x100 boundary. Thus, it is useful for psects whose contents cannot span a 0x100 word boundary. STRCODE — defines a single memory range that covers the entire program memory. It is useful for psects whose content can appear in any page and can cross page boundaries. CONST — consists of ranges that are 0x100 words long and aligned on a 0x100 boundary. Thus, it is useful for psects whose contents cannot span a 0x100 word boundary. 5.15.3.2 DATA MEMORY CLASSES RAM — consist of ranges that cover all the general purpose RAM memory of the target device, but excluding any common (unbanked) memory. Thus, it is useful for psects that must be placed in general-purpose banked RAM, but can be placed in any bank. BIGRAM — consists of a single memory range that is designed to cover the linear data memory of enhanced mid-range devices, or the entire available memory space of PIC18 devices. It is suitable for any psect whose contents are accessed using linear addressing or which does not need to be contained in a single data bank. ABS1 — consist of ranges that cover all the general purpose RAM memory of the target device, including any common (unbanked) memory. Thus, it is useful for psects that must be placed in general purpose RAM, but can be placed in any bank or the common memory, BANKx (where x is a bank number) — each consist of a single range that covers the general purpose RAM in that bank, but excluding any common (unbanked) memory. COMMON — consists of a single memory range that covers the common (unbanked) RAM, if present. SFRx (where x is a bank number) — each consists of a single range that covers the SFR memory in that bank. These classes would not typically be used by programmers as they do not represent general purpose RAM. DS50002053D-page 250 2012-2013 Microchip Technology Inc. C Language Features 5.15.3.3 MISCELLANEOUS CLASSES CONFIG — consists of a single range that covers the memory reserved for configuration bit data in the hex file. This class would not typically be used by programmers as it does not represent general purpose RAM. IDLOC — consists of a single range that covers the memory reserved for ID location data in the hex file. This class would not typically be used by programmers as it does not represent general purpose RAM. EEDATA — consists of a single range that covers the EEPROM memory of the target device, if present. This class would typically be used for psects that contain data that is to be programmed into the EEPROM. 5.15.4 Changing and Linking the Allocated Section Section 5.15.2 “Compiler-Generated Psects” lists the default sections the compiler uses to hold objects and code. You can change the default section of a function or variable several ways. This is typically done when the object has special and unique linking requirements that cannot be addressed by existing compiler features. Both the __section() specifier1 or the #pragma psect (Section 5.14.4.8 “The #pragma psect Directive”) allow you to have a variable or function redirected into a user-define psect (section). There are several differences in their operation which are described below. Use of the pragma will create a user-defined psect with identical psect flags to the original (standard) psect referenced in the directive. A consequence of this is this the psect created using the pragma must be linked into the same memory space as that which would hold the original psect; in particular, the psect must be linked into the same data bank, for the case of RAM-based psects. The compiler assumes this is the case. So, for example, you cannot use the pragma to place a variable normally linked into bank 0 data memory into a user-defined psect and then link that psect into bank 2. New psects created by the specifier will have no flags associated with them. In addition, the compiler will not make assumptions about the final location of the new psect. Thus, for example, you can define a variable, redirect it into a user-defined psect using the __section() specifier and link the psect into any data bank. Note, however, that the code used to access the relocated variable can be less efficient than the code used to access variable without the specifier. Regardless of which method you use to place an object or function into a unique psect, the name of this psect must be a valid identifier in the assembler’s name space, i.e., it cannot have a name which is the same as that of an assembler directive, control, or directive flag. Variables that use the __section() specifier will be cleared or initialized (based on how they are defined) in the usual way by the runtime startup code (see Section 4.4.2 “Startup and Initialization”). For the case of initialized variables, the compiler will automatic allocation an additional psect (whose name will be the same as the new psect prefixed with the letter i), which will contain the initial values. You can need to link this psect explicitly. 1.This specifier is part of the CCI (see Chapter 2. “Common C Interface”), but is also available as part of the native MPLAB XC8 syntax. 2012-2013 Microchip Technology Inc. DS50002053D-page 251 MPLAB® XC8 C Compiler User’s Guide Variables placed in a new section using the pragma will not be cleared or initialized. In fact, initialized variables must never be redirected to a new psect using the pragma. The following are examples of a variable and function allocated to a non-default section. int __section("myData") foobar; int __section("myCode") helper(int mode) { /* ... */ } You must reserve memory, and locate via an explicit linker option, any new psect created with a __section() specifier. So, for example, if you wanted to place the sections created in the above example, you could use the following driver options: -L-pmyData=0200h -L-AMYCODE=50h-3ffh -L-pmyCode=MYCODE which will place the section myData at address 0x200, and the section myCode anywhere in the range 0x50 to 0x3ff represented by the linker class. See Section 7.2 “Operation” for linker options that can be passed using the -L- driver option (Section 4.8.6 “-L-: Adjust Linker Options Directly”). Since user-defined psects created using the pragma inherit the flags of the original psect, they will typically be linked into their original linker class without the need of an explicit linker option. However, if you do link them explicitly, ensure that you follow the original psect linking guidelines indicated in Section 5.15.2 “Compiler-Generated Psects”. 5.15.5 Replacing Library Modules The MPLAB XC8 C compiler comes with a librarian, LIBR, which allows you to unpack a library file and replace modules with your own modified versions. See Section 8.2 “Librarian”. However, you can easily replace a library module that is linked into your program without having to do this. If you add a source file to your project and it contains the definition for a routine with the same name as a library routine, then the library routine will be replaced by your routine. This is due to the way the compiler scans source and library files. When trying to resolve a symbol (a function name, or variable name, for example) the compiler first scans all the source modules for the definition. Only if it cannot resolve the symbol in these files does it then search the library files. If the symbol is defined in a source file, the compiler will never actually search the libraries for this symbol. No error will result if the symbol was present in both your source code and the library files. This cannot be true if a symbol is defined twice in source files and an error can result if there is a conflict in the definitions. All library source code is written in C, and the p-code library files that contain these library routines are actually passed to the code generator, not the linker, but both these applications work in the way described above in resolving library symbols. You cannot replace a C library function with an equivalent written in assembly code using the above method. If this is required, you will need to use the librarian to edit or create a new library file. DS50002053D-page 252 2012-2013 Microchip Technology Inc. C Language Features 5.15.6 Signature Checking The compiler automatically produces signatures for all functions. A signature is a 16-bit value computed from a combination of the function’s return data type, the number of its parameters and other information affecting the calling sequence for the function. This signature is generated and placed the object code whenever a function is referenced or defined. At link time, the linker will report any mismatch of signatures, which will indicate a discrepancy between how the function is defined. MPLAB XC8 is only likely to issue a mismatch error from the linker when the routine is either a precompiled object file or an assembly routine. Other function mismatches are reported by the code generator. It is sometimes necessary to write assembly language routines which are called from C using an extern declaration. Such assembly language functions should include a signature which is compatible with the C prototype used to call them. The simplest method of determining the correct signature for a function is to write a dummy C function with the same prototype and check the assembly list file using the --ASMLIST option (see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”). For example, suppose you have an assembly language routine called _widget which takes a char argument and returns a char. The prototype used to call this function from C would be: extern char widget(char); Where a call to _widget is made in the C code, the signature for a function with one char argument and a char return value would be generated. In order to match the correct signature, the source code for widget needs to contain an assembler SIGNAT directive which defines the same signature value. To determine the correct value, you would write the following code into a dummy file: char widget(char arg1) { } The resultant assembler code seen in the assembly list file includes the following line: SIGNAT _widget,4217 The SIGNAT directive tells the assembler to include a record in the .obj file which associates the signature value 4217 with symbol _widget. The value 4217 is the correct signature for a function with one char argument and a char return value. If this directive is copied into the assembly source file which contains the _widget code, it will associate the correct signature with the function and the linker will be able to check for correct argument passing. If a C source file contains the declaration: extern char widget(long); then a different signature will be generated and the linker will report a signature mis-match which will alert you to the possible existence of incompatible calling conventions. 2012-2013 Microchip Technology Inc. DS50002053D-page 253 MPLAB® XC8 C Compiler User’s Guide 5.15.7 Linker-Defined Symbols The linker defines some special symbols that can be used to determine where some psects where linked in memory. These symbols can be used in code, if required. The link address of a psect can be obtained from the value of a global symbol with name __Lname (two leading underscores) where name is the name of the psect. For example, __LbssBANK0 is the low bound of the bssBANK0 psect. The highest address of a psect (i.e., the link address plus the size) is represented by the symbol __Hname. If the psect has different load and link addresses, the load start address is represented by the symbol __Bname. Not all psects are assigned these symbols, in particular those that are not placed in memory by a -P linker option. See Section 7.2.19 “-Pspec”. Psect names can change from one device to another. Assembly code can use these symbol by globally declaring them, for example: GLOBAL __Lidata and C code could use them by declaring a symbol such as the following. extern char * _Lidata; Note that there is only one leading underscore in the C domain, see Section 5.12.3.1 “Equivalent Assembly Symbols”. As the symbol represents an address, a pointer is the typical type choice. DS50002053D-page 254 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Chapter 6. Macro Assembler 6.1 INTRODUCTION Two macro assemblers are included with the MPLAB XC8 C Compiler to assemble source files for all 8-bit PIC devices. The operation and assembler directives are almost identical for both assemblers. The appropriate assembler application is invoked when you use the compiler driver to build projects. This chapter describes the usage of the assembler and the directives (assembler pseudo-ops and controls) accepted by the assembler in the source files. Although the term “assembler” is almost universally used to describe the tool that converts human-readable mnemonics into machine code, both “assembler” and “assembly” are used to describe the source code which such a tool reads. The latter is more common and is used in this manual to describe the language. Thus you will see the terms assembly language (or just assembly), assembly listing and other assembly terms, but also, assembler options, assembler directive and assembler optimizer. The following topics are examined in this chapter of the user’s guide: • • • • • 6.2 Assembler Usage Options MPLAB XC8 Assembly Language Assembly-Level Optimizations Assembly List Files ASSEMBLER USAGE The assembler is called ASPIC18 for PIC18 devices and ASPIC for all other 8-bit devices. It is available to run on Windows, Linux and Mac OS X systems. Note that the assembler will not produce any messages unless there are errors or warnings – there are no “assembly completed” messages. Typically the command-line driver, xc8, is used to invoke the assembler as it can be passed assembler source files as input; however, the options for the assembler are supplied here for instances where the assembler is being called directly, or when they are specified using the command-line driver option --SETOPTION, see Section 4.8.54 “--SETOPTION: Set the Command-line Options for Application”. The usage of the assembler is similar under all of available operating systems. All command-line options are recognized in either upper or lower case. The basic command format is shown: ASPIC [ options ] files files is a space-separated list of one or more assembler source files. Where more than one source file is specified, the assembler treats them as a single module; i.e., a single assembly will be performed on the concatenation of all the source files specified. The files must be specified in full, no default extensions or suffixes are assumed. options is an optional space-separated list of assembler options, each with a dash – as the first character in the case of single letter options, or double dashes in the case of multi-letter options. The assembler options must be specified on the command line before any files. 2012-2013 Microchip Technology Inc. DS50002053D-page 255 MPLAB® XC8 C Compiler User’s Guide 6.3 OPTIONS The command line options recognized by ASPIC are given in Table 6-1, and a full description of each option followings. TABLE 6-1: ASPIC COMMAND-LINE OPTIONS Option Meaning Default -A produce assembler output produce object code -Cchipinfo define the chipinfo file dat\picc[18].ini -E[file | digit] set error destination/format -Flength specify listing page length -H output HEX values for constants decimal values -I list macro expansions don’t list macros -L[listfile ] produce listing no listing -N disable merging optimizations merging optimizations enabled -O perform optimization no optimization -Ooutfile specify object name srcfile.obj -R specify non-standard ROM -Twidth specify listing page width 80 -V produce line number info no line numbers -VER=version specify full version information for list file title -Wlevel set warning level threshold -X no local symbols in OBJ file --CHIP=device specify device name --DISL=list specify disabled messages --EDF=path specify message file location --EMAX=number specify maximum number of errors --OPT=optimization specify optimization type --VER print version number and stop 6.3.1 66 0 no message disabled 10 -A: Generate Assembly File An assembler file will be produced if this option is used rather than the usual object file format. This is useful when checking the optimized assembler produced using the -O optimization option. By default the output file will an extension .opt, unless the -Ooutfile output option is used to specify another name. 6.3.2 -C: Specify Chip Info File Specify the chipinfo file to use. The chipinfo file is called picc.ini and can be found in the dat directory in the compiler’s installation directory. This file specifies information about the currently selected device. DS50002053D-page 256 2012-2013 Microchip Technology Inc. Macro Assembler 6.3.3 -E: Specify Error Format/File The default format for an error message is in the form: filename: line: message where the error of type message occurred on line line of the file filename. The -E option with no argument will make the assembler use an alternate format for error and warning messages. Specifying a filename as argument will force the assembler to direct error and warning messages to a file with the name specified. 6.3.4 -F: Specify Page Length By default, the assembly listing format is pageless; i.e., the assembler listing output is continuous. The output can be formatted into pages of varying lengths. Each page will begin with a header and title, if specified. The -F option allows a page length to be specified. A zero value of length implies pageless output. The length is specified in a number of lines. 6.3.5 -H: Print Hexadecimal Constant This option specifies that output constants should be shown as hexadecimal values rather than decimal values. This option affects both the assembly list file, as well as assembly output, when requested. 6.3.6 -I: List Macro Expansions This option forces listing of macro expansions and unassembled conditionals which would otherwise be suppressed by a NOLIST assembler control, see Section 6.4.10 “Assembler Controls”. The -L option is still necessary to produce an actual listing output. 6.3.7 -L: Generate an Assembly Listing This option requests the generation of an assembly listing file. If listfile is specified then the listing will be written to that file, otherwise it will be written to the standard output. This option is applied if compiling using xc8, the command-line driver and the --ASMLIST driver option, see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”. 6.3.8 -O: Optimize Assembly This requests the assembler to perform optimization on the assembly code. Note that the use of this option slows the assembly process down, as the assembler must make an additional pass over the input code. Debug information for assembler code generated from C source code can become unreliable in debuggers when this option is used. This option can be applied if compiling using xc8, the command-line driver and the --OPT driver option, see Section 4.8.42 “--OPT: Invoke Compiler Optimizations”. 6.3.9 -O: Specify Output File By default the assembler determines the name of the object file to be created by stripping any suffix or extension from the first source filename and appending .obj. The -O option allows the user to override the default filename and specify a new name for the object file. 2012-2013 Microchip Technology Inc. DS50002053D-page 257 MPLAB® XC8 C Compiler User’s Guide 6.3.10 -T: Specify Listing Page Width This option allows specification of the assembly list file page width, in characters. width should be a decimal number greater than 41. The default width is 80 characters. 6.3.11 -V: Produce Assembly Debug Information This option will include line number and filename information in the object file produced by the assembler. Such information can be used by debuggers. Note that the line numbers will correspond with assembler code lines in the assembler file. This option should not be used when assembling an assembly file produced by the code generator. In that case, debug information should relate back to the original C source, not the intermediate assembly code. 6.3.12 -VER: Specify Version Information This option allows the full version information, including optional text to indicate beta builds or release candidate builds, to be passed to the assembler. This information is only used in the title of the assembly list file and is not reflected in the output to the --VER option. 6.3.13 -X: Strip Local Symbols The object file created by the assembler contains symbol information, including local symbols; i.e., symbols that are neither public nor external. The -X option will prevent the local symbols from being included in the object file, thereby reducing the file size. 6.3.14 --CHIP: Specify Device Name This option defines the device that is being used. The device type can also be indicated by use of the device directive in the assembler source file, see Section 6.4.9.20 “PROCESSOR”. You can also add your own processors to the compiler via the compiler’s chipinfo file. This option is applied if compiling using xc8, the command-line driver and the --CHIP driver option, see Section 4.8.18 “--CHIP: Define Device”. 6.3.15 --DISL: Disable Messages This option is mainly used by the command-line driver, xc8, to disable particular message numbers. It takes a comma-separate list of message numbers that will be disabled during compilation. This option is applied if compiling using xc8, the command-line driver and the --MSGDISABLE driver option, see Section 4.8.38 “--MSGDISABLE: Disable Warning Messages”. See Section 4.6 “Compiler Messages” for full information about the compiler’s messaging system. 6.3.16 --EDF: Set Message File Path This option is mainly used by the command-line driver, xc8, to specify the path of the message description file. The default file is located in the dat directory in the compiler’s installation directory. See Section 4.6 “Compiler Messages”, for full information about the compiler’s messaging system. DS50002053D-page 258 2012-2013 Microchip Technology Inc. Macro Assembler 6.3.17 --EMAX: Specify Maximum Number of Errors This option is mainly used by the command-line driver, xc8, to specify the maximum number of errors that can be encountered before the assembler terminates. The default number is 10 errors. This option is applied if compiling using xc8, the command-line driver and the --ERRORS driver option, see Section 4.8.28 “--ERRORS: Maximum Number of Errors”. See Section 4.6 “Compiler Messages”, for full information about the compiler’s messaging system. 6.3.18 --OPT: Specify Optimization Type This option complements the assembler -O option and indicates specific information about optimizations required. The suboptions: speed, space and debug can be specified to indicate preferences related to procedural abstraction. Abstraction is enabled when the space option is set, and disabled when speed is set. The debug suboption limits the application of some optimizations which otherwise can severely corrupt debug information used by debuggers. 6.3.19 --VER: Print Version Number This option prints information relating to the version and build of the assembler application. The assembler will terminate after processing this option, even if other options and files are present on the command line. 2012-2013 Microchip Technology Inc. DS50002053D-page 259 MPLAB® XC8 C Compiler User’s Guide 6.4 MPLAB XC8 ASSEMBLY LANGUAGE The source language accepted by the macro assembler, ASPIC, is described below. All opcode mnemonics and operand syntax are specific to the target device. Although the PIC17 family instruction set is supported at the assembler level, the code generator cannot produce code for these devices so no C projects can target these devices. Additional mnemonics and assembler directives are documented in this section. The same assembler application is used for compiler-generated intermediate assembly and hand-written assembly source code. 6.4.1 Assembly Instruction Deviations The MPLAB XC8 assembler uses a slightly modified form of assembly language to that specified by the Microchip data sheets. The following information details changes to the instruction format, and pseudo instructions that can be used in addition to the device instruction set. These deviations can be used in assembly code in-line with C code or in hand-written assembly modules. 6.4.1.1 DESTINATION LOCATION Certain PIC instructions use the operands “,0” or “,1” to specify the destination for the result of that operation. ASPIC uses the more-readable operands “,w” and “,f” to specify the destination register. The W register is selected as the destination when using the “,w” operand, and the file register is selected when using the “,f” operand or if no destination operand is specified. The case of the letter in the destination operand in not important. For example: MOVF _foo,w ADDWF _foo,f ;the ,f destination is optional The numerical destination operands cannot be used with ASPIC. The assembler also uses the destination select operand “,b” to indicate that PIC18 instructions should use the bank select register (BSR) when accessing the specified file register address. The “,c” operand indicates that the address is in the common memory, which is known as the access bank on PIC18 devices. Alternatively, an operand can be preceded by the characters “c:” to indicate that the address resides in common memory. These operands and prefix affect the RAM access bit in the instruction. For example: ADDWF _foo,c BTFSC c:_foo,3 If the instruction does not specify a destination select operand, or the common memory prefix, the instruction will used banked access. These operands and prefix are not applicable with operands to the MOVFF instruction, which takes two untruncated addresses and which always works independently of the BSR. If an instruction address is absolute and the address is within the access bank, the common memory destination selectors “,c” or “c:” are not required. The assembler will determine from the address that access bank memory is being accessed. DS50002053D-page 260 2012-2013 Microchip Technology Inc. Macro Assembler For example, the following instructions show the W register being moved to first, an absolute location; and, then, to an address represented by an identifier. The op codes for these instructions, assuming that the address assigned to _foo is 0x516 and to _bar is 0x55, are shown below. 6EE5 6E55 0105 6F16 6F16 MOVWF 0FE5h MOVWF _bar,c BANKSEL(_foo) MOVWF BANKMASK(_foo),b MOVWF BANKMASK(_foo) ;write to access bank location 0xFE5 ;write to access bank location 0x55 ;set up BSR to access _foo ;write to _foo (banked) ;defaults to banked access Notice that the first two instructions have the RAM access bit (bit 8 of the op-code) cleared, but that the bit is set in the last two instructions. 6.4.1.2 BANK AND PAGE SELECTION The BANKSEL pseudo instruction can be used to generate instructions to select the bank of the operand specified. The operand should be the symbol or address of an object that resides in the data memory. Depending on the target device, the generated code will either contain one or more bit instructions to set/clear bits in the appropriate register, or use a MOVLB instruction (in the case of enhanced mid-range or PIC18 devices). As this pseudo instruction can expand to more than one instruction on mid-range or baseline parts, it should not immediately follow a BTFSX instruction on those devices. For example: MOVLW 20 BANKSEL(_foobar) ;select bank for next file instruction MOVWF BANKMASK(_foobar) ;write data and mask address In the same way, the PAGESEL pseudo instruction can be used to generate code to select the page of the address operand. For the current page, you can use the location counter, $, as the operand. Depending on the target device, the generated code will either contain one or more instructions to set/clear bits in the appropriate register, or use a MOVLP instruction in the case of enhanced mid-range PIC devices. As the directive could expand to more than one instruction, it should not immediately follow a BTFSX instruction. For example: FCALL _getInput PAGESEL $ ;select this page This directive is accepted when compiling for PIC18 targets but has no effect and does not generate any code. Support is purely to allow easy migration across the 8-bit devices. 2012-2013 Microchip Technology Inc. DS50002053D-page 261 MPLAB® XC8 C Compiler User’s Guide 6.4.1.3 ADDRESS MASKING A macro, BANKMASK(), can be used with an identifier; so, it is usable as an operand to instructions that expect a file register address. The macro does this by ANDing out the bank information using a suitable mask. It is available once you include the file. An example of this macro is given in Section 6.4.1.2 “Bank and Page Selection”. All MPLAB XC8 assembly identifiers represent a full address. This address includes the bank information for the object it represents. Virtually all instructions in the 8-bit PIC instruction sets that take a file register operand expect this operand value to be an offset into the currently selected bank. As the device families have different bank sizes, the width of this offset is different for each family. Use of this macro increases assembly code portability across Microchip devices, since it adjusts the mask to suit the bank size of the target device. Do not use this macro with either operand to the PIC18’s MOVFF instruction, which requires two full, banked addresses to be specified, or with any other instruction that expects a full address. 6.4.1.4 MOVFW PSEUDO INSTRUCTION The MOVFW pseudo instruction implemented by MPLAB C18 is not implemented in MPLAB XC8. You will need to use the standard PIC instruction that performs an identical function. Note that the MPLAB C18 instruction: MOVFW foobar maps directly to the standard PIC instruction: MOVF foobar,w 6.4.1.5 MOVIW/MOVWI INSTRUCTIONS Both the MOVIW and MOVWI instructions have operands which differ in syntax to that indicated in the data sheet. These instructions are only available with enhanced mid-range devices. The indexed Indirect operands to these instructions have the FSR offset specified first in square brackets, followed by the FSR name, for example: MOVIW [6]FSR0 MOVWI [0x10]FSR1 The pre/post increment/decrement form of these instructions use the name of the FSR register, not the indirection register (INDF), for example: MOVIW ++FSR0 MOVWI FSR1++ MOVWI FSR0-- 6.4.1.6 INTERRUPT RETURN MODE The RETFIE PIC18 instruction can be followed by “f” (no comma) to indicate that the shadow registers should be retrieved and copied to their corresponding registers on execution. Without this modifier, the registers are not updated from the shadow registers. This replaces the “0” and “1” operands indicated in the device data sheet. The following examples show both forms and the opcodes they generate. 0011 0010 RETFIE f RETFIE ;shadow registers copied ;return without copy The baseline and mid-range devices do not allow such a syntax. DS50002053D-page 262 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.1.7 LONG JUMPS AND CALLS The assembler recognizes several mnemonics that expand into regular PIC MCU assembly instructions. The mnemonics are FCALL and LJMP. On baseline and mid-range parts, these instructions expand into regular CALL and GOTO instructions respectively, but also ensure the instructions necessary to set the bits in PCLATH (for mid-range devices) or STATUS (for baseline devices) will be generated when the destination is in another page of program memory. Whether the page selection instructions are generated, and exactly where they will be located, is dependent on the surrounding source code. Page selection instructions can appear immediately before the call or jump, or be generated as part of, and immediately after, a previous FCALL/LJMP instruction. On PIC18 devices, these mnemonics are present purely for compatibility with smaller 8-bit devices and are always expanded as regular PIC18 CALL and GOTO instructions. These additional mnemonics should be used where possible as they make assembly code independent of the final position of the routines that are to be executed. If the call or jump is determined to be within the current page, the additional code to set the PCLATH bits can be optimized away. Note that assembly code that is added in-line with C code is never optimized and assembly modules require a specific option to enable optimization, see Section 4.8.42 “--OPT: Invoke Compiler Optimizations”. Unoptimized FCALL and LJMP instruction will always generate page selection code. The following mid-range PIC example shows an FCALL instruction in the assembly list file. You can see that the FCALL instruction has expanded to five instructions. In this example, there are two bit instructions that set/clear bits in the PCLATH register. Bits are also set/cleared in this register after the call to reselect the page that was selected before the FCALL. 13 14 0079 007A 15 007F 3021 120A 120A 3400 158A 118A 2000 movlw fcall 33 _phantom retlw 0 Since FCALL and LJMP instructions can expand into more than one instruction, they should never be preceded by an instruction that can skip, e.g., a BTFSC instruction. The FCALL and LJMP instructions assume that the psect that contains them is smaller than a page. Do not use these instructions to transfer control to a label in the current psect if it is larger than a page. The default linker options will not permit code psects to be larger than a page. On PIC18 devices, the regular CALL instruction can be followed by a “,f” to indicate that the W, STATUS and BSR registers should be pushed to their respective shadow registers. This replaces the “,1” syntax indicated on the device data sheet. 2012-2013 Microchip Technology Inc. DS50002053D-page 263 MPLAB® XC8 C Compiler User’s Guide 6.4.1.8 RELATIVE BRANCHES The PIC18 devices implement conditional relative branch instructions, e.g., BZ, BNZ. These instructions have a limited jump range compared to the GOTO instruction. Note that in some instance, the assembler can change a relative branch instruction to be a relative branch with the opposite condition over a GOTO instruction. For example: BZ error ;next can become: BNZ l18 GOTO error l18: ;next This is functionally identical and is performed so that the conditional branch can use the same destination range as the GOTO instruction. 6.4.2 Statement Formats Legal statement formats are shown in Table 6-2: "ASPIC Statement Formats". The label field is optional and, if present, should contain one identifier. A label can appear on a line of its own, or precede a mnemonic as shown in the second format. The third format is only legal with certain assembler directives, such as MACRO, SET and EQU. The name field is mandatory and should contain one identifier. If the assembly file is first processed by the C preprocessor, see Section 4.8.10 “-P: Preprocess Assembly Files”, then it can also contain lines that form valid preprocessor directives. See Section 5.14.1 “C Language Comments”, for more information on the format for these directives. There is no limitation on what column or part of the line in which any part of the statement should appear. TABLE 6-2: ASPIC STATEMENT FORMATS Format # Field1 Field2 Field3 Field4 Format 1 label: Format 2 label: mnemonic operands ; comment Format 3 name pseudo-op operands ; comment Format 4 ; comment only Format 5 empty line 6.4.3 Characters The character set used is standard 7 bit ASCII. Alphabetic case is significant for identifiers, but not mnemonics and reserved words. Tabs are treated as equivalent to spaces. 6.4.3.1 DELIMITERS All numbers and identifiers must be delimited by white space, non-alphanumeric characters or the end of a line. DS50002053D-page 264 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.3.2 SPECIAL CHARACTERS There are a few characters that are special in certain contexts. Within a macro body, the character & is used for token concatenation. To use the bitwise & operator within a macro body, escape it by using && instead. In a macro argument list, the angle brackets < and > are used to quote macro arguments. 6.4.4 Comments An assembly comment is initiated with a semicolon that is not part of a string or character constant. If the assembly file is first processed by the C preprocessor, see Section 4.8.10 “-P: Preprocess Assembly Files”, then the file can also contain C or C++ style comments using the standard /* ... */ and // syntax. 6.4.4.1 SPECIAL COMMENT STRINGS Several comment strings are appended to compiler-generated assembly instructions by the code generator. These comments are typically used by the assembler optimizer. The comment string ;volatile is used to indicate that the memory location being accessed in the instruction is associated with a variable that was declared as volatile in the C source code. Accesses to this location which appear to be redundant will not be removed by the assembler optimizer if this string is present. This comment string can also be used in hand-written assembly source to achieve the same effect for locations defined and accessed in assembly code. The comment string ;wreg free is placed on some CALL instructions. The string indicates that the W register was not loaded with a function parameter; i.e., it is not in use. If this string is present, optimizations can be made to assembler instructions before the function call, which loads the W register redundantly. 2012-2013 Microchip Technology Inc. DS50002053D-page 265 MPLAB® XC8 C Compiler User’s Guide 6.4.5 Constants 6.4.5.1 NUMERIC CONSTANTS The assembler performs all arithmetic with signed 32-bit precision. The default radix for all numbers is 10. Other radices can be specified by a trailing base specifier, as given in Table 6-3. TABLE 6-3: ASPIC NUMBERS AND BASES Radix Binary Format Digits 0 and 1 followed by B Octal Digits 0 to 7 followed by O, Q, o or q Decimal Digits 0 to 9 followed by D, d or nothing Hexadecimal Digits 0 to 9, A to F preceded by 0x or followed by H or h Hexadecimal numbers must have a leading digit (e.g., 0ffffh) to differentiate them from identifiers. Hexadecimal digits are accepted in either upper or lower case. Note that a binary constant must have an upper case B following it, as a lower case b is used for temporary (numeric) label backward references. In expressions, real numbers are accepted in the usual format, and are interpreted as IEEE 32-bit format. 6.4.5.2 CHARACTER CONSTANTS AND STRINGS A character constant is a single character enclosed in single quotes ’. Multi-character constants, or strings, are a sequence of characters, not including carriage return or newline characters, enclosed within matching quotes. Either single quotes ’ or double quotes " can be used, but the opening and closing quotes must be the same. 6.4.6 Identifiers Assembly identifiers are user-defined symbols representing memory locations or numbers. A symbol can contain any number of characters drawn from the alphabetics, numerics, and the special characters: dollar, $; question mark, ?; and underscore, _. The first character of an identifier cannot be numeric. The case of alphabetics is significant, e.g., Fred is not the same symbol as fred. Some examples of identifiers are shown here: An_identifier an_identifier an_identifier1 $ ?$_12345 An identifier cannot be one of the assembler directives, keywords, or psect flags. An identifier that begins with at least one underscore character can be accessed from C code. Care must be taken with such symbols that they do not interact with C code identifiers. Identifiers that do not begin with an underscore can only be accessed from the assembly domain. See Section 5.12.3.1 “Equivalent Assembly Symbols” for the mapping between the C and assembly domains. DS50002053D-page 266 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.6.1 SIGNIFICANCE OF IDENTIFIERS Users of other assemblers that attempt to implement forms of data typing for identifiers should note that this assembler attaches no significance to any symbol, and places no restrictions or expectations on the usage of a symbol. The names of psects (program sections) and ordinary symbols occupy separate, overlapping name spaces, but other than this, the assembler does not care whether a symbol is used to represent bytes, words or sports cars. No special syntax is needed or provided to define the addresses of bits or any other data type, nor will the assembler issue any warnings if a symbol is used in more than one context. The instruction and addressing mode syntax provide all the information necessary for the assembler to generate correct code. 6.4.6.2 ASSEMBLER-GENERATED IDENTIFIERS Where a LOCAL directive is used in a macro block, the assembler will generate a unique symbol to replace each specified identifier in each expansion of that macro. These unique symbols will have the form ??nnnn where nnnn is a 4-digit number. The user should avoid defining symbols with the same form. 6.4.6.3 LOCATION COUNTER The current location within the active program section is accessible via the symbol $. This symbol expands to the address of the currently executing instruction (which is different to the address contained in the program counter (PC) register when executing this instruction). Thus: GOTO $ ;endless loop will represent code that will jump to itself and form an endless loop. By using this symbol and an offset, a relative jump destination can be specified. Any address offset added to $ has the native addressability of the target device. So, for baseline and mid-range devices, the offset is the number of instructions away from the current location, as these devices have word-addressable program memory. For PIC18 instructions, which use byte addressable program memory, the offset to this symbol represents the number of bytes from the current location. As PIC18 instructions must be word aligned, the offset to the location counter should be a multiple of 2. All offsets are rounded down to the nearest multiple of 2. For example: GOTO MOVLW MOVWF $+2 8 _foo ;skip... ;to here for PIC18 devices, or ;to here for baseline and mid-range devices will skip the MOVLW instruction on baseline or mid-range devices. On PIC18 devices, GOTO $+2 will jump to the following instruction; i.e., act like a NOP instruction. 6.4.6.4 REGISTER SYMBOLS Code in assembly modules can gain access to the special function registers by including pre-defined assembly header files. The appropriate file can be included by add the line: #include to the assembler source file. Note that the file must be included using a C pre-processor directive and hence the option to preprocess assembly files must be enabled when compiling, see Section 4.8.10 “-P: Preprocess Assembly Files”. This header file contains appropriate commands to ensure that the header file specific for the target device is included into the source file. These header files contain EQU declarations for all byte or multi-byte sized registers and #define macros for named bits within byte registers. 2012-2013 Microchip Technology Inc. DS50002053D-page 267 MPLAB® XC8 C Compiler User’s Guide 6.4.6.5 SYMBOLIC LABELS A label is a symbolic alias that is assigned a value that is equal to the current address within the current psect. Labels are not assigned a value until link time. A label definition consists of any valid assembly identifier followed by a colon, :. The definition can appear on a line by itself or it can be positioned before a statement. Here are two examples of legitimate labels interspersed with assembly code. frank: MOVLW 1 GOTO fin simon44: CLRF _input Here, the label frank will ultimately be assigned the address of the MOVLW instruction, and simon44 the address of the CLRF instruction. Regardless of how they are defined, the assembler list file produced by the assembler will always show labels on a line by themselves. Note that the colon following the label is mandatory for PIC18 assembly, but is recommended in assembly for all other devices. Symbols that are not interpreted as instructions are assumed to be labels. Mistyped assembly instructions can sometimes be treated as labels without an error message being issued. Thus the code: mistake: MOVLW 23h MOVWF 37h REUTRN ; oops defines a symbol called REUTRN, which was intended to be the RETURN instruction. This cannot occur with PIC18 assembly code, as the colon following a label is mandatory; the compiler would report an error when reached the line containing REUTRN. Labels can be used (and are preferred) in assembly code, rather than using an absolute address with other instructions. In this way, they can be used as the target location for jump-type instructions or to load an address into a register. Like variables, labels have scope. By default, they can be used anywhere in the module in which they are defined. They can be used by code located before their definition. To make a label accessible in other modules, use the GLOBAL directive. See Section 6.4.9.1 “GLOBAL” for more information. DS50002053D-page 268 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.7 Expressions The operands to instructions and directives are comprised of expressions. Expressions can be made up of numbers, identifiers, strings and operators. Operators can be unary (one operand, e.g., not) or binary (two operands, e.g., +). The operators allowable in expressions are listed in Table 6-4. TABLE 6-4: ASPIC OPERATORS Operator Purpose Example * multiplication MOVLW 4*33,w + addition BRA $+1 - subtraction DB 5-2 / division MOVLW 100/4 = or eq equality IF inp eq 66 > or gt signed greater than IF inp > 40 >= or ge signed greater than or equal to IF inp ge 66 < or lt signed less than IF inp < 40 <= or le signed less than or equal to IF inp le 66 <> or ne signed not equal to IF inp <> 40 low low byte of operand MOVLW low(inp) high high byte of operand MOVLW high(1008h) highword high 16 bits of operand DW highword(inp) mod modulus MOVLW 77mod4 & or and bitwise AND CLRF inp&0ffh ^ bitwise XOR (exclusive or) MOVF inp^80,w | bitwise OR MOVF inp|1,w not bitwise complement MOVLW not 055h,w << or shl shift left DB inp>>8 >> or shr shift right MOVLW inp shr 2,w rol rotate left DB inp rol 1 ror rotate right DB inp ror 1 float24 24-bit version of real operand DW float24(3.3) nul tests if macro argument is null The usual rules governing the syntax of expressions apply. The operators listed can all be freely combined in both constant and relocatable expressions. The linker permits relocation of complex expressions, so the results of expressions involving relocatable identifiers cannot be resolved until link time. 2012-2013 Microchip Technology Inc. DS50002053D-page 269 MPLAB® XC8 C Compiler User’s Guide 6.4.8 Program Sections Program sections, or psects, are simply a section of code or data. They are a way of grouping together parts of a program (via the psect’s name) even though the source code cannot be physically adjacent in the source file, or even where spread over several modules. For an introductory guide to psects, see Section 5.15.1 “Program Sections”. A psect is identified by a name and has several attributes. The PSECT assembler directive is used to define a psect. It takes as arguments a name and an optional comma-separated list of flags. See Section 5.15.2 “Compiler-Generated Psects” for a list of all psects that the code generator defines. Chapter 7. “Linker” has more information on the operation of the linker and on options that can be used to control psect placement in memory. The assembler associates no significance to the name of a psect. The linker, also, is not aware of which psects are compiler-generated or which are user-defined. Unless defined as abs (absolute), psects are relocatable. Code or data that is not explicitly placed into a psect will become part of the default (unnamed) psect. When writing assembly code, you can use the existing compiler-generated psects, described in Section 5.15.2 “Compiler-Generated Psects”, or create your own. You will not need to adjust the linker options if you are using compiler-generated psects. If you create your own psects, try to associate them with an existing linker class (see Section 5.15.3 “Default Linker Classes” and Section 6.4.9.3.3 “Class”) otherwise you can need to specify linker options for them to be allocated correctly. Note, that the length and placement of psects is important. It is easier to write code if all executable code is located in psects that do not cross any device pages boundaries; so, too, if data psects do not cross bank boundaries. The location of psects (where they are linked) must match the assembly code that accesses the psect contents. DS50002053D-page 270 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.9 Assembler Directives Assembler directives, or pseudo-ops, are used in a similar way to instruction mnemonics. With the exception of PAGESEL and BANKSEL, these directives do not generate instructions. The DB, DW and DDW directives place data bytes into the current psect. The directives are listed in Table 6-5, and are detailed below in the following sections. TABLE 6-5: ASPIC ASSEMBLER DIRECTIVES Directive Purpose GLOBAL make symbols accessible to other modules or allow reference to other modules’ symbols END end assembly PSECT declare or resume program section ORG set location counter within current psect EQU define symbol value SET define or re-define symbol value DB define constant byte(s) DW define constant word(s) DDW define double-width constant word(s) (pic18 devices only) DS reserve storage DABS define absolute storage IF conditional assembly ELSIF alternate conditional assembly ELSE alternate conditional assembly ENDIF end conditional assembly FNCALL inform the linker that one function calls another FNROOT inform the linker that a function is the “root” of a call graph MACRO macro definition ENDM end macro definition LOCAL define local tabs ALIGN align output to the specified boundary BANKSEL generate code to select bank of operand PAGESEL generate set/clear instruction to set PCLATH bits for this page PROCESSOR define the particular chip for which this file is to be assembled. REPT repeat a block of code n times IRP repeat a block of code with a list IRPC repeat a block of code with a character list SIGNAT define function signature 2012-2013 Microchip Technology Inc. DS50002053D-page 271 MPLAB® XC8 C Compiler User’s Guide 6.4.9.1 GLOBAL The GLOBAL directive declares a list of comma-separated symbols. If the symbols are defined within the current module, they are made public. If the symbols are not defined in the current module, they are made references to public symbols defined in external modules. Thus to use the same symbol in two modules the GLOBAL directive must be used at least twice: once in the module that defines the symbol to make that symbol public, and again in the module that uses the symbol to link in with the external definition. For example: GLOBAL 6.4.9.2 lab1,lab2,lab3 END The END directive is optional, but if present should be at the very end of the program. It will terminate the assembly and not even blank lines should follow this directive. If an expression is supplied as an argument, that expression will be used to define the entry point of the program. This is stored in a start record in the object file produced by the assembler. Whether this is of any use will depend on the linker. The default runtime startup code that is defined by the compiler will contain an END directive with a start address. As only one start address can be specified for each project, you generally do not need to define this address – you can use the END directive with no entry point in any file. For example: END start_label ;defines the entry point or END DS50002053D-page 272 ;do not define entry point 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.9.3 PSECT The PSECT directive declares or resumes a program section. For an introductory guide to psects, see Section 5.15.1 “Program Sections”. The directive takes as argument a name and, optionally, a comma-separated list of flags. The allowed flags specify attributes of the psect. They are listed in Table 6-6. The psect name is in a separate name space to ordinary assembly symbols, so a psect can use the same identifier as an ordinary assembly identifier. However, a psect name cannot be one of the assembler directives, keywords, or psect flags. Once a psect has been declared, it can be resumed later by another PSECT directive; however, the flags need not be repeated and will be propagated from the earlier declaration. If two PSECT directives are encountered with contradicting flags, then an error is generated. TABLE 6-6: PSECT FLAGS Flag Meaning abs psect is absolute bit psect holds bit objects class=name specify class name for psect delta=size size of an addressing unit global psect is global (default) inline psect contents (function) can be inlined when called keep psect will not be deleted after inlining limit=address upper address limit of psect local psect is not global merge=allow allow or prevent merging of this psect ovrld psect will overlap same psect in other modules pure psect is to be read-only reloc=boundary start psect on specified boundary size=max maximum size of psect space=area represents area in which psect will reside split=allow allow or prevent splitting of this psect with=psect place psect in the same page as specified psect Some examples of the use of the PSECT directive follow: PSECT fred PSECT bill,size=100h,global PSECT joh,abs,ovrld,class=CODE,delta=2 6.4.9.3.1 Abs The abs flag defines the current psect as being absolute; i.e., it is to start at location 0. This does not mean that this module’s contribution to the psect will start at 0, since other modules can contribute to the same psect. See also Section 6.4.9.3.11 “Ovrld”. An abs-flagged psect is not relocatable and an error will result if a linker option is issued that attempts to place such a psect at any location. 6.4.9.3.2 Bit The bit flag specifies that a psect holds objects that are 1 bit long. Such psects will have a scale value of 8 to indicate that there are 8 addressable units to each byte of storage and all addresses associated with this psect will be bit address, not byte addresses. The scale value is indicated in the map file; see Section 7.4 “Map Files”. 2012-2013 Microchip Technology Inc. DS50002053D-page 273 MPLAB® XC8 C Compiler User’s Guide 6.4.9.3.3 Class The class flag specifies a corresponding linker class name for this psect. A class is a range of addresses in which psects can be placed. Class names are used to allow local psects to be located at link time, since they cannot always be referred to by their own name in a -P linker option (as would be the case if there are more than one local psect with the same name). Class names are also useful where psects need only be positioned anywhere within a range of addresses rather than at a specific address. The association of a class with a psect that you have defined typically means that you do not need to supply a custom linker option to place it in memory. See Section 7.2.1 “-Aclass =low-high,...” for information on how linker classes are defined. 6.4.9.3.4 Delta The delta flag defines the size of the addressable unit. In other words, the number of data bytes that are associated with each address. With PIC mid-range and baseline devices, the program memory space is word addressable; so, psects in this space must use a delta of 2. That is to say, each address in program memory requires 2 bytes of data in the HEX file to define their contents. So, addresses in the HEX file will not match addresses in the program memory. The data memory space on these devices is byte addressable; so, psects in this space must use a delta of 1. This is the default delta value. All memory spaces on PIC18 devices are byte addressable; so, a delta of 1 (the default) should be used for all psects on these devices. The redefinition of a psect with conflicting delta values can lead to phase errors being issued by the assembler. 6.4.9.3.5 Global A psect defined as global will be combined with other global psects with the same name at link time. Psects are grouped from all modules being linked. Psects are considered global by default, unless the local flag is used. 6.4.9.3.6 Inline The inline flag is used by the code generator to tell the assembler that the contents of a psect can be inlined. If this operation is performed, the contents of the inline psect will be copied and used to replace calls to the function defined in the psect. 6.4.9.3.7 Keep Psects that are candidates for inlining (see Section 6.4.9.3.6 “Inline”) can be deleted after the inlining takes place. This flag ensures that the psect is not deleted after any inlining by the assembler optimizer. 6.4.9.3.8 Limit The limit flag specifies a limit on the highest address to which a psect can extend. If this limit is exceeded when it is positioned in memory, an error will be generated. 6.4.9.3.9 Local A psect defined as local will not be combined with other local psects from other modules at link time, even if there are others with the same name. Where there are two local psects in the one module, they reference the same psect. A local psect cannot have the same name as any global psect, even one in another module. DS50002053D-page 274 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.9.3.10 Merge This flag can be assigned 0 or 1, or not specified. When assigned 0, the psect will never be merged by the assembly optimizer during optimizations. If assigned the value 1, the psect can be merged if other psect attributes allow it and the optimizer can see an advantage doing so. If this flag is not specified, then merging will not take place. Typically, merging is only performed on code-based psects (text psects). 6.4.9.3.11 Ovrld A psect defined as ovrld will have the contribution from each module overlaid, rather than concatenated at link time. This flag in combination with the abs flag (see Section 6.4.9.3.1 “Abs”) defines a truly absolute psect; i.e., a psect within which any symbols defined are absolute. 6.4.9.3.12 Pure The pure flag instructs the linker that this psect will not be modified at runtime. So, for example, be placed in ROM. This flag is of limited usefulness since it depends on the linker and target system enforcing it. 6.4.9.3.13 Reloc The reloc flag allows specification of a requirement for alignment of the psect on a particular boundary. For example the flag reloc=100h would specify that this psect must start on an address that is a multiple of 100h. PIC18 instructions must be word aligned, so a reloc value of 2 must be used for any psect that contains executable code. All other sections, and all sections for all other devices, can use the default reloc value of 1. 6.4.9.3.14 Size The size flag allows a maximum size to be specified for the psect, e.g., size=100h. This will be checked by the linker after psects have been combined from all modules. 6.4.9.3.15 Space The space flag is used to differentiate areas of memory that have overlapping addresses, but are distinct. Psects that are positioned in program memory and data memory have a different space value to indicate that the program space address 0, for example, is a different location to the data memory address 0. The memory spaces associated with the space flag numbers are shown in Table 6-7. TABLE 6-7: SPACE FLAG NUMBERS Space Flag Number Memory Space 0 Program memory 1 Data memory 2 Reserved 3 EEPROM Devices that have a banked data space do not use different space values to identify each bank. A full address that includes the bank number is used for objects in this space. So, each location can be uniquely identified. For example, a device with a bank size of 0x80 bytes will use address 0 to 0x7F to represent objects in bank 0, and then addresses 0x80 to 0xFF to represent objects in bank 1, etc. 2012-2013 Microchip Technology Inc. DS50002053D-page 275 MPLAB® XC8 C Compiler User’s Guide 6.4.9.3.16 Split This flag can be assigned 0 or 1, or not specified. When assigned 0, the psect will never be split by the assembly optimizer during optimizations. If assigned the value 1, the psect can be split if other psect attributes allow it and the psect is too large to fit in available memory. If this flag is not specified, then the splitability of this psect is based on whether the psect can be merged, see Section 6.4.9.3.10 “Merge”. 6.4.9.3.17 With The with flag allows a psect to be placed in the same page with another psect. For example the flag with=text will specify that this psect should be placed in the same page as the text psect. The term withtotal refers to the sum of the size of each psect that is placed “with” other psects. 6.4.9.4 ORG The ORG directive changes the value of the location counter within the current psect. This means that the addresses set with ORG are relative to the base address of the psect, which is not determined until link time. Note: The much-abused ORG directive does not move the location counter to the absolute address you specify. Only if the psect in which this directive is placed is absolute and overlaid will the location counter be moved to the address specified. To place objects at a particular address, place them in a psect of their own and link this at the required address using the linkers -P option, see Section 7.2.19 “-Pspec”. The ORG directive is not commonly required in programs. The argument to ORG must be either an absolute value, or a value referencing the current psect. In either case, the current location counter is set to the value determined by the argument. It is not possible to move the location counter backward. For example: ORG 100h will move the location counter to the beginning of the current psect plus 100h. The actual location will not be known until link time. In order to use the ORG directive to set the location counter to an absolute value, the directive must be used from within an absolute, overlaid psect. For example: PSECT absdata,abs,ovrld ORG 50h ;this is guaranteed to reside at address 50h 6.4.9.5 EQU This pseudo-op defines a symbol and equates its value to an expression. For example thomas EQU 123h The identifier thomas will be given the value 123h. EQU is legal only when the symbol has not previously been defined. See also, Section 6.4.9.6 “SET”, which allows for redefinition of values. This directive performs a similar function to the preprocessor’s #define directive, see Section 5.14.1 “C Language Comments”. DS50002053D-page 276 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.9.6 SET This pseudo-op is equivalent to EQU (Section 6.4.9.5 “EQU”) except that allows a symbol to be re-defined without error. For example: thomas SET 0h This directive performs a similar function to the preprocessor’s #define directive, see Section 5.14.1 “C Language Comments”. 6.4.9.7 DB The DB directive is used to initialize storage as bytes. The argument is a comma-separated list of expressions, each of which will be assembled into one byte and assembled into consecutive memory locations. Examples: alabel: DB ’X’,1,2,3,4, If the size of an address unit in the program memory is 2 bytes, as it will be for baseline and mid-range devices (see Section 6.4.9.3.4 “Delta”), the DB pseudo-op will initialize a word with the upper byte set to zero. So, the above example will define bytes padded to the following words. 0058 0001 0002 0003 0004 However, on PIC18 devices (PSECT directive’s delta flag should be 1), no padding will occur and the following data will appear in the HEX file. 58 01 02 03 04 6.4.9.8 DW The DW directive operates in a similar fashion to DB, except that it assembles expressions into 16-bit words. Example: DW -1, 3664h, ’A’ 6.4.9.9 DDW The DDW directive operates in a similar fashion to DW, except that it assembles expressions into double-width (32-bit) words. Example: DDW ’d’, 12345678h, 0 6.4.9.10 DS This directive reserves, but does not initialize, memory locations. The single argument is the number of bytes to be reserved. This directive is typically used to reserve memory location for RAM-based objects in the data memory. If used in a psect linked into the program memory, it will move the location counter, but not place anything in the HEX file output. Note that because the size of an address unit in the program memory is 2 bytes (see Section 6.4.9.3.4 “Delta”), the DS pseudo-op will actually reserve an entire word. A variable is typically defined by using a label and then the DS directive to reserve locations at the label location. Examples: alabel: DS 23 xlabel: DS 2+3 2012-2013 Microchip Technology Inc. ;Reserve 23 bytes of memory ;Reserve 5 bytes of memory DS50002053D-page 277 MPLAB® XC8 C Compiler User’s Guide 6.4.9.11 DABS This directive allows one or more bytes of memory to be reserved at the specified address. The general form of the directive is: DABS memorySpace, address, bytes[ ,symbol] where memorySpace is a number representing the memory space in which the reservation will take place, address is the address at which the reservation will take place, and bytes is the number of bytes that is to be reserved. The symbol is optional and refers to the name of the object at the address. Use of symbol in the directive will aid debugging. The symbol is automatically made globally accessible and is equated to the address specified in the directive. So, for example, the following directive uses a symbol: DABS 1,0x100,4,foo that is identical to the following directives: GLOBAL foo foo EQU 0x100 DABS 1,0x100,4 This directive differs to the DS directive in that it can be used to reserve memory at any location, not just within the current psect. Indeed, these directives can be placed anywhere in the assembly code and do not contribute to the currently selected psect in any way. The memory space number is the same as the number specified with the space flag option to psects (see Section 6.4.9.3.15 “Space”). The code generator issues a DABS directive for every user-defined absolute C variable, or for any variables that have been allocated an address by the code generator. The linker reads this DABS-related information from object files and ensures that the reserved addresses are not used for other memory placement. 6.4.9.12 IF, ELSIF, ELSE AND ENDIF These directives implement conditional assembly. The argument to IF and ELSIF should be an absolute expression. If it is non-zero, then the code following it up to the next matching ELSE, ELSIF or ENDIF will be assembled. If the expression is zero, then the code up to the next matching ELSE or ENDIF will not be output. At an ELSE, the sense of the conditional compilation will be inverted, while an ENDIF will terminate the conditional assembly block. These directives do not implement a runtime conditional statement in the same way that the C statement if() does; they are only evaluated at compile time. In addition, assembly code in both true and false cases is always scanned and interpreted, but the machine code corresponding to instructions is output only if the condition matches. This implies that assembler directives (e.g., EQU) will be processed regardless of the state of the condition expression, and so, should not be used inside an IF construct. DS50002053D-page 278 2012-2013 Microchip Technology Inc. Macro Assembler For example: IF ABC GOTO aardvark ELSIF DEF GOTO denver ELSE GOTO grapes ENDIF In this example, if ABC is non-zero, the first GOTO instruction will be assembled but not the second or third. If ABC is zero and DEF is non-zero, the second GOTO instruction will be assembled but the first and third will not. If both ABC and DEF are zero, the third GOTO instruction will be assembled. Note in the above example, only one GOTO instruction will appear in the output; which one will be determined by the values assigned to ABC and DEF. Conditional assembly blocks can be nested. 6.4.9.13 MACRO AND ENDM These directives provide for the definition of assembly macros, optionally with arguments. See Section 6.4.9.5 “EQU” for simple association of a value with an identifier, or Section 5.14.1 “C Language Comments” for the preprocessor’s #define macro directive, which can also work with arguments. The MACRO directive should be preceded by the macro name and optionally followed by a comma-separated list of formal arguments. When the macro is used, the macro name should be used in the same manner as a machine opcode, followed by a list of arguments to be substituted for the formal parameters. For example: ;macro: ;args: ; ;descr: movlf arg1 - the literal value to load arg2 - the NAME of the source variable Move a literal value into a nominated file register movlf MACRO arg1,arg2 MOVLW arg1 MOVWF arg2 mod 080h ENDM When used, this macro will expand to the 2 instructions in the body of the macro, with the formal parameters substituted by the arguments. Thus: movlf 2,tempvar expands to: MOVLW 2 MOVWF tempvar mod 080h The & character can be used to permit the concatenation of macro arguments with other text, but is removed in the actual expansion. For example: loadPort MACRO port, value MOVLW value MOVWF PORT&port ENDM will load PORTA if port is A when called, etc. The special meaning of the & token in macros implies that you can not use the bitwise AND operator, (also represented by &), in assembly macros; use the and form of this operator instead. A comment can be suppressed within the expansion of a macro (thus saving space in the macro storage) by opening the comment with a double semicolon, ;;. 2012-2013 Microchip Technology Inc. DS50002053D-page 279 MPLAB® XC8 C Compiler User’s Guide When invoking a macro, the argument list must be comma-separated. If it is desired to include a comma (or other delimiter such as a space) in an argument then angle brackets < and > can be used to quote If an argument is preceded by a percent sign, %, that argument will be evaluated as an expression and passed as a decimal number, rather than as a string. This is useful if evaluation of the argument inside the macro body would yield a different result. The nul operator can be used within a macro to test a macro argument, for example: IF nul ... ELSE ... ENDIF arg3 ; argument was not supplied. ; argument was supplied See Section 6.4.9.14 “LOCAL” for use of unique local labels within macros. By default, the assembly list file will show macro in an unexpanded format; i.e., as the macro was invoked. Expansion of the macro in the listing file can be shown by using the EXPAND assembler control; see Section 6.4.10.3 “EXPAND”. 6.4.9.14 LOCAL The LOCAL directive allows unique labels to be defined for each expansion of a given macro. Any symbols listed after the LOCAL directive will have a unique assembler generated symbol substituted for them when the macro is expanded. For example: down MACRO count LOCAL more more: DECFSZ count GOTO more ENDM when expanded, will include a unique assembler generated label in place of more. For example: down foobar expands to: ??0001 DECFSZ foobar GOTO ??0001 If invoked a second time, the label more would expand to ??0002, and multiply defined symbol errors will be averted. 6.4.9.15 ALIGN The ALIGN directive aligns whatever is following, data storage or code etc., to the specified offset boundary within the current psect. The boundary is specified as a number of bytes following the directive. For example, to align output to a 2-byte (even) address within a psect, the following could be used. ALIGN 2 Note that what follows will only begin on an even absolute address if the psect begins on an even address; i.e., alignment is done within the current psect. See Section 6.4.9.3.13 “Reloc” for psect alignment. The ALIGN directive can also be used to ensure that a psect’s length is a multiple of a certain number. For example, if the above ALIGN directive was placed at the end of a psect, the psect would have a length that was always an even number of bytes long. DS50002053D-page 280 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.9.16 REPT The REPT directive temporarily defines an unnamed macro, then expands it a number of times as determined by its argument. For example: REPT 3 ADDWF fred,w ENDM will expand to: ADDWF fred,w ADDWF fred,w ADDWF fred,w See also, Section 6.4.9.17 “IRP and IRPC”. 6.4.9.17 IRP AND IRPC The IRP and IRPC directives operate in a similar way to REPT; however, instead of repeating the block a fixed number of times, it is repeated once for each member of an argument list. In the case of IRP, the list is a conventional macro argument list. In the case or IRPC, it is each character in one argument. For each repetition, the argument is substituted for one formal parameter. For example: IRP number,4865h,6C6Ch,6F00h DW number ENDM would expand to: DW 4865h DW 6C6Ch DW 6F00h Note that you can use local labels and angle brackets in the same manner as with conventional macros. The IRPC directive is similar, except it substitutes one character at a time from a string of non-space characters. For example: IRPC char,ABC DB ’char’ ENDM will expand to: DB ’A’ DB ’B’ DB ’C’ 6.4.9.18 BANKSEL This directive can be used to generate code to select the bank of the operand. The operand should be the symbol or address of an object that resides in the data memory. See Section 6.4.1.2 “Bank and Page Selection”. 6.4.9.19 PAGESEL This directive can be used to generate code to select the page of the address operand. See Section 6.4.1.2 “Bank and Page Selection”. 2012-2013 Microchip Technology Inc. DS50002053D-page 281 MPLAB® XC8 C Compiler User’s Guide 6.4.9.20 PROCESSOR The output of the assembler can vary, depending on the target device. The device name is typically set using the --CHIP option to the command-line driver xc8, see Section 4.8.18 “--CHIP: Define Device”, or using the assembler’s --CHIP option, see Section 6.3.14 “--CHIP: Specify Device Name”. However, it can also be set with this directive, for example: PROCESSOR 16F877 This directive will override any device selected by any command-line option. 6.4.9.21 SIGNAT This directive is used to associate a 16-bit signature value with a label. At link time, the linker checks that all signatures defined for a particular label are the same. The linker will produce an error if they are not. The SIGNAT directive is used by HI-TECH C to enforce link time checking of C function prototypes and calling conventions. Use the SIGNAT directive if you want to write assembly language routines that are called from C. For example: SIGNAT _fred,8192 associates the signature value 8192 with the symbol _fred. If a different signature value for _fred is present in any object file, the linker will report an error. The easiest way to determine the correct signature value for a routine is to write a C routine with the same prototype as the assembly routine and check the signature value determined by the code generator. This will be shown in the assembly list file; see Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers”, and Section 6.5 “Assembly-Level Optimizations”. DS50002053D-page 282 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.10 Assembler Controls Assembler controls can be included in the assembler source to control assembler operation. These keywords have no significance anywhere else in the program. The control is invoked by the directive OPT, followed by the control name. Some keywords are followed by one or more arguments. For example: OPT EXPAND A list of keywords is given in Table 6-8, and each is described further below. TABLE 6-8: ASPIC ASSEMBLER CONTROLS Control Meaning ASMOPT_ON, start and stop assembly optimizations ASMOPT_OFF Format OPT ASMOPT_OFF ;protected code OPT ASMOPT_ON COND*, NOCOND include/do not include conditional code OPT COND in the listing EXPAND, NOEXPAND expand/do not expand macros in the listing output OPT EXPAND INCLUDE textually include another source file OPT INCLUDE < pathname > LIST*, NOLIST define options for listing output/disable OPT LIST [< listopt >, ..., listing output < listopt >] PAGE start a new page in the listing output OPT PAGE SPACE add blank lines to listing OPT SPACE 3 SUBTITLE specify the subtitle of the program OPT SUBTITLE “< subtitle >” TITLE specify the title of the program OPT TITLE “< title >” Note 1: 6.4.10.1 The default options are listed with an asterisk (*) ASMOPT_OFF AND ASMOPT_ON These controls allow the assembler optimizer to be selectively disabled for sections of assembly code. No code is modified after an ASMOPT_OFF control until a subsequent ASMOPT_ON control is encountered. 6.4.10.2 COND Any conditional code is included in the listing output. See also, the NOCOND control in Section 6.4.10.6 “NOCOND”. 6.4.10.3 EXPAND When EXPAND is in effect, the code generated by macro expansions appears in the listing output. See also, the NOEXPAND control in Section 6.4.10.7 “NOEXPAND”. 2012-2013 Microchip Technology Inc. DS50002053D-page 283 MPLAB® XC8 C Compiler User’s Guide 6.4.10.4 INCLUDE This control causes the file specified by pathname to be textually included at that point in the assembly file. The INCLUDE control must be the last control keyword on the line, for example: OPT INCLUDE "options.h" The driver does not pass any search paths to the assembler, so if the include file is not located in the working directory, the pathname must specify the exact location. See also, the driver option -P in (Section 4.8.10 “-P: Preprocess Assembly Files”) that forces the C preprocessor to preprocess the assembly file, thus allowing use of preprocessor directives, such as #include (see Section 5.14.1 “C Language Comments”). 6.4.10.5 LIST If, previously, the listing was turned off using the NOLIST control, the LIST control automatically turns listing on. Alternatively, the LIST control can include options to control the assembly and the listing. The options are listed in Table 6-9. TABLE 6-9: LIST CONTROL OPTIONS List Option Default Description c= nnn 80 Set the page (i.e., column) width. n= nnn 59 Set the page length. t= ON|OFF OFF Truncate listing output lines. The default wraps lines. p=< device > n/a Set the device type. r=< radix > HEX Set the default radix to HEX, dec or oct. x= ON|OFF OFF Turn macro expansion on or off. See also, the NOLIST control in Section 6.4.10.8 “NOLIST”. 6.4.10.6 NOCOND Using this control will prevent conditional code from being included in the assembly list file output. See also, the COND control in Section 6.4.10.2 “COND”. 6.4.10.7 NOEXPAND The NOEXPAND control disables macro expansion in the assembly list file. The macro call will be listed instead. See the EXPAND control in Section 6.4.10.3 “EXPAND”. Assembly macros are discussed in Section 6.4.9.13 “MACRO and ENDM”. 6.4.10.8 NOLIST This control turns the listing output off from a precise point forward. See also, the LIST control in Section 6.4.10.5 “LIST”. 6.4.10.9 PAGE The PAGE control causes a new page to be started in the listing output. A Control-L (form feed) character will also cause a new page when it is encountered in the source. 6.4.10.10 SPACE The SPACE control places a number of blank lines in the listing output, as specified by its parameter. DS50002053D-page 284 2012-2013 Microchip Technology Inc. Macro Assembler 6.4.10.11 SUBTITLE The SUBTITLE control defines a subtitle to appear at the top of every listing page, but under the title. The string should be enclosed in single or double quotes. See also, the TITLE control in Section 6.4.10.12 “TITLE”. 6.4.10.12 TITLE This control keyword defines a title to appear at the top of every listing page. The string should be enclosed in single or double quotes. See also, the SUBTITLE control in Section 6.4.10.11 “SUBTITLE”. 2012-2013 Microchip Technology Inc. DS50002053D-page 285 MPLAB® XC8 C Compiler User’s Guide 6.5 ASSEMBLY-LEVEL OPTIMIZATIONS The assembler performs optimizations on assembly code, in addition to those optimizations performed by the code generator directly on the C code; see Section 5.13 “Optimizations”. The xc8 driver, by default, instructs the assembler to optimize assembly code that is generated from C code, but to refrain from performing optimizations on hand-written assembly source modules. The latter code can be optimized if required; see Section 4.8.42 “--OPT: Invoke Compiler Optimizations”. Assembly added in-line (see Section 5.12.2 “#asm, #endasm and asm()”) with C code is never optimized. The optimizations that can be performed by the assembler include the following. Note, however, that these optimizations are skipped if the compiler is operating in Free mode. Procedural abstraction is only employed by compilers operating in PRO mode. The compiler operating mode selection is made by an option, see Section 4.8.37 “--MODE: Choose Compiler Operating Mode”. Assembly-level optimizations include: • In-lining of small routines – is done so that a call to the routine is not required. Only very small routines (typically a few instructions) will be changed so that code size is not impacted. This will speed code execution without a significant increase in code size. • Procedural abstraction – is performed on assembly code sequences that appear more than once. This is essentially a reverse in-lining process. The code sequences are abstracted into callable routines that use a label, PLx, where x is a number. A call to this routine will replace every instance of the original code sequence. This optimization reduces code size considerably, with a small impact on code speed. It can, however, adversely impact debugging. • Jump-to-jump type optimizations – are made primarily to tidy the output related to conditional code sequences that follow a generic template. Jump-to-jump can leave jump instructions that jump to other jump instructions. • Unreachable code – is removed. Code can become orphaned by other optimizations and cannot be reached during normal execution, e.g., instructions after a return instruction. The presence of any label is considered a possible entry point, and code following a label is always considered reachable. • Peephole optimizations – are performed on every instruction. These optimizations consider the state of execution at, and immediately around, each instruction – hence the name. They either alter or delete one or more instructions at each step. For example, if W is known to contain the value 0, and an instruction moves W to an address (MOVWF), this might be replaceable with a CLRF instruction. • Psect merging – can be performed to allow other optimizations to take place. Code within the same psect is guaranteed to be located in the same program memory page. So, calls and jumps within the psect do not need to have the page selection bits set before executing. Code using the LJMP and FCALL instructions will benefit from this optimization, see Section 6.4.1 “Assembly Instruction Deviations”. Assembly optimizations can often interfere with debugging in some tools, such as MPLAB IDE. It can be necessary to disable them when debugging code, if that is possible. See Section 4.8.42 “--OPT: Invoke Compiler Optimizations”, for more details. The assembler optimizations can drastically reduce code size. However, they typically have little effect on RAM usage. DS50002053D-page 286 2012-2013 Microchip Technology Inc. Macro Assembler 6.6 ASSEMBLY LIST FILES The assembler will produce an assembly list file if instructed. The xc8 driver option --ASMLIST is typically used to request generation of such a file, see Section 4.8.16 “--ASMLIST: Generate Assembler List Files”. The assembly list file shows the assembly output produced by the compiler for both C and assembly source code. If the assembler optimizers are enabled, the assembly output can be different than assembly source code. So, it is still useful for assembly programming. The list file is in a human-readable form and cannot be go any farther in the compilation sequence. It differs from an assembly output file in that it contains address and op-code data. In addition, the assembler optimizer simplifies some expressions and removes some assembler directives from the listing file for clarity, although these directives are included in the true assembly output files. If you are using the assembly list file to look at the code produced by the compiler, you might wish to turn off the assembler optimizer so that all the compiler-generated directives are shown in the list file. Re-enable the optimizer when continuing development. Section 4.8.42 “--OPT: Invoke Compiler Optimizations” gives more information on controlling the optimizers. Provided that the link stage has successfully concluded, the listing file is updated by the linker so that it contains absolute addresses and symbol values. Thus, you can use the assembler list file to determine the position and exact op codes of instructions. Tick marks “'” in the assembly listing, next to addresses or opcodes, indicate that the linker did not update the list file, most likely due to a compiler error, or a compiler option that stopped compilation before the link stage. For example, in the following listing: 85 86 87 000A' 027F 000B' 1D03 000C' 2800' subwf skipz goto 127,w u15 These marks indicate that addresses are just address offsets into their enclosing psect, and that opcodes have not been fixed up. Any address field in the opcode that has not been fixed up is shown with a value of 0. There is a single assembly list file produced by the assembler for each assembly file passed to it. So, there is a single file produced for all the C source code in a project, including p-code based library code. The file also contains some of the C initialization that forms part of the runtime startup code. There is also a single file produced for each assembly source file. Typically, there is at least one assembly file in each project. It contains some of the runtime startup file and is typically named startup.as. 2012-2013 Microchip Technology Inc. DS50002053D-page 287 MPLAB® XC8 C Compiler User’s Guide 6.6.1 General Format The format of the main listing is in the form shown in Figure 6-1. The line numbers purely relate to the assembly list file and are not associated with the lines numbers in the C or assembly source files. Any assembly that begins with a semicolon indicates it is a comment added by the code generator. Such comments contain either the original source code, which corresponds to the generated assembly, or is a comment inserted by the code generator to explain some action taken. Before the output for each function, there is detailed information regarding that function summarized by the code generator. This information relates to register usage, local variable information, functions called, and the calling function. FIGURE 6-1: 1 768 769 770 771 772 773 774 775 776 777 778 6.6.2 GENERAL FORM OF ASSEMBLY LISTING FILE 4 2 0243 3 0243 00A3 0244 0245 3007 05A3 0252 0008 ;sp2_inpADC.c: 119: void ADC_start(unsigned char chan) ;sp2_inpADC.c: 120: { _ADC_start: ; Regs used in _ADC_start: [reg0,reg3] 3] ] movwf ADC_start@chan 1 line number ;sp2_inpADC.c: 121: chan &= 0x07; 2 address movlw 7 andwf ADC_start@chan 3 op code 5 ;sp2_inpADC.c: 128: } instruction 4 source comment ; ========= function _ADC_start ends ======== 5 assembly Psect Information The assembly list file can be used to determine the name of the psect in which a data object or section of code has been placed by the compiler. For labels (symbols), check the symbol table at the end of the file. It indicates the name of the psect in which it resides, as well as the address associated with the symbol. For other code, find the code in the list file. You can usually search for the C statement associated with the code. Look for the first PSECT assembler directive above this code. This name associated with this directive is the psect in which the code is placed, see Section 6.4.9.3 “PSECT”. 6.6.3 Function Information For each C function, printed before the function’s assembly label (search for the function’s name that is immediately followed by a colon :), is general information relating to the resources used by that function. A typical printout is shown in Figure 6-2: “Function Information”. Most of the information is self-explanatory, but special comments follow. The locations shown use the format offset[space]. For example, a location of 42[BANK0] means that the variable was located in the bank 0 memory space and that it appears at an offset of 42 bytes into the compiled stack component in this space, see Section 5.5.2.2.1 “Compiled Stack Operation”. Whenever pointer variables are shown, they are often accompanied by the targets that the pointer can reference, these targets appear after the arrow ->. See, also, Section 6.6.5 “Pointer Reference Graph”. The auto and parameter section of this information is especially useful because the size of pointers is dynamic; see Section 5.4.5 “Pointer Types”. This information shows the actual number of bytes assigned to each pointer variable. DS50002053D-page 288 2012-2013 Microchip Technology Inc. Macro Assembler The tracked objects are generally not used. It indicates the known state of the currently selected RAM bank on entry to the function and at its exit points. It also indicates the bank selection bits that did, or did not, change in the function. The hardware stack information shows how many stack levels were taken up by this function alone, and the total levels used by this function and any functions it calls. Functions that use a non-reentrant model are those that allocate auto and parameter variables to a compiled stack and which are, hence, not reentrant. If a function is marked as being reentrant, it allocates stack-based variables to the software stack and can be reentrantly called. FIGURE 6-2: FUNCTION INFORMATION 4064 ;; 1 *************** function _render ***************** 4065 ;; Defined at: 4066 ;; line 29 in file "draw.c" 2 3 Parameters: 4067 ;; Size Location Type 4068 ;; None 4069 ;; Auto vars: Size Location Type 4070 ;; lll 4 42[BANK0 ] long 4071 ;; x 2 46[BANK0 ] volatile int 4 4072 ;; cp 1 41[BANK0 ] PTR unsigned char 4073 ;; -> inputData(2), 4074 ;; Type 5 Return value: Size Location 4075 ;; None void 4076 ;; 6 Registers used: 4077 ;; wreg, fsr0l, fsr0h, status,2, status,0, pclath, cstack 4078 ;; 7 Tracked objects: 4079 ;; On entry : 17F/0 4080 ;; On exit : 0/0 4081 ;; Unchanged: FFE00/0 4082 ;; Data sizes: COMMON BANK0 BANK1 BANK2 4083 ;; Params: 0 0 0 0 4084 ;; Locals: 0 7 0 0 8 4085 ;; Temps: 0 5 0 0 4086 ;; Totals: 0 12 0 0 1 function's name 4087 ;;Total ram usage: 12 bytes 4088 ;; Hardware stack levels used: 1 2 file name and line number of definition 4089 ;; 4 9 Hardware stack levels required when called: 4090 ;; This function calls: 3 size, location and type of parameters 4091 ;; _lrv 4 size, location and type of auto variables 4092 ;; ___altofl 10 4093 ;; ___awdiv 5 size, location and type of return value 4094 ;; ___awmod 11 6 registers that the function code used 4095 ;; This function is called by: 4096 ;; _main 7 selected GPR bank on entry and exit rant model r 4097 12 ;; This function uses a non-reentrant 8 RAM memory summary for entire function 9 hardware stack requirements 10 functions called by this function 11 which functions call this function 12 how the function was encoded 2012-2013 Microchip Technology Inc. DS50002053D-page 289 MPLAB® XC8 C Compiler User’s Guide 6.6.4 Switch Statement Information Along with the generated code for each switch statement is information about how that statement was encoded. There are several strategies the compiler can use for switch statements. The compiler determines the appropriate strategy, see Section 5.6.3 “Switch Statements”, or you can indicate a preference for a particular type of strategy using a pragma, see Section 5.14.4.10 “The #pragma switch Directive”. The information printed will look similar to that shown in Figure 6-3. FIGURE 6-3: 206 207 208 209 210 211 212 213 214 215 SWITCH STATEMENT INFORMATION 1 ; Switch size 1, requested type "space" 3 4 ; ; ; ; ; ; ; ; ; Number of cases is 4, Range of values is 0 to 254 switch strategies available: Name Instructions Cycles simple_byte 13 7 (average) jumptable 260 6 (fixed) rangetable 259 6 (fixed) spacedrange 516 9 (fixed) locatedrange 255 3 (fixed) Chosen strategy egy e gy is simple_byte 2 1 size of the switched value 2 number and range of the case values 3 all switch strategies and their attributes 4 the strategy choosen for this switch statement 6.6.5 Pointer Reference Graph Other important information contained in the assembly list file is the pointer reference graph (look for pointer list with targets: in the list file). This is a list of each pointer contained in the program and each target the pointer can reference through the program. The size and type of each target is indicated, as well as the size and type of the pointer variable itself. For example, the following shows a pointer called task_tmr in the C code. It is local to the function timer_intr(). It is also a pointer to an unsigned int, and it is one byte wide. There is only one target to this pointer and it is the member timer_count in the structure called task. This target variable resides in the BANK0 class and is two bytes wide. timer_intr@task_tmr PTR unsigned int size(1); Largest target is 2 -> task.timer_count(BANK0[2]), The pointer reference graph shows both pointers to data objects and pointers to functions. DS50002053D-page 290 2012-2013 Microchip Technology Inc. Macro Assembler 6.6.6 Call Graph The other important information in the assembly list file is the call graph. This is produced for all 8-bit devices, which can use a compiled stack to facilitate stack-based variables (function parameters, auto and temporary variables). See Section 5.5.2.2.1 “Compiled Stack Operation”, for more detailed information on compiled stack operation. Call graph tables, showing call information on a function-by-function basis, are presented in the map file, followed by more traditional call graphs for the entire program. The call graphs are built by the code generator, and are used to allow overlapping of functions’ auto-parameter blocks (APBs) in the compiled stack. The call graphs are not used when functions use the software stack (see Section 5.5.2.2.2 “Software Stack Operation”). You can obtain the following information from studying the call graph. • The functions in the program that are “root” nodes marking the top of a call tree, and that are called spontaneously • The functions that the linker deemed were called, or can have been called, during program execution (and those which were called indirectly via a pointer) • The program’s hierarchy of function calls • The size of the auto and parameter areas within each function’s APB • The offset of each function’s APB within the compiled stack • The estimated call tree depth. These features are discussed in sections that follow. 2012-2013 Microchip Technology Inc. DS50002053D-page 291 MPLAB® XC8 C Compiler User’s Guide 6.6.6.1 CALL GRAPH TABLES A typical call graph table can look like the extract shown in Figure 6-4. Look for Call Graph Tables: in the list file. FIGURE 6-4: CALL GRAPH FORM Call Graph Tables: --------------------------------------------------------------------------------(Depth) Function Calls Base Space Used Autos Params Refs --------------------------------------------------------------------------------(0) _main 12 12 0 34134 43 BANK0 5 5 0 0 BANK1 7 7 0 _aOut _initSPI --------------------------------------------------------------------------------(1) _aOut 2 0 2 68 2 BANK0 2 0 2 _SPI _GetDACValue (ARG) --------------------------------------------------------------------------------(1) _initSPI 0 0 0 0 --------------------------------------------------------------------------------(2) _SPI 2 2 0 23 0 BANK0 2 2 0 ... Estimated maximum stack depth 6 --------------------------------------------------------------------------------- The graph table starts with the function main(). Note that the function name will always be shown in the assembly form, thus the function main() appears as the symbol _main. main() is always a root of a call tree. Interrupt functions will form separate trees. All the functions that main() calls, or can call, are shown below the function name in the Calls column. So in this example, main() calls aOut() and initSPI(). These have been grouped in the orange box in the figure. If a star (*) appears next to the function’s name, this implies the function has been called indirectly via a pointer. A function’s inclusion into the call graph does not imply the function was actually called, but there is a possibility that the function was called. For example, code such as: int test(int a) { if(a) foo(); else bar(); } will list foo() and bar() under test(), as either can be called. If a is always true, then the function bar() will never be called, even though it appears in the call graph. In addition to the called functions, information relating to the memory allocated in the compiled stack for main() is shown. This memory will be used for the stack-based variables that are defined in main(), as well as a temporary location for the function’s return value, if appropriate. DS50002053D-page 292 2012-2013 Microchip Technology Inc. Macro Assembler In the orange box for main() you can see that it defines 12 auto and temporary variable (under the Autos column). It defines no parameters – main() never has parameters – under the Params column. There is a total of 34134 references in the assembly code to local objects in main(), shown under the Refs column. The Used column indicates the total number of bytes of local storage, i.e., the sum of the Autos and Params columns. Rather than the compiled stack being one block of memory in one memory space, it can be broken up into multiple blocks placed in different memory spaces to utilize all of the available memory on the target device. This breakdown is shown under the memory summary line for each function. In this example, it shows that 5 bytes of auto objects for main() are placed in the bank 0 component of the compiled stack (Space column), at an offset of 43 (Base column) into this stack. It also shows that 7 bytes of auto objects were placed in the bank 1 data component of the compiled stack at an offset of 0. The name listed under the Space column, is the same name as the linker class which will hold this section of the stack. Below the information for main() (outside the orange box) you will see the same information repeated for the functions that main() called, i.e., aOut()and initSPI(). For clarity, only the first few functions of this program are shown in the figure. Before the name of each function, and in brackets, is the call stack depth for that particular function. A function can be called from many places in a program, and it can have a different stack depth in the call graph at each call. The maximum call depth is always shown for a function, regardless of its position in the call table. The main() function will always have a depth of 0. The starting call depth for interrupt functions assumes a worst case and will start at the start depth of the previous tree plus one. After each tree in the call graph, there is an indication of the maximum stack depth that might be realized by that tree. This stack depth is not printed if any functions in the graph use the software stack. (In that case, a single stack depth estimate is printed for the entire program at the end of the graphs.) In the example shown, the estimated maximum stack depth is 6. Check the associated data sheet for the depth of your device’s hardware stack (see Section 5.3.4 “Stacks”). The stack depth indicated can be used as a guide to the stack usage of the program. No definitive value can be given for the program’s total stack usage for several reasons: • Certain parts of the call tree can never be reached, reducing that tree’s stack usage. • The exact contribution of interrupt (or other) trees to the main() tree cannot be determined as the point in main’s call tree at which the interrupt (or other function invocation) will occur cannot be known; • The assembler optimizer can have replaced function calls with jumps to functions, reducing that tree’s stack usage. • The assembler’s procedural abstraction optimizations can have added in calls to abstracted routines. (Checks are made to ensure this does not exceed the maximum stack depth.) The compiler can be configured to manage the hardware stack for PIC10/12/16 devices only, see Section 4.8.51 “--RUNTIME: Specify Runtime Environment”. When this mode is selected, the compiler will convert calls to jumps if it thinks the maximum stack depth of the device is being exceeded. The stack depth estimate listed in the call table will reflect the stack savings made by this feature. Thus, the stack depth and call depth cannot be the same. Note that main() is jumped to by the runtime startup, not called; so, main() itself does not consume a level of stack. See also Section 5.10.1 “Runtime Startup Code”. 2012-2013 Microchip Technology Inc. DS50002053D-page 293 MPLAB® XC8 C Compiler User’s Guide The code generator produces a warning if the maximum stack depth appears to have been exceeded and the stack is not being managed by the compiler. For the above reasons, this warning, too, is intended to be a only a guide to potential stack problems. 6.6.6.2 CALL GRAPH CRITICAL PATHS Immediately prior to the call graph tables in the list file are the critical paths for memory usage identified in the call graphs. A critical path is printed for each memory space and for each call graph. Look for a line similar to Critical Paths under _main in BANK0, which, for this example, indicates the critical path for the main function (the root of one call graph) in bank 0 memory. There will be one call graph for the function main and another for each interrupt function. Each of these will appear for every memory space the device defines. A critical path here represents the biggest range of APBs stacked together in a contiguous block. Essentially, it identifies those functions whose APBs are contributing to the program’s memory usage in that particular memory space. If you can reduce the memory usage of these functions in the corresponding memory space, then you will affect the program’s total memory usage in that memory space. This information can be presented as follows. 3793 3794 3795 3796 3797 ;; Critical Paths under _main in BANK0 ;; ;; _main->_foobar ;; _foobar->___flsub ;; ___flsub->___fladd In this example, it shows that of all the call graph paths starting from the function main, the path in which main calls foobar, which calls flsub, which calls fladd, is using the largest block of memory in bank 0 RAM. The exact memory usage of each function is shown in the call graph tables. The memory used by functions that are not in the critical path will overlap entirely with that in the critical path. Reducing the memory usage of these will have no impact on the memory usage of the entire program. 6.6.6.3 CALL GRAPH GRAPHS Following the call tables are the call graphs, which show the full call tree for main() and any interrupt functions. This is a subset of the information presented in the call tables, and it is shown in a different form. The call graphs will look similar to the one shown in Figure 6-5. FIGURE 6-5: CALL GRAPH GRAPHS Call Graph Graphs: _main (ROOT) _initSPI _aOut _SPI _GetDACValue ___ftadd ___ftpack ___ftmul (ARG) ... DS50002053D-page 294 2012-2013 Microchip Technology Inc. Macro Assembler Indentation is used to indicate the call depth. In the diagram, you can see that main() calls aOut(), which in turn calls GetDACValue(), which in turn calls the library function __ftadd(), etc. If a star (*) appears next to the function’s name, this implies that the function has been called indirectly via a pointer. 6.6.6.4 ARG NODES In both the call trees and the call graph itself, you can see functions listed with the annotation (ARG) after its name. This implies that the call to that function at that point in the call graph is made to obtain an argument to another function. For example, in the following code snippet, the function input() is called to obtain an argument value to the function process(). result = process(input(0x7)); For such code, if it were to appear inside the main() function, the call graph would contain the following. _main (ROOT) _input _process _input (ARG) This indicates that main() calls input() and main() also calls process(), but input() is also called as an argument expression to process(). These argument nodes in the graph do not contribute to the overall stack depth usage of the program, but they are important for the creation of the compiled stack. The call depth stack usage of the tree indicated above would only be 1, not 2, even though the argument node function is at an indicated depth of 2. This is because there is no actual reentrancy in terms of an actual call and a return address being stored on the hardware stack. The compiler must ensure that the parameter area for a function and any of its ‘argument functions’ must be at unique addresses in the compiled stack to avoid data corruption. Note that a function’s return value is also stored in its parameter area; so, that needs to be considered by the compiler even if there are no parameters. A function’s parameters become ‘active’ before the function is actually called (when the arguments are passed) and its return value location remains ‘active’ after the function has returned (while that return value is being processed). In terms of data allocation, the compiler assumes a function has been ‘called’ the moment that any of its parameters have been loaded and is still considered ‘called’ up until its return value is no longer required. Thus, the definition for ‘reentrancy’ is much broader when considering data allocation than it is when considering stack call depth. 6.6.7 Symbol Table At the bottom of each assembly list file is a symbol table. This differs from the symbol table presented in the map file (see Section 7.4.2.6 “Symbol Table”) in two ways: • It lists only those symbols associated with the assembly module from which the list file is produced (as opposed to the entire program); and • It lists local as well as global symbols associated with that module. Each symbol is listed along with the address is has been assigned. 2012-2013 Microchip Technology Inc. DS50002053D-page 295 MPLAB® XC8 C Compiler User’s Guide NOTES: DS50002053D-page 296 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Chapter 7. Linker 7.1 INTRODUCTION This chapter describes the theory behind, and the usage of, the linker. The application name of the linker is HLINK. In most instances it will not be necessary to invoke the linker directly, as the compiler driver, xc8, will automatically execute the linker with all the necessary arguments. Using the linker directly is not simple, and should be attempted only by those with a sound knowledge of the compiler, and linking in general. The compiler often makes assumptions about the way in which the program will be linked. If the psects are not linked correctly, code failure can result. If it is absolutely necessary to use the linker directly, the best way to start is to copy the linker arguments constructed by the compiler driver, and modify them as is appropriate. This ensures that the necessary startup module and arguments are present. The following topics are examined in this chapter of the MPLAB XC8 C Compiler User’s Guide: • Operation • Relocation and Psects • Map Files 7.2 OPERATION A command to the linker takes the following form: hlink [options] files The options are zero or more linker options, each of which modifies the behavior of the linker in some way. The files is one or more object files, and zero or more object code library names (.lib extension). Note that P-code libraries (.lpp extension) are always passed to the code generator application. They cannot be passed to the linker. 2012-2013 Microchip Technology Inc. DS50002053D-page 297 MPLAB® XC8 C Compiler User’s Guide The options recognized by the linker are listed in Table 7-1 and discussed in the following paragraphs. TABLE 7-1: LINKER COMMAND-LINE OPTIONS Option -8 -Aclass=low-high ,... -Cx -Cpsect=class -Cbaseaddr -Dclass=delta -Dsymfile -Eerrfile -F -G spec -H symfile -H+ symfile -I -J num -K use 8086 style segment:offset address form specify address ranges for a class call graph options specify a class name for a global psect produce binary output file based at baseaddr specify a class delta value produce old-style symbol file write error messages to errfile produce .obj file with only symbol records specify calculation for segment selectors generate symbol file generate enhanced symbol file ignore undefined symbols set maximum number of errors before aborting prevent overlaying function parameter and auto areas -L preserve relocation items in .obj file -LM preserve segment relocation items in .obj file -N sort symbol table in map file by address order -Nc sort symbol table in map file by class address order -Ns -Mmapfile -Ooutfile -Pspec -Qprocessor -S -Sclass=limit[,bound] sort symbol table in map file by space address order -Usymbol -Vavmap -Wwarnlev -Wwidth -X pre-enter symbol in table as undefined -Z --DISL=list --EDF=path --EMAX=number DS50002053D-page 298 Effect generate a link map in the named file specify name of output file specify psect addresses and ordering specify the device type (for cosmetic reasons only) inhibit listing of symbols in symbol file specify address limit, and start boundary for a class of psects use file avmap to generate an Avocet format symbol file set warning level (-9 to 9) set map file width (>=10) remove any local symbols from the symbol file remove trivial local symbols from the symbol file specify disabled messages specify message file location specify maximum number of errors --NORLF do not relocate list file --VER print version number and stop 2012-2013 Microchip Technology Inc. Linker If the standard input is a file, then this file is assumed to contain the command-line argument. Lines can be broken by leaving a backslash \ at the end of the preceding line. In this fashion, HLINK commands of almost unlimited length can be issued. For example, a link command file called x.lnk and containing the following text: -Z -OX.OBJ -MX.MAP \ -Ptext=0,data=0/,bss,nvram=bss/. \ X.OBJ Y.OBJ Z.OBJ can be passed to the linker by one of the following: hlink @x.lnk hlink < x.lnk Several linker options require memory addresses or sizes to be specified. The syntax for all of these is similar. By default, the number is interpreted as a decimal value. To force interpretation as a HEX number, a trailing H, or h, should be added. For example, 765FH will be treated as a HEX number. 7.2.1 -Aclass =low-high,... Normally psects are linked according to the information given to a -P option (see Section 7.2.19 “-Pspec”). But, sometimes it is desirable to have a class of psects linked into more than one non-contiguous address range. This option allows a number of address ranges to be specified as a class. For example: -ACODE=1020h-7FFEh,8000h-BFFEh specifies that psects in the class CODE are to be linked into the given address ranges, unless they are specifically linked otherwise. Where there are a number of identical, contiguous address ranges, they can be specified with a repeat count following an x character. For example: -ACODE=0-0FFFFhx16 specifies that there are 16 contiguous ranges, each 64k bytes in size, starting from address zero. Even though the ranges are contiguous, no psect will straddle a 64k boundary, thus this can result in different psect placement to the case where the option -ACODE=0-0FFFFFh had been specified, which does not include boundaries on 64k multiples. The -A option does not specify the memory space associated with the address. Once a psect is allocated to a class, the space value of the psect is then assigned to the class, see Section 6.4.9.3.15 “Space”. 7.2.2 -Cx This option is now obsolete. 7.2.3 -Cpsect=class This option allows a psect to be associated with a specific class. Normally, this is not required on the command line because psect classes are specified in object files. See Section 6.4.9.3.3 “Class”. 7.2.4 -Dclass=delta This option allows the delta value for psects that are members of the specified class to be defined. The delta value should be a number. It represents the number of bytes per addressable unit of objects within the psects. Most psects do not need this option as they are defined with a delta value. See Section 6.4.9.3.4 “Delta”. 2012-2013 Microchip Technology Inc. DS50002053D-page 299 MPLAB® XC8 C Compiler User’s Guide 7.2.5 -Dsymfile Use this option to produce an old-style symbol file. An old-style symbol file is an ASCII file, where each line has the link address of the symbol followed by the symbol name. 7.2.6 -Eerrfile Error messages from the linker are written to the standard error stream. Under DOS, there is no convenient way to redirect this to a file (the compiler drivers will redirect standard errors, if standard output is redirected). This option makes the linker write all error messages to the specified file instead of the screen, which is the default standard error destination. 7.2.7 -F Normally the linker will produce an object file that contains both program code and data bytes, and symbol information. Sometimes you want to produce a symbol-only object file that can be used again in a subsequent linker run to supply symbol values. The -F option suppresses data and code bytes from the output file, leaving only the symbol records. This option can be used when part of one project (i.e., a separate build) is to be shared with another, as might be the case with a bootloader and application. The files for one project are compiled using this linker option to produce a symbol-only object file. That file is then linked with the files for the other project. 7.2.8 -Gspec When linking programs using segmented, or bank-switched psects, there are two ways the linker can assign segment addresses, or selectors, to each segment. A segment is defined as a contiguous group of psects where each psect in sequence has both its link and load addresses concatenated with the previous psect in the group. The segment address or selector for the segment is the value derived when a segment type relocation is processed by the linker. By default the segment selector is generated by dividing the base load address of the segment by the relocation quantum of the segment, which is based on the reloc= flag value given to psects at the assembler level, see Section 6.4.9.3.13 “Reloc”. The -G option allows an alternate method for calculating the segment selector. The argument to -G is a string similar to: A /10h-4h where A represents the load address of the segment and / represents division. This means “Take the load address of the psect, divide by 10 HEX, then subtract 4”. This form can be modified by substituting N for A, * for / (to represent multiplication), and adding, rather than subtracting, a constant. The token N is replaced by the ordinal number of the segment, which is allocated by the linker. For example: N*8+4 means “take the segment number, multiply by 8, then add 4”. The result is the segment selector. This particular example would allocate segment selectors in the sequence 4, 12, 20, ... for the number of segments defined. The selector of each psect is shown in the map file. See Section 7.4.2.2 “Psect Information Listed by Module”. 7.2.9 -Hsymfile This option instructs the linker to generate a symbol file. The optional argument symfile specifies the name of the file to receive the data. The default file name is l.sym. DS50002053D-page 300 2012-2013 Microchip Technology Inc. Linker 7.2.10 -H+symfile This option will instruct the linker to generate an enhanced symbol file, which provides, in addition to the standard symbol file, class names associated with each symbol and a segments section which lists each class name and the range of memory it occupies. This format is recommended if the code is to be run in conjunction with a debugger. The optional argument symfile specifies a file to receive the symbol file. The default file name is l.sym. 7.2.11 -I Usually, failure to resolve a reference to an undefined symbol is a fatal error. Using this option causes undefined symbols to be treated as warnings, instead. 7.2.12 -Jerrcount The linker will stop processing object files after a certain number of errors (other than warnings). The default number is 10, but the -J option allows this to be altered. 7.2.13 -K This option should not be used. It is for older compilers that use a compiled stack. In those cases, the linker tries to overlay function auto and parameter blocks to reduce the total amount of RAM required. For debugging purposes, that feature can be disabled with this option. However, doing so will increase the data memory requirements. This option has no effect when compiled stack allocation is performed by the code generator. This is the case for OCG (PRO-Standard-Free mode) compilers, and this option should not be used. 7.2.14 -L When the linker produces an output file it does not usually preserve any relocation information, since the file is now absolute. In some circumstances a further “relocation” of the program is done at load time. The -L option generates, in the output file, one null relocation record for each relocation record in the input. 7.2.15 -LM Similar to the above option, this preserves relocation records in the output file, but only segment relocations. 7.2.16 -Mmapfile This option causes the linker to generate a link map in the named file, or on the standard output, if the file name is omitted. The format of the map file is illustrated in Section 7.4 “Map Files”. 7.2.17 -N, -Ns and-Nc By default the symbol table in the map file is sorted by name. The -N option causes it to be sorted numerically, based on the value of the symbol. The -Ns and -Nc options work similarly except that the symbols are grouped by either their space value, or class. 7.2.18 -Ooutfile This option allows specification of an output file name for the linker. The default output file name is l.obj. Use of this option overrides that default. 2012-2013 Microchip Technology Inc. DS50002053D-page 301 MPLAB® XC8 C Compiler User’s Guide 7.2.19 -Pspec Psects are linked together and assigned addresses based on information supplied to the linker via -P options. The argument to the -P option consists basically of comma -separated sequences thus: -Ppsect =lnkaddr+min/ldaddr+min,psect=lnkaddr/ldaddr,... There are several variations, but essentially each psect is listed with its desired link and load addresses, and a minimum value. All values can be omitted, in which case a default will apply, depending on previous values. If present, the minimum value, min, is preceded by a + sign. It sets a minimum value for the link or load address. The address will be calculated as described below, but if it is less than the minimum then it will be set equal to the minimum. The link and load addresses are either numbers, or the names of other psects, classes, or special tokens. If the link address is a negative number, the psect is linked in reverse order with the top of the psect appearing at the specified address minus one. Psects following a negative address will be placed before the first psect in memory. If a psect’s link address is omitted, it will be derived from the top of the previous psect. For example, in the following: -Ptext=100h,data,bss the text psect is linked at 100h (its load address defaults to the same). The data psect will be linked (and loaded) at an address which is 100 HEX plus the length of the text psect, rounded up as necessary if the data psect has a reloc value associated with it (see Section 6.4.9.3.13 “Reloc”). Similarly, the bss psect will concatenate with the data psect. Again: -Ptext=-100h,data,bss will link in ascending order bss, data then text with the top of the text psect appearing at address 0ffh. If the load address is omitted entirely, it defaults to the same as the link address. If the slash / character is supplied, but no address is supplied after it, the load address will concatenate with the previous psect. For example: -Ptext=0,data=0/,bss will cause both text and data to have a link address of zero; text will have a load address of zero, and data will have a load address starting after the end of text. The bss psect will concatenate with data in terms of both link and load addresses. The load address can be replaced with a dot character, “.”. This tells the linker to set the load address of this psect to the same as its link address. The link or load address can also be the name of another (previously linked) psect. This will explicitly concatenate the current psect with the previously specified psect, for example: -Ptext=0,data=8000h/,bss/. -Pnvram=bss,heap This example shows text at zero, data linked at 8000h but loaded after text; bss is linked and loaded at 8000h plus the size of data, and nvram and heap are concatenated with bss. Note here the use of two -P options. Multiple -P options are processed in order. If -A options (see Section 7.2.1 “-Aclass =low-high,...”) have been used to specify address ranges for a class then this class name can be used in place of a link or load address, and space will be found in one of the address ranges. For example: -ACODE=8000h-BFFEh,E000h-FFFEh -Pdata=C000h/CODE DS50002053D-page 302 2012-2013 Microchip Technology Inc. Linker This will link data at C000h, but find space to load it in the address ranges associated with the CODE class. If no sufficiently large space is available in this class, an error will result. Note that in this case the data psect will still be assembled into one contiguous block, whereas other psects in the class CODE will be distributed into the address ranges wherever they will fit. This means that if there are two or more psects in class CODE, they can be intermixed in the address ranges. Any psects allocated by a -P option will have their load address range subtracted from the address ranges associate with classes in the same memory space. This allows a range to be specified with the -A option without knowing in advance how much of the lower part of the range, for example, will be required for other psects. The final link and load address of psects are shown in the map file. See Section 7.4.2.2 “Psect Information Listed by Module”. 7.2.20 -Qprocessor This option allows a device type to be specified. This is purely for information placed in the map file. The argument to this option is a string describing the device. There are no behavioral changes attributable to the device type. 7.2.21 -S This option prevents symbol information relating from being included in the symbol file produced by the linker. Segment information is still included. 7.2.22 -Sclass =limit[,bound] A class of psects can have an upper address limit associated with it. The following example places a limit on the maximum address of the CODE class of psects to one less than 400h. -SCODE=400h Note that to set an upper limit to a psect, this must be set in assembler code using the psect limit flag, see Section 6.4.9.3.8 “Limit”). If the bound (boundary) argument is used, the class of psects will start on a multiple of the bound address. This example below places the FARCODE class of psects at a multiple of 1000h, but with an upper address limit of 6000h. -SFARCODE=6000h,1000h 7.2.23 -Usymbol This option will enter the specified symbol into the linker’s symbol table as an undefined symbol. This is useful for linking entirely from libraries, or for linking a module from a library where the ordering has been arranged so that by default a later module will be linked. 7.2.24 -Vavmap To produce an Avocet format symbol file, the linker needs to be given a map file to allow it to map psect names to Avocet memory identifiers. The avmap file will normally be supplied with the compiler, or created automatically by the compiler driver as required. 7.2.25 -Wnum The -W option can be used to set the warning level, in the range -9 to 9, or the width of the map file, for values of num >= 10. -W9 will suppress all warning messages. -W0 is the default. Setting the warning level to -9 (-W-9) will give the most comprehensive warning messages. 2012-2013 Microchip Technology Inc. DS50002053D-page 303 MPLAB® XC8 C Compiler User’s Guide 7.2.26 -X Local symbols can be suppressed from a symbol file with this option. Global symbols will always appear in the symbol file. 7.2.27 -Z Some local symbols are compiler generated and not of interest in debugging. This option will suppress from the symbol file all local symbols that have the form of a single alphabetic character, followed by a digit string. The set of letters that can start a trivial symbol is currently “klfLSu“. The -Z option will strip any local symbols starting with one of these letters, and followed by a digit string. 7.2.28 --DISL=message numbers Disable Messages This option is mainly used by the command-line driver, xc8, to disable particular message numbers. It takes a comma-separate list of message numbers that will be disabled during compilation. This option is applied if compiling using xc8, the command-line driver and the --MSGDISABLE driver option, see Section 4.8.38 “--MSGDISABLE: Disable Warning Messages”. See Section 4.6 “Compiler Messages” for full information about the compiler’s messaging system. 7.2.29 --EDF=message file: Set Message File Path This option is mainly used by the command-line driver, xc8, to specify the path of the message description file. The default file is located in the dat directory in the compiler’s installation directory. See Section 4.6 “Compiler Messages” for full information about the compiler’s messaging system. 7.2.30 --EMAX=number: Specify Maximum Number of Errors This option is mainly used by the command-line driver, xc8, to specify the maximum number of errors that can be encountered before the assembler terminates. The default number is 10 errors. This option is applied if compiling using xc8, the command-line driver and the --ERRORS driver option, see Section 4.8.28 “--ERRORS: Maximum Number of Errors”. See Section 4.6 “Compiler Messages” for full information about the compiler’s messaging system. 7.2.31 --NORLF: Do Not Relocate List File Use of this option prevents the linker applying fixups to the assembly list file produced by the assembler. This option is normally using by the command line driver, xc8, when performing pre-link stages, but is omitted when performing the final link step so that the list file shows the final absolute addresses. If you are attempting to resolve fixup errors, this option should be disabled so as to fixup the assembly list file and allow absolute addresses to be calculated for this file. If the compiler driver detects the presence of a preprocessor macro __DEBUG which is equated to 1, then this option will be disabled when building. This macro is set when choosing a Debug build in MPLAB IDE, so always have this selected if you encounter such errors. DS50002053D-page 304 2012-2013 Microchip Technology Inc. Linker 7.2.32 --VER: Print Version Number This option printed information relating to the version and build of the linker. The linker will terminate after processing this option, even if other options and files are present on the command line. 2012-2013 Microchip Technology Inc. DS50002053D-page 305 MPLAB® XC8 C Compiler User’s Guide 7.3 RELOCATION AND PSECTS This section looks at the input files that the linker has to work with. The linker can read both relocatable object files and object-file libraries (.lib extension). The library files are a collection of object files packaged into a single unit. So, essentially, we only need consider the format of object files. Each object file consists of a number of records. Each record has a type that indicates what sort of information it holds. Some record types hold general information about the target device and its configuration, other records types can hold data; and others, program debugging information, for example. A lot of the information in object files relates to psects. Psects are an assembly domain construct and are essentially a block of something, either instructions or data. Everything that contributes to the program is located in a psect. See Section 6.4.8 “Program Sections”, for an introductory guide. There is a particular record type that is used to hold the data in psects. The bulk of each object file consists of psect records containing the executable code and variables etc. We are now in a position to look at the fundamental tasks the linker performs, which are: • combining all the relocatable object files into one • relocation of psects contained in the object files into memory • fixup of symbolic references in the psects There are at least two object files that are passed to the linker. One is produced from all the C code in the project, including C library code. There is only one of these files since the code generator compiles and combines all the C code of the program and produces just the one assembly output. The other file passed to the linker will be the object code produced from the runtime startup code, see Section 4.4.2 “Startup and Initialization”. If there are assembly source files in the project, then there will also be one object file produced for each source file, and these will be passed to the linker. Existing object files, or object file libraries can also be specified in a project; and if present, these will also be passed to the linker. The output of the linker is also an object file, but there is only a single file produced. The file is absolute, since relocation will have been performed by the linker. The output file consists of the information from all input object files, merged together. Relocation consists of placing the psect data into the memory of the target device. The target device memory specification is passed to the linker by the way of linker options. These options are generated by the command-line driver, xc8. There are no linker scripts or means of specifying options in any source file. The default linker options rarely need adjusting. But, they can be changed, if required, with caution, using the driver option -L-, see Section 4.8.6 “-L-: Adjust Linker Options Directly”. When the psects are placed at actual memory locations, symbolic references made in the psects data can be replaced with absolute values. This is a process called fixup. For each psect record in the object file, there is a corresponding relocation record that indicates which bytes (or bits) in the psect record need to be adjusted once relocation is complete. The relocation records also specify how the values are to be determined. A linker fixup overflow error can occur if the value determined by the linker is too large to fit in the “hole” reserved for the value in the psect. See (477) fixup overflow in expression (location 0x* (0x*+*), size *, value 0x*) (Linker) for information on finding the cause of these errors. DS50002053D-page 306 2012-2013 Microchip Technology Inc. Linker 7.4 MAP FILES The map file contains information relating to the relocation of psects and the addresses assigned to symbols within those psects. 7.4.1 Generation If compilation is being performed via MPLAB IDE, a map file is generated by default. If you are using the driver from the command line, you need to use the -M option to request that the map file be produced, see Section 7.2.16 “-Mmapfile”. Map files have the extension .map. Map files are produced by the linker. If the compilation process is stopped before the linker is executed, then no map file is produced. The linker produces a map file, even if it encounters errors. The file, then, helps you track down the cause of the errors. However, if the linker ultimately reports too many errors, the linker did not run to completion, the map file was not created. You can use the --ERRORS option (see Section 4.8.28 “--ERRORS: Maximum Number of Errors”) on the command line to increase the number of errors encountered before the linker exits. 7.4.2 Contents The sections in the map file, in order of appearance, are as follows. • • • • • • • • • • • the compiler name and version number a copy of the command line used to invoke the linker the version number of the object code in the first file linked the machine type a psect summary sorted by the psect’s parent object file a psect summary sorted by the psect’s CLASS a segment summary unused address ranges summary the symbol table information summary for each function information summary for each module Portions of an example map file, along with explanatory text, are shown in the following sections. 7.4.2.1 GENERAL INFORMATION At the top of the map file is general information relating to the execution of the linker. When analyzing a program, always confirm the compiler version number shown in the map file if you have more than one compiler version installed to ensure the desired compiler is being executed. The device selected with the --CHIP option (Section 4.8.18 “--CHIP: Define Device”), or the one selected in your IDE, should appear after the Machine type entry. The object code version relates to the file format used by relocatable object files produced by the assembler. Unless either the assembler or linker have been updated independently, this should not be of concern. 2012-2013 Microchip Technology Inc. DS50002053D-page 307 MPLAB® XC8 C Compiler User’s Guide A typical map file can begin something like the following. This example has been cut down for clarity. --edf=/home/jeff/Microchip/XC8/1.00/dat/en_msgs.txt -cs -h+main.sym -z \ -Q16F946 -ol.obj -Mmain.map -ver=XC8 -ACONST=00h-0FFhx32 \ -ACODE=00h-07FFhx4 -ASTRCODE=00h-01FFFh -AENTRY=00h-0FFhx32 \ -ASTRING=00h-0FFhx32 -ACOMMON=070h-07Fh -ABANK0=020h-06Fh \ -ABANK1=0A0h-0EFh -ABANK2=0120h-016Fh -ABANK3=01A0h-01EFh \ -ARAM=020h-06Fh,0A0h-0EFh,0120h-016Fh,01A0h-01EFh \ -AABS1=020h-07Fh,0A0h-0EFh,0120h-016Fh,01A0h-01EFh -ASFR0=00h-01Fh \ -ASFR1=080h-09Fh -ASFR2=0100h-011Fh -ASFR3=0180h-019Fh \ -preset_vec=00h,intentry,init,end_init -ppowerup=CODE -pfunctab=CODE \ -ACONFIG=02007h-02007h -pconfig=CONFIG -DCONFIG=2 -AIDLOC=02000h-02003h \ -pidloc=IDLOC -DIDLOC=2 -AEEDATA=00h-0FFh/02100h -peeprom_data=EEDATA \ -DEEDATA=2 -DCODE=2 -DSTRCODE=2 -DSTRING=2 -DCONST=2 -DENTRY=2 -k \ startup.obj main.obj Object code version is 3.10 Machine type is 16F946 The Linker command line shows all the command-line options and files that were passed to the linker for the last build. Remember, these are linker options, not command-line driver options. The linker options are necessarily complex. Fortunately, they rarely need adjusting from their default settings. They are formed by the command-line driver, xc8, based on the selected target device and the specified driver options. You can often confirm that driver options were valid by looking at the linker options in the map file. For example, if you ask the driver to reserve an area of memory, you should see a change in the linker options used. If the default linker options must be changed, this can be done indirectly through the driver using the driver -L- option, see Section 4.8.6 “-L-: Adjust Linker Options Directly”. If you use this option, always confirm the change appears correctly in the map file. 7.4.2.2 PSECT INFORMATION LISTED BY MODULE The next section in the map file lists those modules that have made a contribution to the output, and information regarding the psects that these modules have defined. See Section 5.15.1 “Program Sections” for an introductory explanation of psects. This section is heralded by the line that contains the headings: Name Link Load Length Selector Space Scale Under this on the far left is a list of object files. These object files include both files generated from source modules and those that were extracted from object library files (.lib extension). In the latter case, the name of the library file is printed before the object file list. Note that since the code generator combines all C source files (and p-code libraries), there is be one object file representing the entire C part of the program. The object file corresponding to the runtime startup code is normally present in this list. The information in this section of the map file can be used to confirm that a module is making a contribution to the output file and to determine the exact psects that each module defines. Shown are all the psects (under the Name column) that were linked into the program from each object file, and information about that psect. The linker deals with two kinds of addresses: link and load. Generally speaking, the link address of a psect is the address by which it is accessed at runtime. DS50002053D-page 308 2012-2013 Microchip Technology Inc. Linker The load address, which is often the same as the link address, is the address at which the psect starts within the output file (HEX or binary file etc.). If a psect is used to hold bits, the load address is irrelevant and is, instead, used to hold the link address (in bit units) converted into a byte address. The Length of the psect is shown in the units that are used by that psect. The Selector is less commonly used and is of no concern when compiling for PIC devices. The Space field is important as it indicates the memory space in which the psect was placed. For Harvard architecture machines, with separate memory spaces (such as the PIC10/12/16 devices), this field must be used in conjunction with the address to specify an exact storage location. A space of 0 indicates the program memory, and a space of 1 indicates the data memory. See Section 6.4.9.3.15 “Space”. The Scale of a psect indicates the number of address units per byte. This remains blank if the scale is 1, and shows 8 for psects that hold bit objects. The load address of psects that hold bits is used to display the link address converted into units of bytes, rather than the load address. See Section 6.4.9.3.2 “Bit”. For example, the following appears in a map file. ext.obj Name text bss rbit Link 3A 4B 50 Load 3A 4B A Length Selector 22 30 10 4B 2 0 Space 0 1 1 Scale 8 This indicates that one of the files that the linker processed was called ext.obj. (This can have been derived from C code or a source file called ext.as.) This object file contained a text psect, as well as psects called bss and rbit. The psect text was linked at address 3A and bss at address 4B. At first glance, this seems to be a problem, given that text is 22 words long. However, they are in different memory areas, as indicated by the space flag (0 for text and 1 for bss), and so they do not occupy the same memory. The psect rbit contains bit objects, and this can be confirmed by looking at the scale value, which is 8. Again, at first glance it seems that there could be an issue with rbit linked over the top of bss. Their space flags are the same, but since rbit contains bit objects, its link address is in units of bits. The load address field of rbit psect displays the link address converted to byte units, i.e., 50h/8 => Ah. Underneath the object file list there can be a label COMMON. This shows the contribution to the program from program-wide psects, in particular that used by the compiled stack. 2012-2013 Microchip Technology Inc. DS50002053D-page 309 MPLAB® XC8 C Compiler User’s Guide 7.4.2.3 PSECT INFORMATION LISTED BY CLASS The next section in the map file shows the same psect information but grouped by the psects’ class. This section is heralded by the line that contains the headings: TOTAL Name Link Load Length Under this are the class names followed by those psects which belong to this class, see Section 6.4.9.3.3 “Class”. These psects are the same as those listed by module in the above section; there is no new information contained in this section, just a different presentation. 7.4.2.4 SEGMENT LISTING The class listing in the map file is followed by a listing of segments. Typically this section of the map file can be ignored by the user. A segment is a conceptual grouping of contiguous psects in the same memory space, and is used by the linker as an aid in psect placement. There is no segment assembler directive and segments cannot be controlled in any way. This section is heralded by the line that contains the headings: SEGMENTS Name Load Length Top Selector Space Class The name of a segment is derived from the psect in the contiguous group with the lowest link address. This can lead to confusion with the psect with the same name. Do not read psect information from this section of the map file. Again, this section of the map file can be ignored. 7.4.2.5 UNUSED ADDRESS RANGES The last of the memory summaries show the memory that has not been allocated, and is unused. The linker is aware of any memory allocated by the code generator (for absolute variables), and so this free space is accurate. This section follows the heading: UNUSED ADDRESS RANGES and is followed by a list of classes and the memory that is still available in each class. If there is more than one memory range available in a class, each range is printed on a separate line. Any paging boundaries located within a class are not displayed. But, the column Largest block shows the largest contiguous free space (which takes into account any paging in the memory range). If you are looking to see why psects cannot be placed into memory (e.g., cant-find-space type errors) then this is important information to study. Note that the memory associated with a class can overlap that in others, thus the total free space is not simply the addition of all the unused ranges. DS50002053D-page 310 2012-2013 Microchip Technology Inc. Linker 7.4.2.6 SYMBOL TABLE The next section in the map file lists the global symbols that the program defines. This section has the heading: Symbol Table and is followed by two columns in which the symbols are alphabetically listed. As always with the linker, any C derived symbol is shown with its assembler equivalent symbol name. See Section 5.12.3 “Interaction between Assembly and C Code”. The symbols listed in this table are: • Global assembly labels • Global EQU /SET assembler directive labels • Linker-defined symbols Assembly symbols are made global via the GLOBAL assembler directive, see Section 6.4.9.1 “GLOBAL” for more information. Linker-defined symbols act like EQU directives. However, they are defined by the linker during the link process, and no definition for them appears in any source or intermediate file. See Section 5.15.7 “Linker-Defined Symbols”. Each symbol is shown with the psect in which they are placed, and the value (usually an address) that the symbol has been assigned. There is no information encoded into a symbol to indicate whether it represents code or data, nor in which memory space it resides. If the psect of a symbol is shown as (abs), this implies that the symbol is not directly associated with a psect. Such is the case for absolute C variables, or any symbols that are defined using an EQU directive in assembly. Note that a symbol table is also shown in each assembler list file. (See Section 4.8.15 “--ADDRQUAL: Set Compiler Response to Memory Qualifiers” for information on generating these files.) These differ to that shown in the map file as they also list local symbols, and they only show symbols defined in the corresponding module. 7.4.2.7 FUNCTION INFORMATION Following the symbol table is information relating to each function in the program. This information, and its presentation, is identical to the function information displayed in the assembly list file. However, the information from all functions is collated in the one location. See Section 6.6.3 “Function Information” for detailed descriptions of this information. 2012-2013 Microchip Technology Inc. DS50002053D-page 311 MPLAB® XC8 C Compiler User’s Guide 7.4.2.8 MODULE INFORMATION The final section in the map file shows code usage summaries for each module. Each module in the program will show information similar to the following. Module main.c Function Class Link Load Size init main getInput CODE CODE CODE 07D8 07E5 07D9 0000 0000 0000 1 13 4 main.c estimated size: 18 The module name is listed (main.c in the above example). The special module name shared is used for data objects allocated to program memory and to code that is not specific to any particular module. Next, the functions defined by each module are listed. Both user-defined and library functions are shown. The code for each function is placed in a psect. The class in which that psect is located is listed in this section (see Section 5.15.3 “Default Linker Classes”), along with the psect’s link address, load address and its size. The units of the size are the native addressing unit of the device, i.e., bytes for PIC18 devices and Words for other 8-bit devices. After the function list is an estimated size for the program memory component of that module. DS50002053D-page 312 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Chapter 8. Utilities 8.1 INTRODUCTION This chapter discusses some of the utility applications that are bundled with the compiler. The applications discussed in this chapter are those more commonly used, but you do not typically need to execute them directly. Most of their features are invoked indirectly by the command line driver that is based on the command-line arguments or MPLAB X IDE project property selections. The following applications are described in this chapter of the MPLAB XC8 C Compiler User’s Guide: • Librarian • HEXMATE 2012-2013 Microchip Technology Inc. DS50002053D-page 313 MPLAB® XC8 C Compiler User’s Guide 8.2 LIBRARIAN The librarian program, LIBR, has the function of combining several files into a single file known as a library. The reasons you might want to use a library in a project are: • there will be fewer files to link • the file content will be accessed faster • libraries uses less disk space The librarian can build p-code libraries (.lpp extension) from p-code files (.p1 extension), or object code libraries (.lib extension) from object files (.obj extension). P-code libraries should be only created if all the library source code is written in C. Object code libraries should be used for assembly code that is to be built into a library. With both library types, only those modules required by a program will be extracted and included in the program output. 8.2.1 The Library Format The modules in a library are simply concatenated, but a directory of the modules and symbols in the library is maintained at the beginning of a library file. Since this directory is smaller than the sum of the modules, on the first pass the linker can perform faster searches by just reading the directory, and not all the modules. On the second pass, it needs to read only those modules which are required, seeking them over the others. This all minimizes disk I/O when linking. It should be noted that the library format is not a general purpose archiving mechanism as is used by some other compiler systems. This has the advantage that the format can be optimized toward speeding up the linkage process. 8.2.2 Using the Librarian Library files can be built directly using the command-line driver; see Section 4.8.44 “--OUTPUT= type: Specify Output File Type”. In this case, the driver will invoke LIBR with the appropriate options saving you from having to use the librarian directly. You might wish to perform this step manually, or you might need to look at the contents of library files, for example. This section shows how the librarian can be executed from the command-line. The librarian cannot be called from IDEs, such as MPLAB IDE. The librarian program is called LIBR, and the formats of commands to it are as follows: LIBR [options] LIBR [options] k k file.lpp [file1.p1 file2.p1...] file.lib [file1.obj file2.obj ...] options is zero or more librarian options that affect the output of the program. These are listed in Table 8-1. TABLE 8-1: LIBRARIAN COMMAND-LINE OPTIONS Option -P width -W DS50002053D-page 314 Effect Specify page width Suppress non-fatal errors 2012-2013 Microchip Technology Inc. Utilities A key letter, k, denotes the command requested of the librarian (replacing, extracting, or deleting modules, listing modules or symbols). These commands are listed in Table 8-2. TABLE 8-2: LIBRARIAN KEY LETTER COMMANDS Key Meaning r Replace modules d Delete modules x Extract modules m List modules s List modules with symbols o Re-order modules The first file name listed after the key is the name of the library file to be used. The following files, if required, are the modules of the library that is required by the command specified. If you are building a p-code library, the modules listed must be p-code files. If you are building an object file library, the modules listed must be object files. When replacing or extracting modules, the names of the modules to be replaced or extracted must be specified. If no names are supplied, all the modules in the library will be replaced or extracted respectively. Adding a file to a library is performed by requesting the librarian to replace it in the library. Since it is not present, the module will be appended to the library. If the r key is used and the library does not exist, it will be created. When using the d key letter, the named modules will be deleted from the library. In this instance, it is an error not to give any module names. The m and s key letters will list the named modules and, in the case of the s key letter, the global symbols defined or referenced within. A D or U letter is used to indicate whether each symbol is defined in the module, or referenced but undefined. As with the r and x key letters, an empty list of modules means all the modules in the library. The o key takes a list of module names and re-orders the matching modules in the library file so that they have the same order as the one listed on the command line. Modules that are not listed are left in their existing order, and will appear after the re-ordered modules. 8.2.2.1 EXAMPLES Here are some examples of usage of the librarian. The following command: LIBR s pic-stdlib-d24.lpp ctime.p1 lists the global symbols in the modules ctime.p1, as shown here: ctime.p1 D D D D D _moninit _localtime _gmtime _asctime _ctime The D letter before each symbol indicates that these symbols are defined by the module. Using the command above without specifying the module name will list all the symbols defined (or undefined) in the library. The following command deletes the object modules a.obj, b.obj and c.obj from the library lcd.lib: LIBR d lcd.lib a.obj b.obj c.obj 2012-2013 Microchip Technology Inc. DS50002053D-page 315 MPLAB® XC8 C Compiler User’s Guide 8.2.3 Supplying Arguments Since it is often necessary to supply many object file arguments to LIBR, arguments will be read from standard input if no command-line arguments are given. If the standard input is attached to the console, LIBR will prompt for input. Multiple line input can be given by using a backslash as a continuation character on the end of a line. If standard input is redirected from a file, LIBR will take input from the file, without prompting. For example: libr libr> r file.lib 1.obj 2.obj 3.obj \ libr> 4.obj 5.obj 6.obj will perform much the same as if the object files had been typed on the command line. The libr> prompts were printed by LIBR itself, the remainder of the text was typed as input. libr >8)&0xFF); sumB = (sumB & 0xFF) + ((sumB>>8)&0xFF); crcLoop = 0; } data++; } while (--len); if(crcLoop){ sum = (sum & 0xFF) + ((sum>>8)&0xFF); sumB = (sumB & 0xFF) + ((sumB>>8)&0xFF); } sum = (sum & 0xFF) + ((sum>>8)&0xFF); sumB = (sumB & 0xFF) + ((sumB>>8)&0xFF); sumB <<= 8; sum = (sum & 0xFF) | sumB; return sum & 0xFFFF; } // data: bytes to process // len: number of bytes to process // addr: address of starting byte of *data unsigned long fletcher16(unsigned char*data, unsigned long addr, unsigned long len){ unsigned long sum = 0, sumB = 0; unsigned int crcLoop = 0; do { addr %= 2; sum += (*data << (addr*8)); if(addr){ // Do this only every second byte sumB += sum; if(++crcLoop == 256){ sum = (sum & 0xFFFF) + ((sum>>16)&0xFFFF); sumB = (sumB & 0xFFFF) + ((sumB>>16)&0xFFFF); crcLoop = 0; } } data++; addr++; } while (--len); if(crcLoop){ sum = (sum & 0xFFFF) + ((sum>>16)&0xFFFF); sumB = (sumB & 0xFFFF) + ((sumB>>16)&0xFFFF); } // Repeat this just in case there was an overflow in the last addition 2012-2013 Microchip Technology Inc. DS50002053D-page 321 MPLAB® XC8 C Compiler User’s Guide sum = (sum & 0xFFFF) + ((sum>>16)&0xFFFF); sumB = (sumB & 0xFFFF) + ((sumB>>16)&0xFFFF); sumB <<= 16; sum = (sum & 0xFFFF) | sumB; return sum; } 8.3.1.6 -FILL The -FILL option is used for filling unused memory locations with a known value. The usage of this option is: -FILL=[const_width:]fill_expr[@address[:end_address]] where: • const_width has the form wn and signifies the width (n bytes) of each constant in fill_expr. If const_width is not specified, the default value is the native width of the architecture. That is, -FILL=w1:1 with fill every byte with the value 0x01. • fill_expr can use the syntax (where const and increment are n-byte constants): - const fill memory with a repeating constant; i.e., -FILL=0xBEEF becomes 0xBEEF, 0xBEEF, 0xBEEF, 0xBEEF - const+=increment fill memory with an incrementing constant; i.e., -FILL=0xBEEF+=1 becomes 0xBEEF, 0xBEF0, 0xBEF1, 0xBEF2 - const-=increment fill memory with a decrementing constant; i.e., -FILL=0xBEEF-=0x10 becomes 0xBEEF, 0xBEDF, 0xBECF, 0xBEBF - const,const,...,const fill memory with a list of repeating constants; i.e., -FILL=0xDEAD,0xBEEF becomes 0xDEAD,0xBEEF,0xDEAD,0xBEEF • The options following fill_expr result in the following behavior: - @address fill a specific address with fill_expr; i.e., -FILL=0xBEEF@0x1000 puts 0xBEEF at address 1000h - @address:end_address fill a range of memory with fill_expr; i.e., -FILL=0xBEEF@0:0xFF puts 0xBEEF in unused addresses between 0 and 255 All constants can be expressed in (unsigned) binary, octal, decimal or hexadecimal, as per normal C syntax, for example, 1234 is a decimal value, 0xFF00 is hexadecimal and FF00 is illegal. DS50002053D-page 322 2012-2013 Microchip Technology Inc. Utilities 8.3.1.7 -FIND This option is used to detect and log occurrences of an opcode or partial code sequence. The usage of this option is: -FIND=Findcode [mMask]@Start-End [/Align][w][t”Title”] where: • Findcode is the hexadecimal code sequence to search for and is entered in little endian byte order. • Mask is optional. It specifies a bit mask applied over the Findcode value to allow a less restrictive search. It is entered in little endian byte order. • Start and End limit the address range to search. • Align is optional. It specifies that a code sequence can only match if it begins on an address that is a multiple of this value. • w, if present, will cause HEXMATE to issue a warning whenever the code sequence is detected. • Title is optional. It allows a title to be given to this code sequence. Defining a title will make log-reports and messages more descriptive and more readable. A title will not affect the actual search results. Here are some examples. The option -FIND=3412@0-7FFF/2w will detect the code sequence 1234h when aligned on a 2 (two) byte address boundary, between 0h and 7FFFh. w indicates that a warning will be issued each time this sequence is found. In this next example, -FIND=3412M0F00@0-7FFF/2wt”ADDXY”, the option is the same as in last example but the code sequence being matched is masked with 000Fh, so HEXMATE will search for any of the opcodes 123xh, where x is any digit. If a byte-mask is used, is must be of equal byte-width to the opcode it is applied to. Any messaging or reports generated by HEXMATE will refer to this opcode by the name, ADDXY, as this was the title defined for this search. If HEXMATE is generating a log file, it will contain the results of all searches. -FIND accepts whole bytes of HEX data from 1 to 8 bytes in length. Optionally, -FIND can be used in conjunction with REPLACE or DELETE (as described below). 8.3.1.8 -FIND...,DELETE If the DELETE form of the -FIND option is used, any matching sequences will be removed. This function should be used with extreme caution and is not normally recommended for removal of executable code. 8.3.1.9 -FIND...,REPLACE If the REPLACE form of the -FIND option is used, any matching sequences will be replaced, or partially replaced, with new codes. The usage for this sub-option is: -FIND...,REPLACE=Code [mMask] where: • Code is a little endian hexadecimal code to replace the sequences that match the -FIND criteria. • Mask is an optional bit mask to specify which bits within Code will replace the code sequence that has been matched. This can be useful if, for example, it is only necessary to modify 4 bits within a 16-bit instruction. The remaining 12 bits can masked and left unchanged. 2012-2013 Microchip Technology Inc. DS50002053D-page 323 MPLAB® XC8 C Compiler User’s Guide 8.3.1.10 -FORMAT The -FORMAT option can be used to specify a particular variant of INHX format or adjust maximum record length. The usage of this option is: -FORMAT=Type [,Length] where: • Type specifies a particular INHX format to generate. • Length is optional and sets the maximum number of bytes per data record. A valid length is between 1 and 16, with 16 being the default. Consider the case of a bootloader trying to download an INHX32 file, which fails because it cannot process the extended address records that are part of the INHX32 standard. You know that this bootloader can only program data addressed within the range 0 to 64k, and that any data in the HEX file outside of this range can be safely disregarded. In this case, by generating the HEX file in INHX8M format the operation might succeed. The HEXMATE option to do this would be -FORMAT=INHX8M. Now, consider if the same bootloader also required every data record to contain exactly 8 bytes of data. This is possible by combining the -FORMAT with -FILL options. Appropriate use of -FILL can ensure that there are no gaps in the data for the address range being programmed. This will satisfy the minimum data length requirement. To set the maximum length of data records to 8 bytes, just modify the previous option to become -FORMAT=INHX8M,8. The possible types that are supported by this option are listed in Table 8-5. Note that INHX032 is not an actual INHX format. Selection of this type generates an INHX32 file, but will also initialize the upper address information to zero. This is a requirement of some device programmers. TABLE 8-5: INHX TYPES USED IN -FORMAT OPTION Type Description INHX8M cannot program addresses beyond 64K INHX32 can program addresses beyond 64K with extended linear address records INHX032 INHX32 with initialization of upper address to zero 8.3.1.11 -HELP Using -HELP will list all HEXMATE options. Entering another HEXMATE option as a parameter of -HELP will show a detailed help message for the given option. For example: -HELP=string will show additional help for the -STRING HEXMATE option. 8.3.1.12 -LOGFILE The -LOGFILE option saves HEX file statistics to the named file. For example: -LOGFILE=output.hxl will analyze the HEX file that HEXMATE is generating, and save a report to a file named output.hxl. DS50002053D-page 324 2012-2013 Microchip Technology Inc. Utilities 8.3.1.13 -MASK Use this option to logically AND a memory range with a particular bitmask. This is used to ensure that the unimplemented bits in program words (if any) are left blank. The usage of this option is as follows: -MASK=hexcode@start-end where hexcode is a hexadecimal value that will be ANDed with data within the start to end address range. Multibyte mask values can be entered in little endian byte order. 8.3.1.14 -OFILE The generated Intel HEX output will be created in this file. For example: -Oprogram.hex will save the resultant output to program.hex. The output file can take the same name as one of its input files; but, by doing so, it will replace the input file entirely. 8.3.1.15 -SERIAL This option will store a particular HEX value at a fixed address. The usage of this option is: -SERIAL=Code [+/-Increment]@Address [+/-Interval][rRepetitions] where: • Code is a hexadecimal value to store and is entered in little-endian byte order. • Increment is optional and allows the value of Code to change by this value with each repetition (if requested). • Address is the location to store this code, or the first repetition thereof. • Interval is optional and specifies the address shift per repetition of this code. • Repetitions is optional and specifies the number of times to repeat this code. For example: -SERIAL=000001@EFFE will store HEX code 00001h to address EFFEh. Another example: -SERIAL=0000+2@1000+10r5 will store 5 codes, beginning with value 0000 at address 1000h. Subsequent codes will appear at address intervals of +10h and the code value will change in increments of +2h. 8.3.1.16 -SIZE Using the -SIZE option will report the number of bytes of data within the resultant HEX image to standard output. The size will also be recorded in the log file if one has been requested. 2012-2013 Microchip Technology Inc. DS50002053D-page 325 MPLAB® XC8 C Compiler User’s Guide 8.3.1.17 -STRING The -STRING option will embed an ASCII string at a fixed address. The usage of this option is: -STRING@Address [tCode]=”Text” where: • Address is the location to store this string. • Code is optional and allows a byte sequence to trail each byte in the string. This can allow the bytes of the string to be encoded within an instruction. • Text is the string to convert to ASCII and embed. For example: -STRING@1000=”My favorite string” will store the ASCII data for the string, My favorite string (including the null character terminator), at address 1000h. And again: -STRING@1000t34=”My favorite string” will store the same string, with every byte in the string being trailed with the HEX code 34h. 8.3.1.18 -STRPACK This option performs the same function as -STRING, but with two important differences. First, only the lower seven bits from each character are stored. Pairs of 7-bit characters are then concatenated and stored as a 14-bit word rather than in separate bytes. This is known as string packing. This is usually only useful for devices where program space is addressed as 14-bit words (PIC10/12/16 devices). The second difference is that -STRING’s t specifier is not applicable with the -STRPACK option. DS50002053D-page 326 2012-2013 Microchip Technology Inc. MPLAB® XC8 C COMPILER USER’S GUIDE Appendix A. Library Functions A.1 INTRODUCTION The functions and preprocessor macros within the standard compiler library are alphabetically listed in this chapter. The synopsis indicates the header file in which a declaration or definition for function or macro is found. It also shows the function prototype for functions, or the equivalent prototype for macros. Note that where printf() is shown in example code, this assumes that the putch() function has been defined to suit the peripheral that will act as the stdout stream. Initialization of that peripheral must also be performed before you attempt to print. __BUILTIN_SOFTWARE_BREAKPOINT Synopsis #include void __builtin_software_breakpoint(void); Description This builtin unconditionally inserts code into the program output which triggers a software breakpoint when the code is executed using a debugger. The software breakpoint code is only generated for mid-range and PIC18 devices. Baseline devices do not support software breakpoints in this way, and the builtin will be ignored if used with these devices. Example #include int main (void) { __builtin_software_breakpoint(); ... // stop here to begin See also __debug_break() 2012-2013 Microchip Technology Inc. DS50002053D-page 327 MPLAB® XC8 C Compiler User’s Guide __CONFIG (BASELINE & MID-RANGE DEVICES) Synopsis #include __CONFIG(data) Description This macro is provided for legacy support only. Use the #pragma config for new projects. This macro is used to program the Configuration fuses that set the device’s operating modes. The macro assumes the argument is a16-bit value, which will be used to program the Configuration bits. 16-bit masks have been defined to describe each programmable attribute available on each device. These masks can be found in the chip-specific header files included via . Multiple attributes can be selected by ANDing them together. Example #include __CONFIG(RC & UNPROTECT) void main (void) { } See also __EEPROM_DATA(), __IDLOC(), __IDLOC7(), CONFIG() (PIC18), #pragma config DS50002053D-page 328 2012-2013 Microchip Technology Inc. Library Functions __CONFIG (PIC18) Synopsis #include __CONFIG(num, data) Description This macro is provided for legacy support only. Use the #pragma config for new projects. This macro is used to program the Configuration fuses that set the device’s operating modes. The macro accepts the number corresponding to the Configuration register it is to program, then the 16-bit value it is to update it with. 16-bit masks have been defined to describe each programmable attribute available on each device. These masks can be found in the chip-specific header files included via . Multiple attributes can be selected by ANDing them together. Example #include __CONFIG(1,RC & OSCEN) __CONFIG(2,WDTPS16 & BORV45) __CONFIG(4, DEBUGEN) void main (void) { } See also __EEPROM_DATA(), __IDLOC(), __IDLOC7(), CONFIG() (baseline & mid-range devices), #pragma config 2012-2013 Microchip Technology Inc. DS50002053D-page 329 MPLAB® XC8 C Compiler User’s Guide __DEBUG_BREAK Synopsis #include void __debug_break(void); Description This macro conditionally inserts code into the program output which triggers a software breakpoint when the code is executed using a debugger. The code is only generated for debug builds (see Section 3.3.9 “What is Different About an MPLAB X IDE Debug Build?”) and is omitted for production builds. The software breakpoint code is only generated for mid-range and PIC18 devices. Baseline devices do not support software breakpoints in this way, and the macro will be ignored if used with these devices. Example #include int main (void) { __debug_break(); ... // stop here to begin See also __builtin_software_breakpoint() __DELAY_MS, __DELAY_US, __DELAYWDT_US, __DELAYWDT_MS Synopsis __delay_ms(x) // __delay_us(x) // __delaywdt_ms(x) __delaywdt_us(x) request a delay in request a delay in // request a delay // request a delay milliseconds microseconds in milliseconds in microseconds Description It is often more convenient to request a delay in time-based terms, rather than in cycle counts. The macros __delay_ms(x) and __delay_us(x) are provided to meet this need. These macros convert the time-based request into instruction cycles that can be used with _delay(n). In order to achieve this, these macros require the prior definition of preprocessor symbol _XTAL_FREQ, which indicates the system frequency. This symbol should equate to the oscillator frequency (in hertz) used by the system. On PIC18 devices only, you can use the alternate WDT-form of these functions, which uses the CLRWDT instruction as part of the delay code. See the _delay function. The macro argument must be a constant expression. An error will result if these macros are used without defining the oscillator frequency symbol, the delay period requested is too large, or the delay period is not a constant. See also _delay() DS50002053D-page 330 2012-2013 Microchip Technology Inc. Library Functions __EEPROM_DATA Synopsis #include __EEPROM_DATA(a,b,c,d,e,f,g,h) Description This macro is used to store initial values in the device’s EEPROM registers at the time of programming. The macro must be given blocks of 8 bytes to write each time it is called, and can be called repeatedly to store multiple blocks. __EEPROM_DATA() will begin writing to EEPROM address zero, and auto-increments the address written to by 8 each time it is used. Example #include __EEPROM_DATA(0x00,0x01,0x02,0x03,0x04,0x05,0x06,0x07) __EEPROM_DATA(0x08,0x09,0x0A,0x0B,0x0C,0x0D,0x0E,0x0F) void main (void) { } See also __CONFIG() 2012-2013 Microchip Technology Inc. DS50002053D-page 331 MPLAB® XC8 C Compiler User’s Guide __IDLOC Synopsis #include __IDLOC(x) Description This macro is provided for legacy support only. Use the #pragma config for new projects. This macro places data into the device’s special locations, outside of addressable memory, that is reserved for ID. This would be useful for storage of serial numbers, etc. The macro will attempt to write 4 nibbles of data to the 4 locations reserved for ID purposes. Example #include /* will store 1, 5, F and 0 in the ID registers */ __IDLOC(15F0); void main (void) { } See also __IDLOC7(), __CONFIG() DS50002053D-page 332 2012-2013 Microchip Technology Inc. Library Functions __IDLOC7 Synopsis #include __IDLOC7(a,b,c,d) Description This macro is provided for legacy support only. Use the #pragma config for new projects. This macro places data into the device’s special locations, outside of addressable memory, that is reserved for ID. This would be useful for storage of serial numbers, etc. The macro will attempt to write 7 bits of data to each of the 4 locations reserved for ID purposes. Example #include /* will store 7Fh, 70, 1 and 5Ah in the ID registers */ __IDLOC(0x7F,70,1,0x5A); void main (void) { } Note Not all devices permit 7-bit programming of the ID locations. Refer to the device data sheet to see whether this macro can be used on your particular device. See also __IDLOC(), __CONFIG() 2012-2013 Microchip Technology Inc. DS50002053D-page 333 MPLAB® XC8 C Compiler User’s Guide _DELAY() , _DELAYWDT Synopsis #include void _delay(unsigned long cycles); void _delaywdt(unsigned long cycles); Description This is an in-line function that is expanded by the code generator. When called, this routine expands to an in-line assembly delay sequence. The sequence will consist of code that delays for the number of instruction cycles that is specified as the argument. The argument must be a constant expression. The _delay in-line function can use loops and the NOP instruction to implement the delay. On PIC18 devices only, the _delaywdt in-line function performs the same task, but can use the CLRWDT instruction, as well as loops to achieve the specified delay. An error will result if the delay period requested is not a constant expression or is too large (approximately 179,200 for PIC18 devices, and 50,659,000 instructions for other 8-bit PIC devices). For very large delays, call this function multiple times. Example #include void main (void) { control |= 0x80; _delay(10); // delay for 10 cycles control &= 0x7F; } See Also _delay3(), __delay_us(), __delay_ms() DS50002053D-page 334 2012-2013 Microchip Technology Inc. Library Functions __OSCCAL_VAL Synopsis #include unsigned char __osccal_val(void); Description This is a pseudo-function that is defined by the code generator to be a label only. The label’s value is equated to the address of the RETLW instruction, which encapsulates the oscillator configuration value. This function is only available for those devices that are shipped with such a value stored in program memory. Calls to the function will return the device’s oscillator configuration value, which can then be used in any expression, if required. Note that this function is automatically called by the runtime start-up code (unless you have explicitly disabled this option, see Section 4.8.51 “--RUNTIME: Specify Runtime Environment”) and you do not need to call it to calibrate the internal oscillator. Example #include void main (void) { unsigned char c; c = __osccal_val(); } 2012-2013 Microchip Technology Inc. DS50002053D-page 335 MPLAB® XC8 C Compiler User’s Guide _DELAY3() Synopsis #include void _delay3(unsigned char cycles); Description This is an in-line function that is expanded by the code generator. When called, this routine expands to an in-line assembly delay sequence.The sequence will consist of code that delays for 3 times the number of cycles that is specified as argument. The argument can be a byte-sized constant or variable. Example #include void main (void) { control |= 0x80; _delay3(10); // delay for 30 cycles control &= 0x7F; } See Also _delay ABS Synopsis #include int abs (int j) Description The abs() function returns the absolute value of j. Example #include #include void main (void) { int a = -5; printf(“The absolute value of %d is %d\n”, a, abs(a)); } See Also labs(), fabs() Return Value The absolute value of j. DS50002053D-page 336 2012-2013 Microchip Technology Inc. Library Functions ACOS Synopsis #include double acos (double f) Description The acos() function implements the inverse of cos(); i.e., it is passed a value in the range -1 to +1, and returns an angle in radians thats cosine is equal to that value. Example #include #include /* Print acos() values for -1 to 1 in degrees. */ void main (void) { float i, a; for(i = -1.0; i < 1.0 ; i += 0.1) { a = acos(i)*180.0/3.141592; printf(“acos(%f) = %f degrees\n”, i, a); } } See Also sin(), cos(), tan(), asin(), atan(), atan2() Return Value An angle in radians, in the range 0 to 2012-2013 Microchip Technology Inc. DS50002053D-page 337 MPLAB® XC8 C Compiler User’s Guide ASCTIME Synopsis #include char * asctime (struct tm * t) Description The asctime() function takes the time broken down into the struct tm structure, pointed to by its argument, and returns a 26 character string describing the current date and time in the format: Sun Sep 16 01:03:52 1973\n\0 Note the newline at the end of the string. The width of each field in the string is fixed. The example gets the current time, converts it to a struct tm with localtime(), it then converts this to ASCII and prints it. The time() function will need to be provided by the user (see time() for details). Example #include #include void main (void) { time_t clock; struct tm * tp; time(&clock); tp = localtime(&clock); printf(“%s”, asctime(tp)); } See Also ctime(), gmtime(), localtime(), time() Return Value A pointer to the string. Note The example will require the user to provide the time() routine as it cannot be supplied with the compiler. See time() for more details. DS50002053D-page 338 2012-2013 Microchip Technology Inc. Library Functions ASIN Synopsis #include double asin (double f) Description The asin() function implements the converse of sin(); i.e., it is passed a value in the range -1 to +1, and returns an angle in radians whose sine is equal to that value. Example #include #include void main (void) { float i, a; for(i = -1.0; i < 1.0 ; i += 0.1) { a = asin(i)*180.0/3.141592; printf(“asin(%f) = %f degrees\n”, i, a); } } See Also sin(), cos(), tan(), acos(), atan(), atan2() Return Value An angle in radians, in the range -/2 – /2. 2012-2013 Microchip Technology Inc. DS50002053D-page 339 MPLAB® XC8 C Compiler User’s Guide ASSERT Synopsis #include void assert (int e) Description This macro is used for debugging purposes; the basic method of usage is to place assertions liberally throughout your code at points where correct operation of the code depends upon certain conditions being true initially. An assert() routine can be used to ensure at runtime that an assumption holds true. For example, the following statement asserts that tp is not equal to NULL: assert(tp); If at runtime the expression evaluates to false, the program will abort with a message identifying the source file and line number of the assertion, and the expression used as an argument to it. A fuller discussion of the uses of assert() is impossible in limited space, but it is closely linked to methods of proving program correctness. The assert() macro depends on the implementation of the function _fassert(). By default this prints information using printf(). This routine should be inspected to ensure it meets your application needs. Include the source file containing this function, even if you do not modify it, into your project and then rebuild. The _fassert() function is not built into any library file. Example #include void ptrfunc (struct xyz * tp) { assert(tp != 0); } Note The underlying routine _fassert(...) will need to be implemented by the user. DS50002053D-page 340 2012-2013 Microchip Technology Inc. Library Functions ATAN Synopsis #include double atan (double x) Description This function returns the arc tangent of its argument; i.e., it returns an angle ‘e’ in the range -/2 – /2. Example #include #include void main (void) { printf(“atan(%f) is %f\n”, 1.5, atan(1.5)); } See Also sin(), cos(), tan(), asin(), acos(), atan2() Return Value The arc tangent of its argument. ATAN2 Synopsis #include double atan2 (double x, double x) Description This function returns the arc tangent of y/x. Example #include #include void main (void) { printf(“atan2(%f, %f) is %f\n”, 10.0, -10.0, atan2(10.0, -10.0)); } See Also sin(), cos(), tan(), asin(), acos(), atan() Return Value The arc tangent of y/x. 2012-2013 Microchip Technology Inc. DS50002053D-page 341 MPLAB® XC8 C Compiler User’s Guide ATOF Synopsis #include double atof (const char * s) Description The atof() function scans the character string passed to it, skipping leading blanks. It then converts an ASCII representation of a number to a double. The number can be in decimal, normal floating point or scientific notation. Example #include #include void main (void) { char buf[80]; double i; gets(buf); i = atof(buf); printf(“Read %s: converted to %f\n”, buf, i); } See Also atoi(), atol(), strtod() Return Value A double precision floating-point number. If no number is found in the string, 0.0 will be returned. DS50002053D-page 342 2012-2013 Microchip Technology Inc. Library Functions ATOI Synopsis #include int atoi (const char * s) Description The atoi() function scans the character string passed to it, skipping leading blanks and reading an optional sign. It then converts an ASCII representation of a decimal number to an integer. Example #include #include void main (void) { char buf[80]; int i; gets(buf); i = atoi(buf); printf(“Read %s: converted to %d\n”, buf, i); } See Also xtoi(), atof(), atol() Return Value A signed integer. If no number is found in the string, 0 will be returned. 2012-2013 Microchip Technology Inc. DS50002053D-page 343 MPLAB® XC8 C Compiler User’s Guide ATOL Synopsis #include long atol (const char * s) Description The atol() function scans the character string passed to it, skipping leading blanks. It then converts an ASCII representation of a decimal number to a long integer. Example #include #include void main (void) { char buf[80]; long i; gets(buf); i = atol(buf); printf(“Read %s: converted to %ld\n”, buf, i); } See Also atoi(), atof() Return Value A long integer. If no number is found in the string, 0 will be returned. DS50002053D-page 344 2012-2013 Microchip Technology Inc. Library Functions BSEARCH Synopsis #include void * bsearch (const void * key, void * base, size_t n_memb, size_t size, int (*compar)(const void *, const void *)) Description The bsearch() function searches a sorted array for an element matching a particular key. It uses a binary search algorithm, calling the function pointed to by compar to compare elements in the array. Example #include #include #include struct value { char name[10]; int value; } values[] = { { "foobar", 66 }; { "casbar", 87 }; { "crossbar", 105 }; }; int val_cmp (const void * p1, const void * p2) { return strcmp(((const struct value *)p1)->name, ((const struct value *)p2)->name); } void main (void) { int i = sizeof(values)/sizeof(struct value); struct value * vp; qsort(values, i, sizeof values[0], val_cmp); vp = bsearch(“fred”, values, i, sizeof values[0], val_cmp); if(!vp) printf(“Item ’fred’ was not found\n”); else printf(“Item ’fred’ has value %d\n”, vp->value); } See Also qsort() Return Value A pointer to the matched array element (if there is more than one matching element, any of these can be returned). If no match is found, a null is returned. 2012-2013 Microchip Technology Inc. DS50002053D-page 345 MPLAB® XC8 C Compiler User’s Guide Note The comparison function must have the correct prototype. DS50002053D-page 346 2012-2013 Microchip Technology Inc. Library Functions CEIL Synopsis #include double ceil (double f) Description This routine returns the smallest whole number not less than f. Example #include #include #include void main (void) { double j; j = 2.345 * rand() printf(“The ceiling of %f is %f\n”, j, ceil(j)); } 2012-2013 Microchip Technology Inc. DS50002053D-page 347 MPLAB® XC8 C Compiler User’s Guide CGETS Synopsis #include char * cgets (char * s) Description The cgets() function will read one line of input from the console into the buffer passed as an argument. It does so by repeated calls to getche(). As characters are read, they are buffered, with backspace deleting the previously typed character, and ctrl-U deleting the entire line typed so far. Other characters are placed in the buffer, with a carriage return or line feed (newline) terminating the function. The collected string is null terminated. Example #include #include char buffer[80]; void main (void) { for(;;) { cgets(buffer); if(strcmp(buffer, “exit” == 0) break; cputs(“Type ’exit’ to finish\n”); } } See Also getch(), getche(), putch(), cputs() Return Value The return value is the character passed as the sole argument. DS50002053D-page 348 2012-2013 Microchip Technology Inc. Library Functions CLRWDT Synopsis #include CLRWDT(); Description This macro is used to clear the device’s internal watchdog timer. Example #include void main (void) { WDTCON=1; /* enable the WDT */ CLRWDT(); } COS Synopsis #include double cos (double f) Description This function yields the cosine of its argument, which is an angle in radians. The cosine is calculated by expansion of a polynomial series approximation. Example #include #include #define C 3.141592/180.0 void main (void) { double i; for(i = 0 ; i <= 180.0 ; i += 10) printf(“sin(%3.0f) = %f, cos = %f\n”, i, sin(i*C), cos(i*C)); } See Also sin(), tan(), asin(), acos(), atan(), atan2() Return Value A double in the range -1 to +1. 2012-2013 Microchip Technology Inc. DS50002053D-page 349 MPLAB® XC8 C Compiler User’s Guide COSH, SINH, TANH Synopsis #include double cosh (double f) double sinh (double f) double tanh (double f) Description These functions are the implement hyperbolic equivalents of the trigonometric functions; cos(), sin() and tan(). Example #include #include void main (void) { printf(“%f\n”, cosh(1.5)); printf(“%f\n”, sinh(1.5)); printf(“%f\n”, tanh(1.5)); } Return Value The function cosh() returns the hyperbolic cosine value. The function sinh() returns the hyperbolic sine value. The function tanh() returns the hyperbolic tangent value. DS50002053D-page 350 2012-2013 Microchip Technology Inc. Library Functions CPUTS Synopsis #include void cputs (const char * s) Description The cputs() function writes its argument string to the console, outputting carriage returns before each newline in the string. It calls putch() repeatedly. On a hosted system cputs() differs from puts() in that it writes to the console directly, rather than using file I/O. In an embedded system cputs() and puts() are equivalent. Example #include #include char buffer[80]; void main (void) { for(;;) { cgets(buffer); if(strcmp(buffer, “exit” == 0) break; cputs(“Type ’exit’ to finish\n”; } } See Also cputs(), puts(), putch() 2012-2013 Microchip Technology Inc. DS50002053D-page 351 MPLAB® XC8 C Compiler User’s Guide CTIME Synopsis #include char * ctime (time_t * t) Description The ctime() function converts the time in seconds pointed to by its argument to a string of the same form as described for asctime(). Thus the example program prints the current time and date. Example #include #include void main (void) { time_t clock; time(&clock); printf(“%s”, ctime(&clock)); } See Also gmtime(), localtime(), asctime(), time() Return Value A pointer to the string. Note The example will require the user to provide the time() routine as one cannot be supplied with the compiler. See time() for more detail. DS50002053D-page 352 2012-2013 Microchip Technology Inc. Library Functions DI, EI Synopsis #include void ei (void) void di (void) Description The di() and ei() routines disable and re-enable interrupts respectively. These are implemented as macros. The example shows the use of ei() and di() around access to a long variable that is modified during an interrupt. If this was not done, it would be possible to return an incorrect value, if the interrupt occurred between accesses to successive words of the count value. The ei() macro should never be called in an interrupt function, and there is no need to call di() in an interrupt function. Example #include long count; void interrupt tick (void) { count++; } long getticks (void) { long val; /* Disable interrupts around access to count, to ensure consistency.*/ di(); val = count; ei(); return val; } 2012-2013 Microchip Technology Inc. DS50002053D-page 353 MPLAB® XC8 C Compiler User’s Guide DIV Synopsis #include div_t div (int numer, int denom) Description The div() function computes the quotient and remainder of the numerator divided by the denominator. Example #include #include void main (void) { div_t x; x = div(12345, 66); printf(“quotient = %d, remainder = %d\n”, x.quot, x.rem); } See Also udiv(), ldiv(), uldiv() Return Value Returns the quotient and remainder into the div_t structure. DS50002053D-page 354 2012-2013 Microchip Technology Inc. Library Functions EEPROM ROUTINES Description These functions are now supplied in the PIC18 peripheral library. See the peripheral library documentation (docs directory) for full information on this library function. EVAL_POLY Synopsis #include double eval_poly (double x, const double * d, int n) Description The eval_poly() function evaluates a polynomial, whose coefficients are contained in the array d, at x, for example: y = x*x*d2 + x*d1 + d0. The order of the polynomial is passed in n. Example #include #include void main (void) { double x, y; double d[3] = {1.1, 3.5, 2.7}; x = 2.2; y = eval_poly(x, d, 2); printf(“The polynomial evaluated at %f is %f\n”, x, y); } Return Value A double value, being the polynomial evaluated at x. 2012-2013 Microchip Technology Inc. DS50002053D-page 355 MPLAB® XC8 C Compiler User’s Guide EXP Synopsis #include double exp (double f) Description The exp() routine returns the exponential function of its argument; i.e., ‘e’ to the power of ‘f’. Example #include #include void main (void) { double f; for(f = 0.0 ; f <= 5 ; f += 1.0) printf(“e to %1.0f = %f\n”, f, exp(f)); } See Also log(), log10(), pow() FABS Synopsis #include double fabs (double f) Description This routine returns the absolute value of its double argument. Example #include #include void main (void) { printf("%f %f\n", fabs(1.5), fabs(-1.5)); } See Also abs(), labs() DS50002053D-page 356 2012-2013 Microchip Technology Inc. Library Functions FLASH ROUTINES Description These functions are now supplied in the PIC18 peripheral library. See the peripheral library documentation (docs directory) for full information on this library function. FLOOR Synopsis #include double floor (double f) Description This routine returns the largest whole number not greater than f. Example #include #include void main (void) { printf("%f\n", floor( 1.5 )); printf("%f\n", floor( -1.5)); } FMOD Synopsis #include double fmod (double x, double y) Description The function fmod returns the remainder of x/y as a floating-point quantity. Example #include void main (void) { double rem, x; x = 12.34; rem = fmod(x, 2.1); } Return Value The floating-point remainder of x/y. 2012-2013 Microchip Technology Inc. DS50002053D-page 357 MPLAB® XC8 C Compiler User’s Guide FREXP Synopsis #include double frexp (double f, int * p) Description The frexp() function breaks a floating-point number into a normalized fraction and an integral power of 2. The integer is stored into the int object pointed to by p. Its return value x is in the interval (0.5, 1.0) or zero, and f equals x times 2 raised to the power stored in *p. If f is zero, both parts of the result are zero. Example #include #include void main (void) { double f; int i; f = frexp(23456.34, &i); printf("23456.34 = %f * 2^%d\n", f, i); } See Also ldexp() DS50002053D-page 358 2012-2013 Microchip Technology Inc. Library Functions FTOA Synopsis #include char * ftoa (float f, int * status) Description The function ftoa converts the contents of f into a string which is stored into a buffer which is then return. Example #include #include void main (void) { char * buf; float input = 12.34; int status; buf = ftoa(input, &status); printf("The buffer holds %s\n", buf); } See Also strtol(), itoa(), utoa(), ultoa() Return Value This routine returns a reference to the buffer into which the result is written. 2012-2013 Microchip Technology Inc. DS50002053D-page 359 MPLAB® XC8 C Compiler User’s Guide GETCH Synopsis #include char getch (void) Description The getch() function is provided as an empty stub which can be completed as each project requires. Typically this function will read one byte of data from a peripheral that is associated with stdin, and return this value. Example #include char result; void main (void) { result = getch(); } See Also getche(), getchar() GETCHE Synopsis #include char getche (void) Description The getche() function is provided as an empty stub which can be completed as each project requires. Typically this function will read one byte of data from a peripheral that is associated with stdin, and return this value. Unlike getch(), it echoes this character received. Example #include char result; void main (void) { result = getche(); } See Also getch(), getchar() DS50002053D-page 360 2012-2013 Microchip Technology Inc. Library Functions GETCHAR Synopsis #include int getchar (void) Description The getchar() routine usually reads from stdin, but is implemented as a call to getche(). Example #include void main (void) { int c; while((c = getchar()) != EOF) putchar(c); } See Also getc(), getche() Note This routine calls getche(). 2012-2013 Microchip Technology Inc. DS50002053D-page 361 MPLAB® XC8 C Compiler User’s Guide GETS Synopsis #include char * gets (char * s) Description The gets() function reads a line from standard input into the buffer at s, deleting the newline (compare: fgets()). The buffer is null terminated. In an embedded system, gets() is equivalent to cgets(), and results in getche() being called repeatedly to get characters. Editing (with backspace) is available. Example #include void main (void) { char buf[80]; printf("Type a line: "); if(gets(buf)) puts(buf); } See Also fgets(), freopen(), puts() Return Value It returns its argument, or NULL on end-of-file. DS50002053D-page 362 2012-2013 Microchip Technology Inc. Library Functions GET_CAL_DATA Synopsis #include double get_cal_data (const unsigned char * code_ptr) Description This function returns the 32-bit floating-point calibration data from the PIC MCU 14000 calibration space. Only use this function to access KREF, KBG, VHTHERM and KTC (that is, the 32-bit floating-point parameters). FOSC and TWDT can be accessed directly as they are bytes. Example #include void main (void) { double x; unsigned char y; /* Get the slope reference ratio. */ x = get_cal_data(KREF); /* Get the WDT time-out. */ y = TWDT; } Return Value The value of the calibration parameter Note This function can only be used on the PIC14000. 2012-2013 Microchip Technology Inc. DS50002053D-page 363 MPLAB® XC8 C Compiler User’s Guide GMTIME Synopsis #include struct tm * gmtime (time_t * t) Description This function converts the time pointed to by t which is in seconds since 00:00:00 on Jan 1, 1970, into a broken down time stored in a structure as defined in time.h. The structure is defined in the ‘Data Types’ section. Example #include #include void main (void) { time_t clock; struct tm * tp; time(&clock); tp = gmtime(&clock); printf("It’s %d in London\n", tp->tm_year+1900); } See Also ctime(), asctime(), time(), localtime() Return Value Returns a structure of type tm. Note The example will require the user to provide the time() routine as one cannot be supplied with the compiler. See time() for more detail. DS50002053D-page 364 2012-2013 Microchip Technology Inc. Library Functions ISALNUM, ISALPHA, ISDIGIT, ISLOWER, ET. AL. Synopsis #include int int int int int int int int int int int int isalnum (char isalpha (char isascii (char iscntrl (char isdigit (char islower (char isprint (char isgraph (char ispunct (char isspace (char isupper (char isxdigit(char c) c) c) c) c) c) c) c) c) c) c) c) Description These macros, defined in ctype.h, test the supplied character for membership in one of several overlapping groups of characters. Note that all except isascii() are defined for c, if isascii(c) is true or if c = EOF. isalnum(c) isalpha(c) isascii(c) iscntrl(c) isdigit(c) islower(c) isprint(c) isgraph(c) ispunct(c) isspace(c) isupper(c) isxdigit(c) c is in 0-9 or a-z or A-Z c is in A-Z or a-z c is a 7 bit ASCII character c is a control character c is a decimal digit c is in a-z c is a printing char c is a non-space printable character c is not alphanumeric c is a space, tab or newline c is in A-Z c is in 0-9 or a-f or A-F Example #include #include void main (void) { char buf[80]; int i; gets(buf); i = 0; while(isalnum(buf[i])) i++; buf[i] = 0; printf("’%s’ is the word\n", buf); } See Also toupper(), tolower(), toascii() 2012-2013 Microchip Technology Inc. DS50002053D-page 365 MPLAB® XC8 C Compiler User’s Guide ISDIG Synopsis #include int isdig (int c) Description The isdig() function tests the input character c to see if is a decimal digit (0 – 9) and returns true is this is the case; false otherwise. Example #include void main (void) { char buf[] = "1998a"; if(isdig(buf[0])) printf(" type detected\n"); } See Also isdigit() (listed under isalnum()) Return Value Zero if the character is a decimal digit; a non-zero value otherwise. ITOA Synopsis #include char * itoa (char * buf, int val, int base) Description The function itoa converts the contents of val into a string which is stored into buf. The conversion is performed according to the radix specified in base. buf is assumed to reference a buffer which has sufficient space allocated to it. Example #include #include void main (void) { char buf[10]; itoa(buf, 1234, 16); printf("The buffer holds %s\n", buf); } See Also strtol(), utoa(), ltoa(), ultoa() DS50002053D-page 366 2012-2013 Microchip Technology Inc. Library Functions Return Value This routine returns a copy of the buffer into which the result is written. 2012-2013 Microchip Technology Inc. DS50002053D-page 367 MPLAB® XC8 C Compiler User’s Guide LABS Synopsis #include int labs (long int j) Description The labs() function returns the absolute value of long value j. Example #include #include void main (void) { long int a = -5; printf("The absolute value of %ld is %ld\n", a, labs(a)); } See Also abs() Return Value The absolute value of j. LDEXP Synopsis #include double ldexp (double f, int i) Description The ldexp() function performs the inverse of frexp() operation; the integer i is added to the exponent of the floating-point f and the resultant returned. Example #include #include void main (void) { double f; f = ldexp(1.0, 10); printf("1.0 * 2^10 = %f\n", f); } See Also frexp() DS50002053D-page 368 2012-2013 Microchip Technology Inc. Library Functions Return Value The return value is the integer i added to the exponent of the floating-point value f. 2012-2013 Microchip Technology Inc. DS50002053D-page 369 MPLAB® XC8 C Compiler User’s Guide LDIV Synopsis #include ldiv_t ldiv (long number, long denom) Description The ldiv() routine divides the numerator by the denominator, computing the quotient and the remainder. The sign of the quotient is the same as that of the mathematical quotient. Its absolute value is the largest integer which is less than the absolute value of the mathematical quotient. The ldiv() function is similar to the div() function, the difference being that the arguments and the members of the returned structure are all of type long int. Example #include #include void main (void) { ldiv_t lt; lt = ldiv(1234567, 12345); printf("Quotient = %ld, remainder = %ld\n", lt.quot, lt.rem); } See Also div(), uldiv(), udiv() Return Value Returns a structure of type ldiv_t DS50002053D-page 370 2012-2013 Microchip Technology Inc. Library Functions LOCALTIME Synopsis #include struct tm * localtime (time_t * t) Description The localtime() function converts the time pointed to by t which is in seconds since 00:00:00 on Jan 1, 1970, into a broken down time stored in a structure as defined in time.h. The routine localtime() takes into account the contents of the global integer time_zone. This should contain the number of minutes that the local time zone is westward of Greenwich. On systems where it is not possible to predetermine this value, localtime() will return the same result as gmtime(). Example #include #include char * wday[] = { "Sunday", "Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday" }; void main (void) { time_t clock; struct tm * tp; time(&clock); tp = localtime(&clock); printf("Today is %s\n", wday[tp->tm_wday]); } See Also ctime(), asctime(), time() Return Value Returns a structure of type tm. Note The example will require the user to provide the time() routine as one cannot be supplied with the compiler. See time() for more detail. 2012-2013 Microchip Technology Inc. DS50002053D-page 371 MPLAB® XC8 C Compiler User’s Guide LOG, LOG10 Synopsis #include double log (double f) double log10 (double f) Description The log() function returns the natural logarithm of f. The function log10() returns the logarithm to base 10 of f. Example #include #include void main (void) { double f; for(f = 1.0 ; f <= 10.0 ; f += 1.0) printf("log(%1.0f) = %f\n", f, log(f)); } See Also exp(), pow() Return Value Zero if the argument is negative. DS50002053D-page 372 2012-2013 Microchip Technology Inc. Library Functions LONGJMP Synopsis #include void longjmp (jmp_buf buf, int val) Description The longjmp() function, in conjunction with setjmp(), provides a mechanism for non-local goto’s. To use this facility, setjmp() should be called with a jmp_buf argument in some outer level function. The call from setjmp() will return 0. To return to this level of execution, longjmp() can be called with the same jmp_buf argument from an inner level of execution. However, the function that called setjmp() must still be active when longjmp() is called. Breach of this rule will cause errors, due to the use of a stack containing invalid data. The val argument to longjmp() will be the value apparently returned from the setjmp(). This should normally be non-zero, to distinguish it from the genuine setjmp() call. Example #include #include #include jmp_buf jb; void inner (void) { longjmp(jb, 5); } void main (void) { int i; if(i = setjmp(jb)) { printf("setjmp returned %d\n" i); exit(0); } printf("setjmp returned 0 - good\n"); printf("calling inner...\n"); inner(); printf("inner returned - bad!\n"); } See Also setjmp() Return Value The longjmp() routine never returns. Note The function which called setjmp() must still be active when longjmp() is called. Breach of this rule will cause disaster, due to the use of a stack containing invalid data. 2012-2013 Microchip Technology Inc. DS50002053D-page 373 MPLAB® XC8 C Compiler User’s Guide LTOA Synopsis #include char * ltoa (char * buf, long val, int base) Description The function ltoa converts the contents of val into a string which is stored into buf. The conversion is performed according to the radix specified in base. buf is assumed to reference a buffer which has sufficient space allocated to it. Example #include #include void main (void) { char buf[10]; ltoa(buf, 12345678L, 16); printf("The buffer holds %s\n", buf); } See Also strtol(), itoa(), utoa(), ultoa() Return Value This routine returns a copy of the buffer into which the result is written. DS50002053D-page 374 2012-2013 Microchip Technology Inc. Library Functions MEMCHR Synopsis #include void * memchr (const void * block, int val, size_t length) Description The memchr() function is similar to strchr() except that instead of searching null-terminated strings, it searches a block of memory specified by length for a particular byte. Its arguments are a pointer to the memory to be searched, the value of the byte to be searched for, and the length of the block. A pointer to the first occurrence of that byte in the block is returned. Example #include #include unsigned int ary[] = {1, 5, 0x6789, 0x23}; void main (void) { char * cp; cp = memchr(ary, 0x89, sizeof ary); if(!cp) printf("Not found\n"); else printf("Found at offset %u\n", cp - (char *)ary); } See Also strchr() Return Value A pointer to the first byte matching the argument if one exists; NULL otherwise. 2012-2013 Microchip Technology Inc. DS50002053D-page 375 MPLAB® XC8 C Compiler User’s Guide MEMCMP Synopsis #include int memcmp (const void * s1, const void * s2, size_t n) Description The memcmp() function compares two blocks of memory, of length n, and returns a signed value similar to strncmp(). Unlike strncmp() the comparison does not stop on a null character. Example #include #include void main (void) { int buf[10], cow[10], i; buf[0] = 1; buf[2] = 4; cow[0] = 1; cow[2] = 5; buf[1] = 3; cow[1] = 3; i = memcmp(buf, cow, 3*sizeof(int)); if(i < 0) printf("Less than\n"); else if(i > 0) printf("Greater than\n"); else printf("Equal\n"); } See Also strncpy(), strncmp(), strchr(), memset(), memchr() Return Value Returns negative one, zero or one, depending on whether s1 points to string which is less than, equal to or greater than the string pointed to by s2 in the collating sequence. DS50002053D-page 376 2012-2013 Microchip Technology Inc. Library Functions MEMCPY Synopsis #include void * memcpy (void * d, const void * s, size_t n) Description The memcpy() function copies n bytes of memory starting from the location pointed to by s to the block of memory pointed to by d. The result of copying overlapping blocks is undefined. The memcpy() function differs from strcpy() in that it copies a specified number of bytes, rather than all bytes up to a null terminator. Example #include #include void main (void) { char buf[80]; memset(buf, 0, sizeof buf); memcpy(buf, "A partial string", 10); printf("buf = ’%s’\n", buf); } See Also strncpy(), strncmp(), strchr(), memset() Return Value The memcpy() routine returns its first argument. 2012-2013 Microchip Technology Inc. DS50002053D-page 377 MPLAB® XC8 C Compiler User’s Guide MEMMOVE Synopsis #include void * memmove (void * s1, const void * s2, size_t n) Description The memmove() function is similar to the function memcpy() except copying of overlapping blocks is handled correctly. That is, it will copy forwards or backwards as appropriate to correctly copy one block to another that overlaps it. See Also strncpy(), strncmp(), strchr(), memcpy() Return Value The function memmove() returns its first argument. MEMSET Synopsis #include void * memset (void * s, int c, size_t n) Description The memset() function fills n bytes of memory starting at the location pointed to by s with the byte c. Example #include #include void main (void) { char abuf[20]; strcpy(abuf, "This is a string"; memset(abuf, ’x’, 5); printf("buf = ’%s’\n", abuf); } See Also strncpy(), strncmp(), strchr(), memcpy(), memchr() DS50002053D-page 378 2012-2013 Microchip Technology Inc. Library Functions MKTIME Synopsis #include time_t mktime (struct tm * tmptr) Description The mktime() function converts and returns the local calendar time referenced by the tm structure tmptr into a time being the number of seconds passed since Jan 1, 1970, or returns -1 if the time cannot be represented. Example #include #include void main (void) { struct tm birthday; birthday.tm_year = 83; // the 5th of can 1983 birthday.tm_mon = 5; birthday.tm_mday = 5; birthday.tm_hour = birthday.tm_min = birthday.tm_sec = 0; printf("you were born approximately %ld seconds after the unix epoch\n", mktime(&birthday)); } See Also ctime(), asctime() Return Value The time contained in the tm structure represented as the number of seconds since the 1970 Epoch, or -1 if this time cannot be represented. 2012-2013 Microchip Technology Inc. DS50002053D-page 379 MPLAB® XC8 C Compiler User’s Guide MODF Synopsis #include double modf (double value, double * iptr) Description The modf() function splits the argument value into integral and fractional parts, each having the same sign as value. For example, -3.17 would be split into the integral part (-3) and the fractional part (-0.17). The integral part is stored as a double in the object pointed to by iptr. Example #include #include void main (void) { double i_val, f_val; f_val = modf( -3.17, &i_val); } Return Value The signed fractional part of value. NOP Synopsis #include NOP(); Description Execute NOP instruction here. This is often useful to fine tune delays or create a handle for breakpoints. The NOP instruction is sometimes required during some sensitive sequences in hardware. Example #include void crude_delay(unsigned char x) { while(x--){ NOP(); /* Do nothing for 3 cycles */ NOP(); NOP(); } } DS50002053D-page 380 2012-2013 Microchip Technology Inc. Library Functions POW Synopsis #include double pow (double f, double p) Description The pow() function raises its first argument, f, to the power p. Example #include #include void main (void) { double f; for(f = 1.0 ; f <= 10.0 ; f += 1.0) printf("pow(2, %1.0f) = %f\n", f, pow(2, f)); } See Also log(), log10(), exp() Return Value f to the power of p. 2012-2013 Microchip Technology Inc. DS50002053D-page 381 MPLAB® XC8 C Compiler User’s Guide PRINTF, VPRINTF Synopsis #include int printf (const char * fmt, ...) #include #include int vprintf (const char * fmt, va_list va_arg) Description The printf() function is a formatted output routine, operating on stdout. It relies on the putch() function to determine the destination of stdout. The putch() function must be written as part of each project, as well as code to initialize any peripherals used by this routine. A stub for putch can be found in the sources directory of the compiler. A typical putch routine to send one byte to the USART can need to be written similar to the following. void putch(char data) { while( ! TXIF) continue; TXREG = data; } Include the source file for putch into your project once it is complete. The printf() routine is passed a format string, followed by a list of zero or more arguments. In the format string are conversion specifications, each of which is used to print out one of the argument list values. Each conversion specification is of the form %fm.nc where the percent symbol % introduces a conversion, followed by zero or more flags, f, (in any order), and followed by an optional width specification m. The n specification is an optional precision specification (introduced by the dot) and c is a letter specifying the type of the conversion. The flags may consist of the following. • A minus sign (’-’), which indicates left rather than right adjustment of the converted value in the field. Where the field width is larger than required for the conversion, blank padding is performed at the left or right as specified. • A plus sign (’+’), which indicates that the sign of the converted value will always be printed, even if the value is positive. • A space (’ ’), which will prefix a space to the converted result if it does not contain a sign (hence this has no effect if the + flag is specified) or if a signed conversion results in no characters. • The digit zero (’0’), which indicates that any padding will be performed with zeros rather than blanks. This flag is ignored if the ’-’ flag has been specified. • A hash character (’#’), which indicates that an alternate format is to be used. The nature of the alternate format is discussed below. Not all formats have alternates. In those cases, the presence of the hash character has no effect. If the character * is used in place of a decimal constant, e.g., in the format %*d, then one integer argument will be taken from the list to provide that value. DS50002053D-page 382 2012-2013 Microchip Technology Inc. Library Functions The types of conversion are: f Floating point - m is the total width and n is the number of digits after the decimal point. If n is omitted it defaults to 6. If the precision is zero, the decimal point will be omitted unless the alternate format is specified. e Print the corresponding argument in scientific notation. Otherwise similar to f. g Use e or f format, whichever gives maximum precision in minimum width. If the alternate format is not specified, any trailing zeros after the decimal point will be removed, and if no digits remain after the decimal point, it will also be removed. o x X u d Integer conversion - in radices 8, 16, 16, 10 and 10 respectively. The conversion is signed in the case of d, unsigned otherwise. The precision value is the total number of digits to print, and can be used to force leading zeros. For example, %8.4x will print at least 4 HEX digits in an 8 wide field. Preceding the key letter with an l indicates that the value argument is a long integer. The letter X prints out hexadecimal numbers using the upper case letters A-F rather than a- f as would be printed when using x. When the alternate format is specified, a leading zero will be supplied for the octal format, and a leading 0x or 0X for the HEX format. s Print a string - the value argument is assumed to be a character. At most n characters from the string will be printed, in a field m characters wide. c The argument is assumed to be a single character and is printed literally. Any other characters used as conversion specifications will be printed. Thus % will produce a single percent sign. The vprintf() function is similar to printf() but takes a variable argument list rather than a list of arguments. See the description of va_start() for more information on variable argument lists. An example of using vprintf() follows. 2012-2013 Microchip Technology Inc. DS50002053D-page 383 MPLAB® XC8 C Compiler User’s Guide Example printf("Total = %4d%", 23); yields ’Total = 23%’ printf("Size is %lx", size); where size is a long, prints size as hexadecimal. printf("Name = %.8s", "a1234567890"); yields ‘Name = a1234567’ printf("xx%*d", 3, 4); yields ‘xx 4’ /* vprintf example */ #include int error (char * s, ...) { va_list ap; va_start(ap, s); printf("Error: "); vprintf(s, ap); putchar(’\n’); va_end(ap); } void main (void) { int i; i = 3; error("testing 1 2 %d", i); } See Also sprintf() Return Value The printf() and vprintf() functions return the number of characters written to stdout. DS50002053D-page 384 2012-2013 Microchip Technology Inc. Library Functions PUTCH Synopsis #include void putch (char c) Description The putch() function is provided as an empty stub which can be completed as each project requires. Typically this function will accept one byte of data and send this to a peripheral which is associated with stdout. Example #include char * x = "This is a string"; void main (void) { char * cp; cp = x; while(*x) putch(*x++); putch(’\n’); } See Also printf(), putchar() 2012-2013 Microchip Technology Inc. DS50002053D-page 385 MPLAB® XC8 C Compiler User’s Guide PUTCHAR Synopsis #include int putchar (int c) Description The putchar() function calls putch() to print one character to stdout, and is defined in stdio.h. Example #include char * x = "This is a string"; void main (void) { char * cp; cp = x; while(*x) putchar(*x++); putchar(’\n’); } See Also putc(), getc(), freopen(), fclose() Return Value The character passed as argument, or EOF if an error occurred. DS50002053D-page 386 2012-2013 Microchip Technology Inc. Library Functions PUTS Synopsis #include int puts (const char * s) Description The puts() function writes the string s to the stdout stream, appending a newline. The null character terminating the string is not copied. Example #include void main (void) { puts("Hello, world!"); } See Also fputs(), gets(), freopen(), fclose() Return Value EOF is returned on error; zero otherwise. 2012-2013 Microchip Technology Inc. DS50002053D-page 387 MPLAB® XC8 C Compiler User’s Guide QSORT Synopsis #include void qsort (void * base, size_t nel, size_t width, int (*func)(const void *, const void *)) Description The qsort() function is an implementation of the quicksort algorithm. It sorts an array of nel items, each of length width bytes, located contiguously in memory at base. The argument func is a pointer to a function used by qsort() to compare items. It calls func with s to two items to be compared. If the first item is considered to be greater than, equal to or less than the second then func should return a value greater than zero, equal to zero or less than zero respectively. Example #include #include int array[] = { 567, 23, 456, 1024, 17, 567, 66 }; int sortem (const void * p1, const void * p2) { return *(int *)p1 - *(int *)p2; } void main (void) { register int i; qsort(aray, sizeof array/sizeof array[0], sizeof array[0], sortem); for(i = 0 ; i != sizeof array/sizeof array[0] ; i++) printf("%d\t", array[i]); putchar(’\n’); } Note The function parameter must be a pointer to a function of type similar to: int func (const void *, const void *) For example, it must accept two const void * parameters, and must be prototyped. DS50002053D-page 388 2012-2013 Microchip Technology Inc. Library Functions RAND Synopsis #include int rand (void) Description The rand() function is a pseudo-random number generator. It returns an integer in the range 0 to 32767, which changes in a pseudo-random fashion on each call. The algorithm will produce a deterministic sequence if started from the same point. The starting point is set using the srand() call. The example shows use of the time() function to generate a different starting point for the sequence each time. Example #include #include #include void main (void) { time_t toc; int i; time(&toc); srand((int)toc); for(i = 0 ; i != 10 ; i++) printf("%d\t", rand()); putchar(’\n’); } See Also srand() Note The example will require the user to provide the time() routine as one cannot be supplied with the compiler. See time() for more detail. 2012-2013 Microchip Technology Inc. DS50002053D-page 389 MPLAB® XC8 C Compiler User’s Guide ROUND Synopsis #include double round (double x) Description The round function round the argument to the nearest integer value, but in floating-point format. Values midway between integer values are rounded up. Example #include void main (void) { double input, rounded; input = 1234.5678; rounded = round(input); } See Also trunc() DS50002053D-page 390 2012-2013 Microchip Technology Inc. Library Functions SETJMP Synopsis #include int setjmp (jmp_buf buf) Description The setjmp() function is used with longjmp() for non-local goto’s. See longjmp() for further information. Example #include #include #include