ARM® Compiler V5.06 For µVision® ARM C And C++ Libraries Floating Point Support User Guide DUI0378G 02 Mdk
User Manual:
Open the PDF directly: View PDF 
.
Page Count: 218
| Download | |
| Open PDF In Browser | View PDF |
ARM® Compiler v5.06 for µVision® Version 5 ARM C and C++ Libraries and Floating-Point Support User Guide Confidential - Draft - Beta Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. ARM DUI0378G_02 ARM® Compiler v5.06 for µVision® ARM® Compiler v5.06 for µVision® ARM C and C++ Libraries and Floating-Point Support User Guide Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Release Information Document History Issue Date Confidentiality Change A May 2007 Non-Confidential Release for RVCT v3.1 Release for µVision B December 2008 Non-Confidential Release for RVCT v4.0 for µVision C June 2011 Non-Confidential Release for ARM Compiler v4.1 for µVision D July 2012 Non-Confidential Release for ARM Compiler v5.02 for µVision E 30 May 2014 Non-Confidential Release for ARM Compiler v5.04 for µVision F 12 December 2014 Non-Confidential Release for ARM Compiler v5.05 for µVision G-02 15 August 2015 Confidential - Draft Release for ARM Compiler v5.06 for µVision Confidential Proprietary Notice This document is CONFIDENTIAL and any use by you is subject to the terms of the agreement between you and ARM or the terms of the agreement between you and the party authorised by ARM to disclose this document to you. This document is protected by copyright and other related rights and the practice or implementation of the information contained in this document may be protected by one or more patents or pending patent applications. No part of this document may be reproduced in any form by any means without the express prior written permission of ARM. No license, express or implied, by estoppel or otherwise to any intellectual property rights is granted by this document unless specifically stated. Your access to the information in this document is conditional upon your acceptance that you will not use or permit others to use the information: (i) for the purposes of determining whether implementations infringe any third party patents; (ii) for developing technology or products which avoid any of ARM’s intellectual property; or (iii) as a reference for modifying existing patents or patent applications or creating any continuation, continuation in part, or extension of existing patents or patent applications; or (iv) for generating data for publication or disclosure to third parties, which compares the performance or functionality of the ARM technology described in this document with any other products created by you or a third party, without obtaining ARM’s prior written consent. THIS DOCUMENT IS PROVIDED “AS IS”. ARM PROVIDES NO REPRESENTATIONS AND NO WARRANTIES, EXPRESS, IMPLIED OR STATUTORY, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF MERCHANTABILITY, SATISFACTORY QUALITY, NON-INFRINGEMENT OR FITNESS FOR A PARTICULAR PURPOSE WITH RESPECT TO THE DOCUMENT. For the avoidance of doubt, ARM makes no representation with respect to, and has undertaken no analysis to identify or understand the scope and content of, third party patents, copyrights, trade secrets, or other rights. This document may include technical inaccuracies or typographical errors. TO THE EXTENT NOT PROHIBITED BY LAW, IN NO EVENT WILL ARM BE LIABLE FOR ANY DAMAGES, INCLUDING WITHOUT LIMITATION ANY DIRECT, INDIRECT, SPECIAL, INCIDENTAL, PUNITIVE, OR CONSEQUENTIAL DAMAGES, HOWEVER CAUSED AND REGARDLESS OF THE THEORY OF LIABILITY, ARISING OUT OF ANY USE OF THIS DOCUMENT, EVEN IF ARM HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. This document consists solely of commercial items. You shall be responsible for ensuring that any use, duplication or disclosure of this document complies fully with any relevant export laws and regulations to assure that this document or any portion thereof is not exported, directly or indirectly, in violation of such export laws. Use of the word “partner” in reference to ARM’s customers is not intended to create or refer to any partnership relationship with any other company. ARM may make changes to this document at any time and without notice. If any of the provisions contained in these terms conflict with any of the provisions of any signed written agreement covering this document with ARM, then the signed written agreement prevails over and supersedes the conflicting provisions of these terms. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2 ARM® Compiler v5.06 for µVision® This document may be translated into other languages for convenience, and you agree that if there is any conflict between the English version of this document and any translation, the terms of the English version of the Agreement shall prevail. Words and logos marked with ® or ™ are registered trademarks or trademarks of ARM Limited or its affiliates in the EU and/or elsewhere. All rights reserved. Other brands and names mentioned in this document may be the trademarks of their respective owners. Please follow ARM’s trademark usage guidelines at http://www.arm.com/about/trademarks/guidelines/index.php Copyright © [2007, 2008, 2011, 2012, 2014, 2015], ARM Limited or its affiliates. All rights reserved. ARM Limited. Company 02557590 registered in England. 110 Fulbourn Road, Cambridge, England CB1 9NJ. LES-PRE-20348 Additional Notices Some material in this document is based on IEEE 754-1985 IEEE Standard for Binary Floating-Point Arithmetic. The IEEE disclaims any responsibility or liability resulting from the placement and use in the described manner. Confidentiality Status This document is Confidential. This document may only be used and distributed in accordance with the terms of the agreement entered into by ARM and the party that ARM delivered this document to. Product Status The information in this document is for a Beta product, that is a product under development. Web Address http://www.arm.com ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3 Contents ARM® Compiler v5.06 for µVision® ARM C and C++ Libraries and Floating-Point Support User Guide Preface About this book ..................................................... ..................................................... 12 Chapter 1 The ARM C and C++ Libraries 1.1 1.2 1.3 1.4 Mandatory linkage with the C library ................................... ................................... C and C++ runtime libraries .......................................... .......................................... C and C++ library features ........................................... ........................................... C++ and C libraries and the std namespace .......................................................... 1.5 1.6 1.7 1.8 1.9 1.10 Multithreaded support in ARM C libraries ................................................................ 1-24 Support for building an application with the C library ....................... ....................... 1-34 Support for building an application without the C library .................... .................... 1-40 Tailoring the C library to a new execution environment ..................... ..................... 1-47 Assembler macros that tailor locale functions in the C library ................ ................ 1-52 Modification of C library functions for error signaling, error handling, and program exit .. .................................................................................................................................. 1-61 Stack and heap memory allocation and the ARM C and C++ libraries ......... ......... 1-62 Tailoring input/output functions in the C and C++ libraries ...................................... 1-69 Target dependencies on low-level functions in the C and C++ libraries .................. 1-70 The C library printf family of functions .................................. .................................. 1-72 The C library scanf family of functions .................................. .................................. 1-73 Redefining low-level library functions to enable direct use of high-level library functions in the C library .................................................... .................................................... 1-74 1.11 1.12 1.13 1.14 1.15 1.16 ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-16 1-17 1-22 1-23 4 1.17 The C library functions fread(), fgets() and gets() .................................................... 1-76 1.18 1.19 1.20 1.21 1.22 1.23 1.24 1.25 1.26 1.27 Re-implementing __backspace() in the C library .......................... .......................... Re-implementing __backspacewc() in the C library ................................................ Redefining target-dependent system I/O functions in the C library ............ ............ Tailoring non-input/output C library functions ............................. ............................. Real-time integer division in the ARM libraries ........................................................ ISO C library implementation definition ................................. ................................. C library functions and extensions ..................................... ..................................... Compiler generated and library-resident helper functions ................... ................... C and C++ library naming conventions ................................. ................................. Using macro__ARM_WCHAR_NO_IO to disable FILE declaration and wide I/O function prototypes .................................................................................................. Using library functions with execute-only memory ......................... ......................... 1.28 Chapter 2 About floating-point support ......................................... ......................................... The software floating-point library, fplib ................................ ................................ Controlling the ARM floating-point environment .................................................... Using C99 signaling NaNs provided by mathlib (_WANT_SNAN) ............ ............ mathlib double and single-precision floating-point functions ................ ................ IEEE 754 arithmetic ............................................... ............................................... Using the Vector Floating-Point (VFP) support libraries ........................................ 3-108 3-109 3-115 3-127 3-128 3-129 3-137 The C and C++ Library Functions reference 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10 4.11 4.12 4.13 4.14 4.15 ARM DUI0378G_02 About microlib .......................................................................................................... 2-97 Differences between microlib and the default C library ..................... ..................... 2-98 Library heap usage requirements of microlib ............................ ............................ 2-100 ISO C features missing from microlib .................................................................... 2-101 Building an application with microlib ...................................................................... 2-103 Configuring the stack and heap for use with microlib ............................................ 2-104 Entering and exiting programs linked with microlib ....................... ....................... 2-105 Tailoring the microlib input/output functions ............................. ............................. 2-106 Floating-point Support 3.1 3.2 3.3 3.4 3.5 3.6 3.7 Chapter 4 1-94 1-95 The ARM C Micro-library 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 Chapter 3 1-77 1-78 1-79 1-81 1-82 1-83 1-89 1-90 1-91 __aeabi_errno_addr() ............................................................................................ alloca() ......................................................... ......................................................... clock() .................................................................................................................... _clock_init() ..................................................... ..................................................... __default_signal_handler() .................................................................................... errno ...................................................................................................................... _findlocale() ..................................................... ..................................................... _fisatty() ........................................................ ........................................................ _get_lconv() ..................................................... ..................................................... getenv() ........................................................ ........................................................ _getenv_init() .................................................... .................................................... __heapstats() .................................................... .................................................... __heapvalid() .................................................... .................................................... lconv structure ................................................... ................................................... localeconv() ..................................................... ..................................................... Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-140 4-141 4-142 4-143 4-144 4-145 4-146 4-147 4-148 4-149 4-150 4-151 4-152 4-153 4-155 5 ARM DUI0378G_02 4.16 _membitcpybl(), _membitcpybb(), _membitcpyhl(), _membitcpyhb(), _membitcpywl(), 4.17 4.18 4.19 4.20 4.21 4.22 4.23 4.24 4.25 4.26 4.27 4.28 4.29 4.30 4.31 4.32 4.33 4.34 4.35 4.36 4.37 4.38 4.39 4.40 4.41 4.42 4.43 4.44 4.45 4.46 4.47 4.48 4.49 4.50 4.51 4.52 4.53 4.54 4.55 4.56 4.57 4.58 4.59 4.60 4.61 _membitcpywb(), _membitmovebl(), _membitmovebb(), _membitmovehl(), _membitmovehb(), _membitmovewl(), _membitmovewb() .................................... 4-156 posix_memalign() .................................................................................................. 4-157 #pragma import(_main_redirection) ................................... ................................... 4-158 __raise() ........................................................ ........................................................ 4-159 _rand_r() ................................................................................................................ 4-160 remove() ................................................................................................................ 4-161 rename() ................................................................................................................ 4-162 __rt_entry ....................................................... ....................................................... 4-163 __rt_errno_addr() ................................................. ................................................. 4-164 __rt_exit() ....................................................... ....................................................... 4-165 __rt_fp_status_addr() ............................................................................................ 4-166 __rt_heap_extend() ............................................... ............................................... 4-167 __rt_lib_init() .......................................................................................................... 4-168 __rt_lib_shutdown() ............................................... ............................................... 4-169 __rt_raise() ............................................................................................................ 4-170 __rt_stackheap_init() .............................................. .............................................. 4-171 setlocale() .............................................................................................................. 4-172 _srand_r() .............................................................................................................. 4-174 strcasecmp() .......................................................................................................... 4-175 strncasecmp() ........................................................................................................ 4-176 strlcat() ......................................................... ......................................................... 4-177 strlcpy() .................................................................................................................. 4-178 _sys_close() ..................................................... ..................................................... 4-179 _sys_command_string() ........................................................................................ 4-180 _sys_ensure() ........................................................................................................ 4-181 _sys_exit() ...................................................... ...................................................... 4-182 _sys_flen() ...................................................... ...................................................... 4-183 _sys_istty() ...................................................... ...................................................... 4-184 _sys_open() ..................................................... ..................................................... 4-185 _sys_read() ............................................................................................................ 4-186 _sys_seek() ..................................................... ..................................................... 4-187 _sys_tmpnam() ...................................................................................................... 4-188 _sys_write() ..................................................... ..................................................... 4-189 system() ........................................................ ........................................................ 4-190 time() .......................................................... .......................................................... 4-191 _ttywrch() ....................................................... ....................................................... 4-192 __user_heap_extend() ............................................. ............................................. 4-193 __user_heap_extent() ............................................. ............................................. 4-194 __user_setup_stackheap() .................................................................................... 4-195 __vectab_stack_and_reset .................................................................................... 4-196 wcscasecmp() ........................................................................................................ 4-197 wcsncasecmp() ...................................................................................................... 4-198 wcstombs() ............................................................................................................ 4-199 Thread-safe C library functions ...................................... ...................................... 4-200 C library functions that are not thread-safe ............................. ............................. 4-202 Legacy function __user_initial_stackheap() ............................. ............................. 4-204 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 6 Chapter 5 ARM DUI0378G_02 Floating-point Support Functions Reference 5.1 _clearfp() ....................................................... ....................................................... 5-206 5.2 5.3 5.4 5.5 5.6 5.7 5.8 5.9 _controlfp() ...................................................... ...................................................... 5-207 __fp_status() .......................................................................................................... 5-209 gamma(), gamma_r() .............................................. .............................................. 5-211 __ieee_status() ...................................................................................................... 5-212 j0(), j1(), jn(), Bessel functions of the first kind ........................... ........................... 5-215 significand(), fractional part of a number ............................... ............................... 5-216 _statusfp() .............................................................................................................. 5-217 y0(), y1(), yn(), Bessel functions of the second kind .............................................. 5-218 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 7 List of Figures ARM® Compiler v5.06 for µVision® ARM C and C++ Libraries and Floating-Point Support User Guide Figure 3-1 Figure 3-2 Figure 5-1 Figure 5-2 ARM DUI0378G_02 IEEE 754 single-precision floating-point format ................................................................... IEEE 754 double-precision floating-point format ................................................................. Floating-point status word layout ......................................................................................... IEEE status word layout ...................................................................................................... Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-129 3-130 5-209 5-212 8 List of Tables ARM® Compiler v5.06 for µVision® ARM C and C++ Libraries and Floating-Point Support User Guide Table 1-1 Table 1-2 Table 1-3 Table 1-4 Table 1-5 Table 1-6 Table 1-7 Table 1-8 Table 1-9 Table 1-10 Table 1-11 Table 1-12 Table 1-13 Table 3-1 Table 3-2 Table 3-3 Table 3-4 Table 3-5 Table 3-6 Table 3-7 Table 3-8 Table 3-9 ARM DUI0378G_02 C library callouts .................................................................................................................... 1-26 Direct semihosting dependencies .......................................................................................... 1-37 Indirect semihosting dependencies ....................................................................................... 1-38 Published API definitions ....................................................................................................... 1-38 Standalone C library functions ............................................................................................... 1-40 Default ISO8859-1 locales ..................................................................................................... 1-52 Default Shift-JIS and UTF-8 locales ...................................................................................... 1-53 Trap and error handling ......................................................................................................... 1-61 Input/output dependencies .................................................................................................... 1-70 Signals supported by the signal() function ............................................................................. 1-84 perror() messages ................................................................................................................. 1-86 Standard C++ library differences ........................................................................................... 1-87 C library extensions ............................................................................................................... 1-89 Arithmetic routines ............................................................................................................... 3-110 Number format conversion routines ..................................................................................... 3-111 Floating-point comparison routines ...................................................................................... 3-112 fplib C99 functions ............................................................................................................... 3-113 FE_EX_INTYPE_MASK operand type flags ....................................................................... 3-122 FE_EX_OUTTYPE_MASK operand type flags ................................................................... 3-122 FE_EX_FN_MASK operation type flags .............................................................................. 3-123 FE_EX_CMPRET_MASK comparison type flags ................................................................ 3-123 Sample single-precision floating-point values ..................................................................... 3-131 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 9 Table 3-10 Sample double-precision floating-point values .................................................................... 3-132 Table 4-1 Table 4-2 Table 5-1 Table 5-2 Table 5-3 Functions that are thread-safe ............................................................................................. Functions that are not thread-safe ....................................................................................... _controlfp argument macros ............................................................................................... Status word bit modification ................................................................................................. Rounding mode control ....................................................................................................... ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-200 4-202 5-207 5-212 5-213 10 Preface This preface introduces the ARM® Compiler v5.06 for µVision® ARM C and C++ Libraries and FloatingPoint Support User Guide. It contains the following: • About this book on page 12. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 11 Preface About this book About this book ARM® Compiler for µVision® ARM C and C++ Libraries and Floating-Point Support User Guide. This manual provides user information for the ARM libraries and floating-point support. It is also available as a PDF. Using this book This book is organized into the following chapters: Chapter 1 The ARM C and C++ Libraries Describes the ARM® C and C++ libraries. Chapter 2 The ARM C Micro-library Describes microlib, the C micro-library. Chapter 3 Floating-point Support Describes ARM support for floating-point computations. Chapter 4 The C and C++ Library Functions reference Describes the standard C and C++ library functions that are extensions to the C Standard or that differ in some way to the standard. Chapter 5 Floating-point Support Functions Reference Describes ARM support for floating-point functions. Glossary The ARM Glossary is a list of terms used in ARM documentation, together with definitions for those terms. The ARM Glossary does not contain terms that are industry standard unless the ARM meaning differs from the generally accepted meaning. See the ARM Glossary for more information. Typographic conventions italic Introduces special terminology, denotes cross-references, and citations. bold Highlights interface elements, such as menu names. Denotes signal names. Also used for terms in descriptive lists, where appropriate. monospace Denotes text that you can enter at the keyboard, such as commands, file and program names, and source code. monospace Denotes a permitted abbreviation for a command or option. You can enter the underlined text instead of the full command or option name. monospace italic Denotes arguments to monospace text where the argument is to be replaced by a specific value. monospace bold Denotes language keywords when used outside example code.Encloses replaceable terms for assembler syntax where they appear in code or code fragments. For example: MRC p15, 0, , , , SMALL CAPITALS Used in body text for a few terms that have specific technical meanings, that are defined in the ARM glossary. For example, IMPLEMENTATION DEFINED, IMPLEMENTATION SPECIFIC, UNKNOWN, and UNPREDICTABLE. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 12 Preface About this book Feedback Feedback on this product If you have any comments or suggestions about this product, contact your supplier and give: • The product name. • The product revision or version. • An explanation with as much information as you can provide. Include symptoms and diagnostic procedures if appropriate. Feedback on content If you have comments on content then send an e-mail to errata@arm.com. Give: • • • • The title ARM® Compiler v5.06 for µVision® ARM C and C++ Libraries and Floating-Point Support User Guide. The number ARM DUI0378G_02. If applicable, the page number(s) to which your comments refer. A concise explanation of your comments. ARM also welcomes general suggestions for additions and improvements. Note ARM tests the PDF only in Adobe Acrobat and Acrobat Reader, and cannot guarantee the quality of the represented document when used with any other PDF reader. Other information • • • • ARM DUI0378G_02 ARM Information Center. ARM Technical Support Knowledge Articles. Support and Maintenance. ARM Glossary. Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 13 Chapter 1 The ARM C and C++ Libraries Describes the ARM® C and C++ libraries. It contains the following sections: • 1.1 Mandatory linkage with the C library on page 1-16. • 1.2 C and C++ runtime libraries on page 1-17. • 1.3 C and C++ library features on page 1-22. • 1.4 C++ and C libraries and the std namespace on page 1-23. • 1.5 Multithreaded support in ARM C libraries on page 1-24. • 1.6 Support for building an application with the C library on page 1-34. • 1.7 Support for building an application without the C library on page 1-40. • 1.8 Tailoring the C library to a new execution environment on page 1-47. • 1.9 Assembler macros that tailor locale functions in the C library on page 1-52. • 1.10 Modification of C library functions for error signaling, error handling, and program exit on page 1-61. • 1.11 Stack and heap memory allocation and the ARM C and C++ libraries on page 1-62. • 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. • 1.13 Target dependencies on low-level functions in the C and C++ libraries on page 1-70. • 1.14 The C library printf family of functions on page 1-72. • 1.15 The C library scanf family of functions on page 1-73. • 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library on page 1-74. • 1.17 The C library functions fread(), fgets() and gets() on page 1-76. • 1.18 Re-implementing __backspace() in the C library on page 1-77. • 1.19 Re-implementing __backspacewc() in the C library on page 1-78. • 1.20 Redefining target-dependent system I/O functions in the C library on page 1-79. • 1.21 Tailoring non-input/output C library functions on page 1-81. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-14 1 The ARM C and C++ Libraries • • • • • • • ARM DUI0378G_02 1.22 Real-time integer division in the ARM libraries on page 1-82. 1.23 ISO C library implementation definition on page 1-83. 1.24 C library functions and extensions on page 1-89. 1.25 Compiler generated and library-resident helper functions on page 1-90. 1.26 C and C++ library naming conventions on page 1-91. 1.27 Using macro__ARM_WCHAR_NO_IO to disable FILE declaration and wide I/O function prototypes on page 1-94. 1.28 Using library functions with execute-only memory on page 1-95. Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-15 1 The ARM C and C++ Libraries 1.1 Mandatory linkage with the C library 1.1 Mandatory linkage with the C library If you write an application in C, you must link it with the C library, even if it makes no direct use of C library functions. This is because the compiler might implicitly generate calls to C library functions to improve your application, even though calls to such functions might not exist in your source code. Even if your application does not have a main() function, meaning that the C library is not initialized, some C library functions are still legitimately available and the compiler might implicitly generate calls to these functions. Related concepts 1.2.1 Summary of the C and C++ runtime libraries on page 1-17. 1.6.4 Using the libraries in a nonsemihosting environment on page 1-35. Related references 1.7.1 Building an application without the C library on page 1-40. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-16 1 The ARM C and C++ Libraries 1.2 C and C++ runtime libraries 1.2 C and C++ runtime libraries ARM provides the C standardlib, C microlib, and C++ runtime libraries to support compiled C and C++. This section contains the following subsections: • 1.2.1 Summary of the C and C++ runtime libraries on page 1-17. • 1.2.2 Compliance with the Application Binary Interface (ABI) for the ARM architecture on page 1-18. • 1.2.3 Increasing portability of object files to other CLIBABI implementations on page 1-18. • 1.2.4 ARM C and C++ library directory structure on page 1-18. • 1.2.5 Selection of ARM C and C++ library variants based on build options on page 1-19. • 1.2.6 Thumb C libraries on page 1-20. 1.2.1 Summary of the C and C++ runtime libraries A summary of the C and C++ runtime libraries provided by ARM. C standardlib This is a C library consisting of: • All functions defined by the ISO C99 library standard. • Target-dependent functions that implement the C library functions in the semihosted execution environment. You can redefine these functions in your own application. • Functions called implicitly by the compiler. • ARM extensions that are not defined by the ISO C library standard, but are included in the library. C microlib This is a C library that can be used as an alternative to C standardlib. It is a micro-library that is ideally suited for deeply embedded applications that have to fit within small-sized memory. The C micro-library, microlib, consists of: • Functions that are highly optimized to achieve the minimum code size. • Functions that are not compliant with the ISO C library standard. • Functions that are not compliant with the 1985 IEEE 754 standard for binary floating-point arithmetic. C++ This is a C++ library that can be used with C standardlib. It consists of: • Functions defined by the ISO C++ library standard. • The Rogue Wave Standard C++ library. • Additional C++ functions not supported by the Rogue Wave library. The C++ libraries depend on the C library for target-specific support. There are no target dependencies in the C++ libraries. Related concepts 1.1 Mandatory linkage with the C library on page 1-16. Related references 1.7.1 Building an application without the C library on page 1-40. Chapter 1 The ARM C and C++ Libraries on page 1-14. Chapter 2 The ARM C Micro-library on page 2-96. Related information ISO C library standard. IEEE Standard for Floating-Point Arithmetic (IEEE 754), 1985 version. What is Semihosting?. Rogue Wave Standard C++ Library Documentation. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-17 1 The ARM C and C++ Libraries 1.2 C and C++ runtime libraries 1.2.2 Compliance with the Application Binary Interface (ABI) for the ARM architecture The ABI for the ARM Architecture is a family of specifications that describes the processor-specific aspects of the translation of a source program into object files. Object files produced by any toolchain that conforms to the relevant aspects of the ABI can be linked together to produce a final executable image or library. Each document in the specification covers a specific area of compatibility. For example, the C Library ABI for the ARM® Architecture (CLIBABI) describes the parts of the C library that are expected to be common to all conforming implementations. The ABI documents contain several areas that are marked as platform specific. To define a complete execution environment these platform-specific details have to be provided. This gives rise to a number of supplemental specifications, for example the ARM GNU/Linux ABI supplement. The Base Standard ABI for the ARM® Architecture (BSABI) enables you to use ARM and Thumb® objects and libraries from different producers that support the ABI for the ARM Architecture. The ARM compilation tools fully support the BSABI, including support for Debug With Arbitrary Record Format (DWARF) 3 debug tables (DWARF Debugging Standard Version 3). The ARM C and C++ libraries conform to the standards described in the BSABI, the CLIBABI, and the C++ ABI for the ARM Architecture (CPPABI). Related tasks 1.2.3 Increasing portability of object files to other CLIBABI implementations on page 1-18. Related information Application Binary Interface (ABI) for the ARM Architecture. DWARF Debugging Standard. 1.2.3 Increasing portability of object files to other CLIBABI implementations You can request full CLIBABI portability to increase the portability of your object files to other implementations of the CLIBABI. Note This reduces the performance of some library operations. There are a number of methods you can use to request full CLIBABI portability. Procedure • Specify #define _AEABI_PORTABILITY_LEVEL 1 before you #include any library headers, such as . • Specify -D_AEABI_PORTABILITY_LEVEL=1 on the compiler command line. Related concepts 1.2.2 Compliance with the Application Binary Interface (ABI) for the ARM architecture on page 1-18. Related information Application Binary Interface (ABI) for the ARM Architecture. -Dname[(parm-list)][=def] compiler option. 1.2.4 ARM C and C++ library directory structure The libraries are installed in the armlib and cpplib subdirectories within install_directory\lib. armlib Contains the variants of the ARM C library, the floating-point arithmetic library (fplib), and the math library (mathlib). ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-18 1 The ARM C and C++ Libraries 1.2 C and C++ runtime libraries cpplib Contains the variants of the Rogue Wave C++ library (cpp_*) and supporting ARM C++ functions (cpprt_*), referred to collectively as the ARM C++ Libraries. The accompanying header files for these libraries are installed in: install_directory\include The environment variable ARMCC5LIB must be set to point to the lib directory, or if this variable is not set, ARMLIB. You must not identify the armlib and cpplib directories separately because this directory structure might change in future releases. The linker finds them from the location of lib. Note • • • • The ARM C libraries are supplied in binary form only. The ARM libraries must not be modified. If you want to create a new implementation of a library function, place the new function in an object file, or your own library, and include it when you link the application. Your version of the function is used instead of the standard library version. Normally, only a few functions in the ISO C library require re-implementation to create a targetdependent application. The source for the Rogue Wave Standard C++ Library is not freely distributable. It can be obtained from Rogue Wave Software Inc., or through ARM, for an additional license fee. Related information Rogue Wave Standard C++ Library Documentation. 1.2.5 Selection of ARM C and C++ library variants based on build options When you build your application, you must make certain choices such as the target architecture, instruction set, and byte order. You communicate these choices to the compiler using build options. The linker then selects appropriate C and C++ library variants compatible with these build options. Choices that influence the ARM C and C++ library variant include the following: Target Architecture and instruction set ARM or Thumb instruction sets. Byte order Big-endian or little-endian. Floating-point support • Software (SoftVFP). • Hardware (VFP). • Software or hardware with half-precision or double-precision extensions. • No floating-point support. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-19 1 The ARM C and C++ Libraries 1.2 C and C++ runtime libraries Position independence Different ways to access your data are as follows: • By absolute address. • Relative to sb (read/write position-independent). • Relative to pc (fpic). Different ways to access your code are as follows: • • By absolute address when appropriate. Relative to pc (read-only position independent). The standard C libraries provide variants to support all of these options. Position-independent C++ code can only be achieved with --apcs=/fpic. Note Position independence is not supported in microlib. When you link your assembler code, C or C++ code, the linker selects appropriate C and C++ library variants compatible with the build options you specified. There is a variant of the ISO C library for each combination of major build options. Related information Code compatibility between separately compiled and assembled modules. --apcs=qualifier...qualifier compiler option. --arm compiler option. --bigend compiler option. --fpu=name compiler option. --littleend compiler option. --thumb compiler option. --fpu=name linker option. --ropi linker option. --rwpi linker option. --arm assembler option. --bigend assembler option. --fpu assembler option. --littleend assembler option. --thumb assembler option. 1.2.6 Thumb C libraries The linker automatically links in the Thumb C library if the objects to be linked contain Thumb instructions. Objects use Thumb instructions if they have been built for: • • • • Thumb code, either using the --thumb option or #pragma thumb. Interworking, using the --apcs /interwork option on architecture ARMv4T. An ARMv6-M architecture target or processor, for example, Cortex®-M1 or Cortex-M0. An ARMv7-M architecture target or processor, for example, Cortex-M3. Despite its name, the Thumb C library might not contain exclusively Thumb code. If ARM instructions are available, the Thumb library might use them to improve the performance of critical functions such as memcpy(), memset(), and memclr(). The bulk of the Thumb C library, however, is coded in Thumb for the best code density. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-20 1 The ARM C and C++ Libraries 1.2 C and C++ runtime libraries For an ARM instruction-only build, compile with the --arm_only option. Note • • The Thumb C library used for ARMv6-M targets contains only 16-bit Thumb code. The Thumb C library used for ARMv7-M targets contains both 16-bit and 32-bit Thumb code. Related information Cortex-R series processors. Cortex-M series processors. --arm compiler option. --thumb compiler option. --arm_only compiler option. #pragma thumb. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-21 1 The ARM C and C++ Libraries 1.3 C and C++ library features 1.3 C and C++ library features The C library uses the standard ARM semihosted environment to provide facilities such as file input/ output. This environment is supported by the ARM DSTREAM debug and trace unit, the ARM RVI debug unit, and the Fixed Virtual Platform (FVP) models. You can re-implement any of the target-dependent functions of the C library as part of your application. This enables you to tailor the C library and, therefore, the C++ library, to your own execution environment. You can also tailor many of the target-independent functions to your own application-specific requirements. For example: • • • The malloc family. The ctype family. All the locale-specific functions. Many of the C library functions are independent of any other function and contain no target dependencies. You can easily exploit these functions from assembler code. Functions in the C library are responsible for: • Creating an environment in which a C or C++ program can execute. This includes: — Creating a stack. — Creating a heap, if required. — Initializing the parts of the library the program uses. • Starting execution by calling main(). • Supporting use of ISO-defined functions by the program. • Catching runtime errors and signals and, if required, terminating execution on error or program exit. Related information What is Semihosting?. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-22 1 The ARM C and C++ Libraries 1.4 C++ and C libraries and the std namespace 1.4 C++ and C libraries and the std namespace All C++ standard library names, including the C library names, if you include them, are defined in the namespace std. Standard library names are defined using the following C++ syntax: #include // instead of stdlib.h This means that you must qualify all the library names using one of the following methods: • Specify the standard namespace, for example: std::printf("example\n"); • Use the C++ keyword using to import a name to the global namespace: using namespace std; printf("example\n"); • Use the compiler option --using_std. Note errno is a macro, so it is not necessary to qualify it with a namespace. Related information --using_std, --no_using_std compiler option. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-23 1 The ARM C and C++ Libraries 1.5 Multithreaded support in ARM C libraries 1.5 Multithreaded support in ARM C libraries Describes the features that are supported by the ARM C libraries for creating multithreaded applications. This section contains the following subsections: • 1.5.1 ARM C libraries and multithreading on page 1-24. • 1.5.2 ARM C libraries and reentrant functions on page 1-24. • 1.5.3 ARM C libraries and thread-safe functions on page 1-25. • 1.5.4 Use of static data in the C libraries on page 1-25. • 1.5.5 Use of the __user_libspace static data area by the C libraries on page 1-26. • 1.5.6 C library functions to access subsections of the __user_libspace static data area on page 1-27. • 1.5.7 Re-implementation of legacy function __user_libspace() in the C library on page 1-27. • 1.5.8 Management of locks in multithreaded applications on page 1-28. • 1.5.9 How to ensure re-implemented mutex functions are called on page 1-30. • 1.5.10 Using the ARM C library in a multithreaded environment on page 1-30. • 1.5.11 Thread safety in the ARM C library on page 1-31. • 1.5.12 Thread safety in the ARM C++ library on page 1-31. • 1.5.13 The floating-point status word in a multithreaded environment on page 1-33. 1.5.1 ARM C libraries and multithreading The ARM C libraries support multithreading, for example, where you are using a Real-Time Operating System (RTOS). In this context, the following definitions are used: Threads Mean multiple streams of execution sharing global data between them. Process Means a collection of all the threads that share a particular set of global data. If there are multiple processes on a machine, they can be entirely separate and do not share any data (except under unusual circumstances). Each process might be a single-threaded process or might be divided into multiple threads. Where you have single-threaded processes, there is only one flow of control. In multithreaded applications, however, several flows of control might try to access the same functions, and the same resources, concurrently. To protect the integrity of resources, any code you write for multithreaded applications must be reentrant and thread-safe. Reentrancy and thread safety are both related to the way functions in a multithreaded application handle resources. Related concepts 1.5.2 ARM C libraries and reentrant functions on page 1-24. 1.5.3 ARM C libraries and thread-safe functions on page 1-25. Related references 1.5.10 Using the ARM C library in a multithreaded environment on page 1-30. 1.5.2 ARM C libraries and reentrant functions A reentrant function does not hold static data over successive calls, and does not return a pointer to static data. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-24 1 The ARM C and C++ Libraries 1.5 Multithreaded support in ARM C libraries For this type of function, the caller provides all the data that the function requires, such as pointers to any workspace. This means that multiple concurrent invocations of the function do not interfere with each other. Note A reentrant function must not call non-reentrant functions. Related concepts 1.5.3 ARM C libraries and thread-safe functions on page 1-25. 1.5.1 ARM C libraries and multithreading on page 1-24. 1.5.3 ARM C libraries and thread-safe functions A thread-safe function protects shared resources from concurrent access using locks. Thread safety concerns only how a function is implemented and not its external interface. In C, local variables are held in processor registers, or if the compiler runs out of registers, are dynamically allocated on the stack. Therefore, any function that does not use static data, or other shared resources, is thread-safe. Related concepts 1.5.2 ARM C libraries and reentrant functions on page 1-24. 1.5.1 ARM C libraries and multithreading on page 1-24. 1.5.11 Thread safety in the ARM C library on page 1-31. 1.5.12 Thread safety in the ARM C++ library on page 1-31. Related references 1.5.8 Management of locks in multithreaded applications on page 1-28. 1.5.10 Using the ARM C library in a multithreaded environment on page 1-30. 1.5.4 Use of static data in the C libraries Static data refers to persistent read/write data that is not stored on the stack or the heap. Callouts from the C library enable access to static data. Static data can be external or internal in scope, and is: • At a fixed address, when compiled with --apcs /norwpi. • At a fixed address relative to the static base, register r9, when compiled with --apcs /rwpi. • At a fixed address relative to the program counter (pc), when compiled with --apcs /fpic. Libraries that use static data might be reentrant, but this depends on their use of the __user_libspace static data area, and on the build options you choose: • When compiled with --apcs /norwpi, read/write static data is addressed in a position-dependent fashion. This is the default. Code from these variants is single-threaded because it uses read/write static data. • When compiled with --apcs /rwpi, read/write static data is addressed in a position-independent fashion using offsets from the static base register sb. Code from these variants is reentrant and can be multithreaded if each thread uses a different static base value. The following describes how the C libraries use static data: • The default floating-point arithmetic libraries fz_* and fj_* do not use static data and are always reentrant. For software floating-point, the f_* and g_* libraries use static data to store the FloatingPoint (FP) status word. For hardware floating-point, the f_* and g_* libraries do not use static data. • All statically-initialized data in the C libraries is read-only. • All writable static data is zero-initialized. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-25 1 The ARM C and C++ Libraries 1.5 Multithreaded support in ARM C libraries • • Most C library functions use no writable static data and are reentrant whether built with default build options, --apcs /norwpi or reentrant build options, --apcs /rwpi. Some functions have static data implicit in their definitions. You must not use these in a reentrant application unless you build it with --apcs /rwpi and the callers use different values in sb. Note Exactly which functions use static data in their definitions might change in future releases. Callouts from the C library enable access to static data. C library functions that use static data can be categorized as: • • • Functions that do not use any static data of any kind, for example fprintf(). Functions that manage a static state, such as malloc(), rand(), and strtok(). Functions that do not manage a static state, but use static data in a way that is specific to the implementation in ARM Compiler, for example isalpha(). When the C library does something that requires implicit static data, it uses a callout to a function you can replace. These functions are shown in the following table. They do not use semihosting. Table 1-1 C library callouts Function Description __rt_errno_addr() Called to get the address of the variable errno __rt_fp_status_addr() Called by the floating-point support code to get the address of the floating-point status word locale functions The function __user_libspace() creates a block of private static data for the library The default implementation of __user_libspace creates a 96-byte block in the ZI region. Even if your application does not have a main() function, the __user_libspace() function does not normally have to be redefined. Note Exactly which functions use static data in their definitions might change in future releases. Related concepts 1.5.7 Re-implementation of legacy function __user_libspace() in the C library on page 1-27. 1.9 Assembler macros that tailor locale functions in the C library on page 1-52. 1.5.1 ARM C libraries and multithreading on page 1-24. Related references 4.26 __rt_fp_status_addr() on page 4-166. Related information Code compatibility between separately compiled and assembled modules. --apcs=qualifier...qualifier compiler option. 1.5.5 Use of the __user_libspace static data area by the C libraries The __user_libspace static data area holds the static data for the C libraries. The C libraries use the __user_libspace area to store a number of different types of data. This is a block of 96 bytes of zero-initialized data, supplied by the C library. It is also used as a temporary stack during C library initialization. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-26 1 The ARM C and C++ Libraries 1.5 Multithreaded support in ARM C libraries The default ARM C libraries use the __user_libspace area to hold: • errno, used by any function that is capable of setting errno. By default, __rt_errno_addr() returns a pointer to errno. • The Floating-Point (FP) status word for software floating-point (exception flags, rounding mode). It is unused in hardware floating-point. By default, __rt_fp_status_addr() returns a pointer to the FP status word. • A pointer to the base of the heap (that is, the __Heap_Descriptor), used by all the malloc-related functions. • The current locale settings, used by functions such as setlocale(), but also used by all other library functions that depend on them. For example, the ctype.h functions have to access the LC_CTYPE setting. The C++ libraries use the __user_libspace area to hold: • The new_handler field and ddtor_pointer: — The new_handler field keeps track of the value passed to std::set_new_handler(). — The ddtor_pointer, that points to a list of destructions to be performed on program exit. For example, objects constructed outside function scope exist for the duration of the program, but require destruction on program exit. The ddtor_pointer is used by __cxa_atexit() and __aeabi_atexit(). • C++ exception handling information for functions such as std::set_terminate() and std::set_unexpected(). Note How the C and C++ libraries use the __user_libspace area might change in future releases. Related concepts 1.5.6 C library functions to access subsections of the __user_libspace static data area on page 1-27. Related information __aeabi_atexit() in C++ ABI for the ARM Architecture. 1.5.6 C library functions to access subsections of the __user_libspace static data area The __user_perproc_libspace() and __user_perthread_libspace() functions return subsections of the __user_libspace static data area. __user_perproc_libspace() Returns a pointer to 96 bytes of 4-byte aligned memory for storing data that is global to an entire process. This data is shared between all threads. __user_perthread_libspace() Returns a pointer to 96 bytes of 4-byte aligned memory for storing data that is local to a particular thread. This means that __user_perthread_libspace() returns a different address depending on the thread it is called from. Related concepts 1.5.7 Re-implementation of legacy function __user_libspace() in the C library on page 1-27. Related references 1.5.5 Use of the __user_libspace static data area by the C libraries on page 1-26. 1.5.7 Re-implementation of legacy function __user_libspace() in the C library The __user_libspace() function returns a pointer to a block of private static data for the C library. This function does not normally have to be redefined. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-27 1 The ARM C and C++ Libraries 1.5 Multithreaded support in ARM C libraries If you are writing an operating system or a process switcher, then typically you use the __user_perproc_libspace() and __user_perthread_libspace() functions (which are always available) rather than re-implement __user_libspace(). If you have legacy source code that re-implements __user_libspace(), you do not have to change the re-implementation for single-threaded processes. However, you are likely to be required to do so for multi-threaded applications. For multi-threaded applications, use either or both of __user_perproc_libspace() and __user_perthread_libspace(), instead of __user_libspace(). Related concepts 1.5.6 C library functions to access subsections of the __user_libspace static data area on page 1-27. 1.5.4 Use of static data in the C libraries on page 1-25. Related references 1.5.10 Using the ARM C library in a multithreaded environment on page 1-30. 1.5.5 Use of the __user_libspace static data area by the C libraries on page 1-26. 1.5.8 Management of locks in multithreaded applications A thread-safe function protects shared resources from concurrent access using locks. There are functions in the C library that you can re-implement, that enable you to manage the locking mechanisms and so prevent the corruption of shared data such as the heap. These functions are mutex functions, where the lifecycle of a mutex is one of initialization, iterative acquisition and releasing of the mutex as required, and then optionally freeing the mutex when it is never going to be required again. The mutex functions wrap onto your own Real-Time Operating System (RTOS) calls, and their function prototypes are: _mutex_initialize() int _mutex_initialize(mutex *m); This function accepts a pointer to a 32-bit word and initializes it as a valid mutex. By default, _mutex_initialize() returns zero for a nonthreaded application. Therefore, in a multithreaded application, _mutex_initialize() must return a nonzero value on success so that at runtime, the library knows that it is being used in a multithreaded environment. Ensure that _mutex_initialize() initializes the mutex to an unlocked state. This function must be supplied if you are using mutexes. _mutex_acquire() void _mutex_acquire(mutex *m); This function causes the calling thread to obtain a lock on the supplied mutex. _mutex_acquire() returns immediately if the mutex has no owner. If the mutex is owned by another thread, _mutex_acquire() must block until it becomes available. _mutex_acquire() is not called by the thread that already owns the mutex. This function must be supplied if you are using mutexes. _mutex_release() void _mutex_release(mutex *m); This function causes the calling thread to release the lock on a mutex acquired by _mutex_acquire(). The mutex remains in existence, and can be re-locked by a subsequent call to mutex_acquire(). _mutex_release() assumes that the mutex is owned by the calling thread. This function must be supplied if you are using mutexes. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-28 1 The ARM C and C++ Libraries 1.5 Multithreaded support in ARM C libraries _mutex_free() void _mutex_free(mutex *m); This function causes the calling thread to free the supplied mutex. Any operating system resources associated with the mutex are freed. The mutex is destroyed and cannot be reused. _mutex_free() assumes that the mutex is owned by the calling thread. This function is optional. If you do not supply this function, the C library does not attempt to call it. The mutex data structure type that is used as the parameter to the _mutex_*() functions is not defined in any of the ARM Compiler toolchain header files, but must be defined elsewhere. Typically, it is defined as part of RTOS code. Functions that call _mutex_*() functions create 4 bytes of memory for holding the mutex data structure. __Heap_Initialize() is one such function. For the C library, a mutex is specified as a single 32-bit word of memory that can be placed anywhere. However, if your mutex implementation requires more space than this, or demands that the mutex be in a special memory area, then you must treat the default mutex as a pointer to a real mutex. A typical example of a re-implementation framework for _mutex_initialize(), _mutex_acquire(), and _mutex_release() is as follows, where SEMAPHORE_ID, CreateLock(), AcquireLock(), and ReleaseLock() are defined in the RTOS, and (...) implies additional parameters: int _mutex_initialize(SEMAPHORE_ID *sid) { /* Create a mutex semaphore */ *sid = CreateLock(...); return 1; } void _mutex_acquire(SEMAPHORE_ID *sid) { /* Task sleep until get semaphore */ AcquireLock(*sid, ...); } void _mutex_release(SEMAPHORE_ID *sid) { /* Release the semaphore. */ ReleaseLock(*sid); } void _mutex_free(SEMAPHORE_ID *sid) { /* Free the semaphore. */ FreeLock(*sid, ...); } Note • • _mutex_release() releases the lock on the mutex that was acquired by _mutex_acquire(), but the mutex still exists, and can be re-locked by a subsequent call to _mutex_acquire(). It is only when the optional wrapper function _mutex_free() is called that the mutex is destroyed. After the mutex is destroyed, it cannot be used without first calling _mutex_initialize() to set it up again. Related concepts 1.5.9 How to ensure re-implemented mutex functions are called on page 1-30. 1.5.11 Thread safety in the ARM C library on page 1-31. 1.5.12 Thread safety in the ARM C++ library on page 1-31. Related references 1.5.10 Using the ARM C library in a multithreaded environment on page 1-30. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-29 1 The ARM C and C++ Libraries 1.5 Multithreaded support in ARM C libraries 1.5.9 How to ensure re-implemented mutex functions are called If your re-implemented _mutex_*() functions are within an object that is contained within a library file, the linker does not automatically include the object. This can result in the _mutex_*() functions being excluded from the image you have built. To ensure that your _mutex_*() functions are called, you can either: • Place your mutex functions in a non-library object file. This helps to ensure that they are resolved at link time. • Place your mutex functions in a library object file, and arrange a non-weak reference to something in the object. • Place your mutex functions in a library object file, and have the linker explicitly extract the specific object from the library on the command line by writing libraryname.a(objectfilename.o) when you invoke the linker. Related concepts 1.5.11 Thread safety in the ARM C library on page 1-31. 1.5.12 Thread safety in the ARM C++ library on page 1-31. Related references 1.5.10 Using the ARM C library in a multithreaded environment on page 1-30. 1.5.8 Management of locks in multithreaded applications on page 1-28. 1.5.10 Using the ARM C library in a multithreaded environment There are a number of requirements you must fulfill before you can use the ARM C library in a multithreaded environment. To use the ARM C library in a multithreaded environment, you must provide: • An implementation of __user_perthread_libspace() that returns a different block of memory for each thread. This can be achieved by either: — Returning a different address depending on the thread it is called from. — Having a single __user_perthread_libspace block at a fixed address and swapping its contents when switching threads. You can use either approach to suit your environment. • • You do not have to re-implement __user_perproc_libspace() unless there is a specific reason to do so. In the majority of cases, there is no requirement to re-implement this function. A way to manage multiple stacks. A simple way to do this is to use the ARM two-region memory model. Using this means that you keep the stack that belongs to the primary thread entirely separate from the heap. Then you must allocate more memory for additional stacks from the heap itself. Thread management functions, for example, to create or destroy threads, to handle thread synchronization, and to retrieve exit codes. Note The ARM C libraries supply no thread management functions of their own so you must supply any that are required. • A thread-switching mechanism. Note The ARM C libraries supply no thread-switching mechanisms of their own. This is because there are many different ways to do this and the libraries are designed to work with all of them. You only have to provide implementations of the mutex functions if you require them to be called. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-30 1 The ARM C and C++ Libraries 1.5 Multithreaded support in ARM C libraries In some applications, the mutex functions might not be useful. For example, a co-operatively threaded program does not have to take steps to ensure data integrity, provided it avoids calling its yield function during a critical section. However, in other types of application, for example where you are implementing preemptive scheduling, or in a Symmetric Multi-Processor (SMP) model, these functions play an important part in handling locks. If all of these requirements are met, you can use the ARM C library in your multithreaded environment. The following behavior applies: • Some functions work independently in each thread. • Some functions automatically use the mutex functions to mediate multiple accesses to a shared resource. • Some functions are still nonreentrant so a reentrant equivalent is supplied. • A few functions remain nonreentrant and no alternative is available. Related concepts 1.5.1 ARM C libraries and multithreading on page 1-24. 1.5.9 How to ensure re-implemented mutex functions are called on page 1-30. 1.5.11 Thread safety in the ARM C library on page 1-31. 1.5.12 Thread safety in the ARM C++ library on page 1-31. Related references 1.5.8 Management of locks in multithreaded applications on page 1-28. 4.59 Thread-safe C library functions on page 4-200. 4.60 C library functions that are not thread-safe on page 4-202. 1.5.11 Thread safety in the ARM C library ARM C library functions are either always thread-safe, never thread-safe, or thread-safe in certain circumstances. In the ARM C library: • Some functions are never thread-safe, for example setlocale(). • Some functions are inherently thread-safe, for example memcpy(). • Some functions, such as malloc(), can be made thread-safe by implementing the _mutex_* functions. • Other functions are only thread-safe if you pass the appropriate arguments, for example tmpnam(). Threading problems might occur when your application makes use of the ARM C library in a way that is hidden, for example, if the compiler implicitly calls functions that you have not explicitly called in your source code. Familiarity with the thread-safe C library functions and C library functions that are not thread-safe can help you to avoid this type of threading problem, although in general, it is unlikely to arise. Related concepts 1.5.9 How to ensure re-implemented mutex functions are called on page 1-30. 1.5.12 Thread safety in the ARM C++ library on page 1-31. Related references 1.5.10 Using the ARM C library in a multithreaded environment on page 1-30. 1.5.8 Management of locks in multithreaded applications on page 1-28. 4.59 Thread-safe C library functions on page 4-200. 4.60 C library functions that are not thread-safe on page 4-202. 1.5.12 Thread safety in the ARM C++ library ARM C++ library functions are either always thread-safe, never thread-safe, or thread-safe in certain circumstances. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-31 1 The ARM C and C++ Libraries 1.5 Multithreaded support in ARM C libraries The following points summarize thread safety in the C++ library: • The function std::set_new_handler() is not thread-safe. This means that some forms of ::operator new and ::operator delete are not thread-safe with respect to std::set_new_handler(): — The default C++ runtime library implementations of the following use malloc() and free() and are thread-safe with respect to each other. They are not thread-safe with respect to std::set_new_handler(). You are permitted to replace them: ::operator new(std::size_t) ::operator new[](std::size_t) ::operator new(std::size_t, const std::nothrow_t&) ::operator new[](std::size_t, const std::nothrow_t) ::operator delete(void*) ::operator delete[](void*) ::operator delete(void*, const std::nothrow_t&) ::operator delete[](void*, const std::nothrow_t&) — The following placement forms are also thread-safe. You are not permitted to replace them: ::operator new(std::size_t, void*) ::operator new[](std::size_t, void*) ::operator delete(void*, void*) ::operator delete[](void*, void*) • • Construction and destruction of global objects are not thread-safe. Construction of local static objects can be made thread-safe if you re-implement the functions __cxa_guard_acquire(), __cxa_guard_release(), __cxa_guard_abort(), __cxa_atexit() and __aeabi_atexit() appropriately. For example, with appropriate re-implementation, the following construction of lsobj can be made thread-safe: struct T { T(); }; void f() { static T lsobj; } • • Throwing an exception is thread-safe if any user constructors and destructors that get called are also thread-safe. The ARM C++ library uses the ARM C library. To use the ARM C++ library in a multithreaded environment, you must provide the same functions that you would be required to provide when using the ARM C library in a multithreaded environment. Rogue Wave Standard C++ library The Rogue Wave Standard C++ library is a part of the ARM C++ library. What applies to the ARM C++ library applies to the Rogue Wave Standard C++ library too. In the Rogue Wave Standard C++ library, specifically: • • All containers and all functions are reentrant, making no use of internal, modifiable static data. — Except for the std::random_shuffle function, which uses static data to record the state of the random number generator. The iostream and locale classes are not thread safe. You must protect shared objects while using the iostream and locale classes, and the std::random_shuffle function. To do this, you might use mutex functions, or co-operative threading. As an example, in a typical case of a pre-emptive multitasking environment, one that uses mutex functions with containers, this means that: • Reader threads can safely share a container if no thread writes to it during the reads. • While a thread writes to a shared container, you must apply locking around the use of the container. • Writer threads can write to different containers safely. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-32 1 The ARM C and C++ Libraries 1.5 Multithreaded support in ARM C libraries • • You must apply locking around the use of random_shuffle. Multiple threads cannot use iostream and locale classes safely unless you apply locking around the use of their objects. Related concepts 1.5.9 How to ensure re-implemented mutex functions are called on page 1-30. 1.5.11 Thread safety in the ARM C library on page 1-31. Related references 1.5.10 Using the ARM C library in a multithreaded environment on page 1-30. 1.5.8 Management of locks in multithreaded applications on page 1-28. Related information __cxa_*, __aeabi_* functions, C++ ABI for the ARM Architecture. Rogue Wave Standard C++ Library Documentation. 1.5.13 The floating-point status word in a multithreaded environment Applicable to variants of the software floating-point libraries that require a status word (--fpmode=ieee_fixed or --fpmode=ieee_full), the floating-point status word is safe to use in a multithreaded environment, even with software floating-point. A status word for each thread is stored in its own __user_perthread_libspace block. Note In a hardware floating-point environment, the floating-point status word is stored in a Vector FloatingPoint (VFP) register. In this case, your thread-switching mechanism must keep a separate copy of this register for each thread. Related concepts 1.5.11 Thread safety in the ARM C library on page 1-31. Related information --fpmode=model compiler option. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-33 1 The ARM C and C++ Libraries 1.6 Support for building an application with the C library 1.6 Support for building an application with the C library Describes the ARM Compiler features that are supported when building an application with the C library. This section contains the following subsections: • 1.6.1 Using the C library with an application on page 1-34. • 1.6.2 Using the C and C++ libraries with an application in a semihosting environment on page 1-34. • 1.6.3 Using $Sub$$ to mix semihosted and nonsemihosted I/O functionality on page 1-35. • 1.6.4 Using the libraries in a nonsemihosting environment on page 1-35. • 1.6.5 C++ exceptions in a non-semihosting environment on page 1-36. • 1.6.6 Direct semihosting C library function dependencies on page 1-36. • 1.6.7 Indirect semihosting C library function dependencies on page 1-37. • 1.6.8 C library API definitions for targeting a different environment on page 1-38. 1.6.1 Using the C library with an application Depending on how you use the C and C ++ libraries with your application, you might have to reimplement particular functions. You can use the C and C ++ libraries with an application in the following ways: • Build a semihosting application that can be debugged in a semihosted environment such as with the ARM DSTREAM debug and trace unit. • Build a non-hosted application that, for example, can be embedded into ROM. • Build an application that does not use main() and does not initialize the library. This application has restricted library functionality, unless you re-implement some functions. Related concepts 1.6.2 Using the C and C++ libraries with an application in a semihosting environment on page 1-34. 1.6.4 Using the libraries in a nonsemihosting environment on page 1-35. Related references 1.7.1 Building an application without the C library on page 1-40. 1.6.2 Using the C and C++ libraries with an application in a semihosting environment If you are developing an application to run in a semihosted environment for debugging, you must have an execution environment that supports ARM or Thumb semihosting, and that has sufficient memory. The execution environment can be provided by either: • Using the standard semihosting functionality that is present by default in, for example, the ARM DSTREAM debug and trace unit. • Implementing your own handler for the semihosting calls. It is not necessary to write any new functions or include files if you are using the default semihosting functionality of the C and C++ libraries. The ARM debug agents support semihosting, but the memory map assumed by the C library might require tailoring to match the hardware being debugged. Related references 1.6.3 Using $Sub$$ to mix semihosted and nonsemihosted I/O functionality on page 1-35. 1.6.6 Direct semihosting C library function dependencies on page 1-36. Related information What is Semihosting?. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-34 1 The ARM C and C++ Libraries 1.6 Support for building an application with the C library 1.6.3 Using $Sub$$ to mix semihosted and nonsemihosted I/O functionality You can use $Sub$$ to provide a mixture of semihosted and nonsemihosted functionality. For example, given an implementation of fputc() that writes directly to a UART, and a semihosted implementation of fputc(), you can provide both of these depending on the nature of the FILE * pointer passed into the function. Example 1-1 Using $Sub$$ to mix semihosting and nonsemihosting I/O functionality int $Super$$fputc(int c, FILE *fp); int $Sub$$fputc(int c, FILE *fp) { if (fp == (FILE *)MAGIC_NUM) // where MAGIC_NUM is a special value that { // is different to all normal FILE * pointer // values. write_to_UART(c); return c; } else { return $Super$$fputc(c, fp); } } Related concepts 1.6.2 Using the C and C++ libraries with an application in a semihosting environment on page 1-34. 1.6.4 Using the libraries in a nonsemihosting environment on page 1-35. Related information Use of $Super$$ and $Sub$$ to patch symbol definitions. ELF for the ARM Architecture. 1.6.4 Using the libraries in a nonsemihosting environment Some C library functions use semihosting. If you use the libraries in a nonsemihosting environment, you must ensure that semihosting function calls are dealt with appropriately. If you do not want to use semihosting, either: • • • Remove all calls to semihosting functions. Re-implement the lower-level functions, for example, fputc(). You are not required to re-implement all semihosting functions. You must, however, re-implement the functions you are using in your application. You must re-implement functions that the C library uses to isolate itself from target dependencies. For example, if you use printf() you must re-implement fputc(). If you do not use the higher-level input/output functions like printf(), you do not have to re-implement the lower-level functions like fputc(). Implement a handler for all of the semihosting calls to be handled in your own specific way. One such example is for the handler to intercept the calls, redirecting them to your own nonsemihosted, that is, target-specific, functions. To guarantee that no functions using semihosting are included in your application, use either: • • ARM DUI0378G_02 IMPORT __use_no_semihosting from armasm assembly language. #pragma import(__use_no_semihosting) from C. Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-35 1 The ARM C and C++ Libraries 1.6 Support for building an application with the C library Note IMPORT __use_no_semihosting is only required to be added to a single assembly source file. Similarly, #pragma import(__use_no_semihosting) is only required to be added to a single C source file. It is unnecessary to add these inserts to every single source file. If you include a library function that uses semihosting and also reference __use_no_semihosting, the library detects the conflicting symbols and the linker reports an error. To determine which objects are using semihosting: 1. Link with armlink --verbose --list err.txt 2. Search err.txt for occurrences of __I$use$semihosting For example: ... Loading member sys_exit.o from c_4.l. reference : __I$use$semihosting definition: _sys_exit ... This shows that the semihosting-using function _sys_exit is linked-in from the C library. To prevent this, you must provide your own implementation of this function. There are no target-dependent functions in the C++ library, although some C++ functions use underlying C library functions that are target-dependent. Related concepts 1.1 Mandatory linkage with the C library on page 1-16. 1.6.5 C++ exceptions in a non-semihosting environment on page 1-36. Related references 1.6.3 Using $Sub$$ to mix semihosted and nonsemihosted I/O functionality on page 1-35. 1.6.7 Indirect semihosting C library function dependencies on page 1-37. 1.6.8 C library API definitions for targeting a different environment on page 1-38. Related information What is Semihosting?. --list=filename linker option. --verbose linker option. 1.6.5 C++ exceptions in a non-semihosting environment The default C++ std::terminate() handler is required by the C++ Standard to call abort(). The default C library implementation of abort() uses functions that require semihosting support. Therefore, if you use exceptions in a non-semihosting environment, you must provide an alternative implementation of abort(). Related concepts 1.6.4 Using the libraries in a nonsemihosting environment on page 1-35. 1.6.6 Direct semihosting C library function dependencies A table showing the functions that depend directly on semihosting. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-36 1 The ARM C and C++ Libraries 1.6 Support for building an application with the C library Table 1-2 Direct semihosting dependencies Function Description __user_setup_stackheap() Sets up and returns the locations of the stack and the heap. You might have to re-implement this function if you are using a scatter file at the link stage. _sys_exit() Error signaling, error handling, and program exit. _ttywrch() _sys_command_string() Tailoring input/output functions in the C and C++ libraries. _sys_close() _sys_iserror() _sys_istty() _sys_flen() _sys_open() _sys_read() _sys_seek() _sys_write() _sys_tmpnam() clock() Tailoring other C library functions. _clock_init() remove() rename() system() time() Related concepts 1.6.2 Using the C and C++ libraries with an application in a semihosting environment on page 1-34. 1.8.1 Initialization of the execution environment and execution of the application on page 1-47. Related references 1.10 Modification of C library functions for error signaling, error handling, and program exit on page 1-61. 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.6.7 Indirect semihosting C library function dependencies on page 1-37. 1.21 Tailoring non-input/output C library functions on page 1-81. 4.54 __user_setup_stackheap() on page 4-195. 1.6.7 Indirect semihosting C library function dependencies A table showing functions that depend indirectly on one or more of the directly dependent functions. You can use this table as an initial guide, but it is recommended that you use either of the following to identify any other functions with indirect or direct dependencies on semihosting at link time: • #pragma import(__use_no_semihosting) in C source code. • IMPORT __use_no_semihosting in armasm assembly language source code. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-37 1 The ARM C and C++ Libraries 1.6 Support for building an application with the C library Table 1-3 Indirect semihosting dependencies Function Usage __raise() Catching, handling, or diagnosing C library exceptions, without C signal support. (Tailoring error signaling, error handling, and program exit.) __default_signal_handler() Catching, handling, or diagnosing C library exceptions, with C signal support. (Tailoring error signaling, error handling, and program exit.) __Heap_Initialize() Choosing or redefining memory allocation. Avoiding the heap and heap-using C library functions supplied by ARM. ferror(), fputc(), __stdout Re-implementing the printf family. (Tailoring input/output functions in the C and C+ + libraries.). __backspace(), fgetc(), __stdin Re-implementing the scanf family. (Tailoring input/output functions in the C and C+ + libraries.). fwrite(), fputs(), puts(), fread(), Re-implementing the stream output family. (Tailoring input/output functions in the fgets(), gets(), ferror() C and C++ libraries.). Related concepts 1.6.4 Using the libraries in a nonsemihosting environment on page 1-35. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. 1.11.5 Avoiding the heap and heap-using library functions supplied by ARM on page 1-67. 1.10 Modification of C library functions for error signaling, error handling, and program exit on page 1-61. 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.21 Tailoring non-input/output C library functions on page 1-81. 1.6.8 C library API definitions for targeting a different environment In addition to the direct and indirect semihosting dependent functions, there are a number of functions and files that might be useful when building for a different environment. The following table shows these functions and files. Table 1-4 Published API definitions File or function Description __main(), __rt_entry() Initializes the runtime environment and executes the user application __rt_lib_init(), __rt_exit(), Initializes or finalizes the runtime library __rt_lib_shutdown() LC_CTYPE locale Defines the character properties for the local alphabet rt_sys.h A C header file describing all the functions whose default (semihosted) implementations use semihosting calls rt_heap.h A C header file describing the storage management abstract data type rt_locale.h A C header file describing the five locale category filing systems, and defining some macros that are useful for describing the contents of locale categories rt_misc.h A C header file describing miscellaneous unrelated public interfaces to the C library rt_memory.s An empty, but commented, prototype implementation of the memory model ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-38 1 The ARM C and C++ Libraries 1.6 Support for building an application with the C library If you are re-implementing a function that exists in the standard ARM library, the linker uses an object or library from your project rather than the standard ARM library. Caution Do not replace or delete libraries supplied by ARM. You must not overwrite the supplied library files. Place your re-implemented functions in separate object files or libraries instead. Related concepts 1.6.4 Using the libraries in a nonsemihosting environment on page 1-35. 1.9 Assembler macros that tailor locale functions in the C library on page 1-52. Related information --list=filename linker option. --verbose linker option. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-39 1 The ARM C and C++ Libraries 1.7 Support for building an application without the C library 1.7 Support for building an application without the C library Describes the ARM Compiler features that are supported and not supported when building an application without the C library. This section contains the following subsections: • 1.7.1 Building an application without the C library on page 1-40. • 1.7.2 Creating an application as bare machine C without the C library on page 1-42. • 1.7.3 Integer and floating-point compiler functions and building an application without the C library on page 1-42. • 1.7.4 Bare machine integer C on page 1-43. • 1.7.5 Bare machine C with floating-point processing on page 1-43. • 1.7.6 Customized C library startup code and access to C library functions on page 1-44. • 1.7.7 Using low-level functions when exploiting the C library on page 1-45. • 1.7.8 Using high-level functions when exploiting the C library on page 1-45. • 1.7.9 Using malloc() when exploiting the C library on page 1-46. 1.7.1 Building an application without the C library If your application does not initialize the C library, a number of functions are not available in your application. Creating an application that has a main() function causes the C library initialization functions to be included as part of __rt_lib_init. If your application does not have a main() function, the C library is not initialized and the following functions are not available in your application: • Low-level stdio functions that have the prefix _sys_. • Signal-handling functions, signal() and raise() in signal.h. • Other functions, such as atexit(). The following table shows header files, and the functions they contain, that are available with an uninitialized library. Some otherwise unavailable functions can be used if the library functions they depend on are re-implemented. Table 1-5 Standalone C library functions Function Description alloca.h Functions in this file work without any library initialization or function re-implementation. You must know how to build an application with the C library to use this header file. assert.h Functions listed in this file require high-level stdio, __rt_raise(), and _sys_exit(). You must be familiar with tailoring error signaling, error handling, and program exit to use this header file. ctype.h Functions listed in this file require the locale functions. errno.h Functions in this file work without the requirement for any library initialization or function re-implementation. fenv.h Functions in this file work without the requirement for any library initialization and only require the re-implementation of __rt_raise(). float.h This file does not contain any code. The definitions in the file do not require library initialization or function reimplementation. inttypes.h Functions listed in this file require the locale functions. limits.h Functions in this file work without the requirement for any library initialization or function re-implementation. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-40 1 The ARM C and C++ Libraries 1.7 Support for building an application without the C library Table 1-5 Standalone C library functions (continued) Function Description locale.h Call setlocale() before calling any function that uses locale functions. For example: setlocale(LC_ALL, "C"); See the contents of locale.h for more information on the following functions and data structures: • • • • setlocale() selects the appropriate locale as specified by the category and locale arguments. lconv is the structure used by locale functions for formatting numeric quantities according to the rules of the current locale. localeconv() creates an lconv structure and returns a pointer to it. _get_lconv() fills the lconv structure pointed to by the parameter. This ISO extension removes the requirement for static data within the library. locale.h also contains constant declarations used with locale functions. math.h For functions in this file to work, you must first call _fp_init() and re-implement __rt_raise(). setjmp.h Functions in this file work without any library initialization or function re-implementation. signal.h Functions listed in this file are not available without library initialization. You must know how to build an application with the C library to use this header file. __rt_raise() can be re-implemented for error and exit handling. You must be familiar with tailoring error signaling, error handling, and program exit. stdarg.h Functions listed in this file work without any library initialization or function re-implementation. stddef.h This file does not contain any code. The definitions in the file do not require library initialization or function reimplementation. stdint.h This file does not contain any code. The definitions in the file do not require library initialization or function reimplementation. stdio.h The following dependencies or limitations apply to these functions: • The high-level functions such as printf(), scanf(), puts(), fgets(), fread(), fwrite(), and perror() depend on lower-level stdio functions fgetc(), fputc(), and __backspace(). You must re-implement these lower-level functions when using the standalone C library. However, you cannot re-implement the _sys_ prefixed functions (for example, _sys_read()) when using the standalone C library because the layer of stdio that calls the _sys_ functions requires library initialization. • • You must be familiar with tailoring the input/output functions in the C and C++ libraries. The printf() and scanf() family of functions require locale. The remove() and rename() functions are system-specific and probably not usable in your application. stdlib.h Most functions in this file work without any library initialization or function re-implementation. The following functions depend on other functions being instantiated correctly: • ato*() requires locale. • strto*() requires locale. • malloc(), calloc(), realloc(), and free() require heap functions. • atexit() is not available when building an application without the C library. string.h Functions in this file work without any library initialization, with the exception of strcoll() and strxfrm(), that require locale. time.h mktime() and localtime() can be used immediately time() and clock() are system-specific and are probably not usable unless re-implemented asctime(), ctime(), and strftime() require locale. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-41 1 The ARM C and C++ Libraries 1.7 Support for building an application without the C library Table 1-5 Standalone C library functions (continued) Function Description wchar.h Wide character library functions added to ISO C by Normative Addendum 1 in 1994. The following dependencies or limitations apply to these functions: • The high-level functions such as swprintf(), vswprintf(), swscanf(), and vswscanf() depend on lowerlevel stdio functions such as fgetwc() and fputwc(). You must re-implement these lower-level functions when using the standalone C library. See 1.13 Target dependencies on low-level functions in the C and C++ libraries on page 1-70 for more information. • The high-level functions such as swprintf(), vswprintf(), swscanf(), and vswscanf() require locale. • All the conversion functions (for example, btowc, wctob, mbrtowc, and wcrtomb) require locale. • wcscoll() and wcsxfrm() require locale. wctype.h Wide character library functions added to ISO C by Normative Addendum 1 in 1994. This requires locale. Related concepts 1.7.2 Creating an application as bare machine C without the C library on page 1-42. 1.9 Assembler macros that tailor locale functions in the C library on page 1-52. 1.7.3 Integer and floating-point compiler functions and building an application without the C library on page 1-42. 1.7.8 Using high-level functions when exploiting the C library on page 1-45. 1.7.7 Using low-level functions when exploiting the C library on page 1-45. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.10 Modification of C library functions for error signaling, error handling, and program exit on page 1-61. 1.7.2 Creating an application as bare machine C without the C library Bare machine C applications do not automatically use the full C runtime environment provided by the C library. Even though you are creating an application without the library, some functions from the library that are called implicitly by the compiler must be included. There are also many library functions that can be made available with only minor re-implementations. Related concepts 1.7.3 Integer and floating-point compiler functions and building an application without the C library on page 1-42. 1.7.6 Customized C library startup code and access to C library functions on page 1-44. 1.7.7 Using low-level functions when exploiting the C library on page 1-45. 1.7.8 Using high-level functions when exploiting the C library on page 1-45. 1.7.4 Bare machine integer C on page 1-43. 1.7.5 Bare machine C with floating-point processing on page 1-43. 1.7.9 Using malloc() when exploiting the C library on page 1-46. Related references 1.7.1 Building an application without the C library on page 1-40. 1.7.3 Integer and floating-point compiler functions and building an application without the C library There are several compiler helper functions that the compiler uses to handle operations that do not have a short machine code equivalent. These functions require __rt_raise(). ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-42 1 The ARM C and C++ Libraries 1.7 Support for building an application without the C library For example, integer divide uses a function that is implicitly called by the compiler if there is no divide instruction available in the target instruction set. (ARMv7-R and ARMv7-M architectures use the instructions SDIV and UDIV in Thumb state. Other versions of the ARM architecture also use compiler functions that are implicitly invoked.) Integer divide, and all the floating-point functions if you use a floating-point mode that involves throwing exceptions, require __rt_raise() to handle math errors. Re-implementing __rt_raise() enables all the math functions, and it avoids having to link in all the signal-handling library code. Related concepts 1.7.2 Creating an application as bare machine C without the C library on page 1-42. 1.25 Compiler generated and library-resident helper functions on page 1-90. Related references 1.7.1 Building an application without the C library on page 1-40. Related information Application Binary Interface (ABI) for the ARM Architecture. 1.7.4 Bare machine integer C If you are writing a program in C that does not use the library and is to run without any environment initialization, there are a number of requirements you must meet. These requirements are: • Re-implement __rt_raise() if you are using the heap. • Not define main(), to avoid linking in the library initialization code. • Write an assembly language veneer that establishes the register state required to run C. This veneer must branch to the entry function in your application. • Provide your own RW/ZI initialization code. • Ensure that your initialization veneer is executed by, for example, placing it in your reset handler. • Build your application using --fpu=none. When you have met these requirements, link your application normally. The linker uses the appropriate C library variant to find any required compiler functions that are implicitly called. Many library facilities require __user_libspace for static data. Even without the initialization code activated by having a main() function, __user_libspace is created automatically and uses 96 bytes in the ZI segment. Related concepts 1.7.2 Creating an application as bare machine C without the C library on page 1-42. 1.7.5 Bare machine C with floating-point processing on page 1-43. Related references 1.7.1 Building an application without the C library on page 1-40. 1.5.5 Use of the __user_libspace static data area by the C libraries on page 1-26. Related information --fpu=name compiler option. 1.7.5 Bare machine C with floating-point processing If you want to use floating-point processing in an application without the C library, there are a number of requirements you must fulfill. These requirements are: • Re-implement __rt_raise() if you are using the heap. • Not define main(), to avoid linking in the library initialization code. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-43 1 The ARM C and C++ Libraries 1.7 Support for building an application without the C library • • • • • Write an assembly language veneer that establishes the register state required to run C. This veneer must branch to the entry function in your application. The register state required to run C primarily comprises the stack pointer. It also consists of sb, the static base register, if Read/Write PositionIndependent (RWPI) code applies. Provide your own RW/ZI initialization code. Ensure that your initialization veneer is executed by, for example, placing it in your reset handler. Use the appropriate FPU option when you build your application. Call _fp_init() to initialize the floating-point status register before performing any floating-point operations. Do not build your application with the --fpu=none option. The floating-point modes --fpmode=ieee_fixed and --fpmode=ieee_full when used with software floating-point support require a floating-point status word. In such cases, you can also define the function __rt_fp_status_addr() to return the address of a writable data word to be used instead of the floatingpoint status register. If you rely on the default library definition of __rt_fp_status_addr(), this word resides in the program data section, unless you define __user_perthread_libspace() (or in the case of legacy code that does not yet use __user_perthread_libspace(), __user_libspace() ). Related concepts 1.7.2 Creating an application as bare machine C without the C library on page 1-42. 1.7.4 Bare machine integer C on page 1-43. 1.5.4 Use of static data in the C libraries on page 1-25. Related references 1.7.1 Building an application without the C library on page 1-40. 1.5.5 Use of the __user_libspace static data area by the C libraries on page 1-26. 1.7.6 Customized C library startup code and access to C library functions If you build an application with customized startup code, you must either avoid functions that require initialization or provide the initialization and low-level support functions. When building an application without the C library, if you create an application that includes a main() function, the linker automatically includes the initialization code necessary for the execution environment. There are situations where this is not desirable or possible. For example, a system running a Real-Time Operating System (RTOS) might have its execution environment configured by the RTOS startup code. You can create an application that consists of customized startup code and still use many of the library functions. You must either: • • Avoid functions that require initialization. Provide the initialization and low-level support functions. The functions you must re-implement depend on how much of the library functionality you require: • If you want only the compiler support functions for division, structure copy, and floating-point arithmetic, you must provide __rt_raise(). This also enables very simple library functions such as those in errno.h, setjmp.h, and most of string.h to work. • If you call setlocale() explicitly, locale-dependent functions are activated. This enables you to use the atoi family, sprintf(), sscanf(), and the functions in ctype.h. • Programs that use floating-point must call _fp_init(). If you select software floating-point in -fpmode=ieee_fixed or --fpmode=ieee_full mode, the program must also provide __rt_fp_status_addr(). • Implementing high-level input/output support is necessary for functions that use fprintf() or fputs(). The high-level output functions depend on fputc() and ferror(). The high-level input functions depend on fgetc() and __backspace(). Implementing these functions and the heap enables you to use almost the entire library. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-44 1 The ARM C and C++ Libraries 1.7 Support for building an application without the C library Related concepts 1.7.2 Creating an application as bare machine C without the C library on page 1-42. 1.6.1 Using the C library with an application on page 1-34. 1.7.4 Bare machine integer C on page 1-43. 1.7.7 Using low-level functions when exploiting the C library on page 1-45. 1.7.8 Using high-level functions when exploiting the C library on page 1-45. 1.7.9 Using malloc() when exploiting the C library on page 1-46. Related references 1.7.1 Building an application without the C library on page 1-40. 1.5.5 Use of the __user_libspace static data area by the C libraries on page 1-26. 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.7.7 Using low-level functions when exploiting the C library If you are using the libraries in an application that does not have a main() function, you must reimplement some functions in the library. Caution __rt_raise() is essential if you are using the heap. Note If rand() is called, srand() must be called first. This is done automatically during library initialization but not when you avoid the library initialization. Related concepts 1.7.8 Using high-level functions when exploiting the C library on page 1-45. 1.7.2 Creating an application as bare machine C without the C library on page 1-42. 1.7.6 Customized C library startup code and access to C library functions on page 1-44. Related references 1.7.1 Building an application without the C library on page 1-40. 1.7.8 Using high-level functions when exploiting the C library High-level I/O functions can be used if the low-level functions are re-implemented. High-level I/O functions are those such as fprintf(), printf(), scanf(), puts(), fgets(), fread(), fwrite(), and perror(). Low-level functions are those such as fputc(), fgetc(), and __backspace(). Most of the formatted output functions also require a call to setlocale(). Anything that uses locale must not be called before first calling setlocale().setlocale() selects the appropriate locale. For example, setlocale(LC_ALL, "C"), where LC_ALL means that the call to setlocale() affects all locale categories, and "C" specifies the minimal environment for C translation. Locale-using functions include the functions in ctype.h and locale.h, the printf() family, the scanf() family, ato*, strto*, strcoll/strxfrm, and most of time.h. Related concepts 1.7.7 Using low-level functions when exploiting the C library on page 1-45. 1.7.2 Creating an application as bare machine C without the C library on page 1-42. 1.7.6 Customized C library startup code and access to C library functions on page 1-44. Related references 1.7.1 Building an application without the C library on page 1-40. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-45 1 The ARM C and C++ Libraries 1.7 Support for building an application without the C library 1.7.9 Using malloc() when exploiting the C library If heap support is required for bare machine C, you must implement _init_alloc() and __rt_heap_extend(). _init_alloc() must be called first to supply initial heap bounds, and __rt_heap_extend() must be provided even if it only returns failure. Without __rt_heap_extend(), certain library functionality is included that causes problems when you are writing bare machine C. Prototypes for both _init_alloc() and __rt_heap_extend() are in rt_heap.h. Related concepts 1.7.2 Creating an application as bare machine C without the C library on page 1-42. 1.7.6 Customized C library startup code and access to C library functions on page 1-44. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-46 1 The ARM C and C++ Libraries 1.8 Tailoring the C library to a new execution environment 1.8 Tailoring the C library to a new execution environment Tailoring the C library to a new execution environment involves re-implementing functions to produce an application for a new execution environment, for example, embedded in ROM or used with an RTOS. Functions whose names start with a single or double underscore are used as part of the low-level implementation. You can re-implement some of these functions. Additional information on these library functions is available in the rt_heap.h, rt_locale.h, rt_misc.h, and rt_sys.h include files and the rt_memory.s assembler file. This section contains the following subsections: • 1.8.1 Initialization of the execution environment and execution of the application on page 1-47. • 1.8.2 C++ initialization, construction and destruction on page 1-48. • 1.8.3 Exceptions system initialization on page 1-48. • 1.8.4 Emergency buffer memory for exceptions on page 1-49. • 1.8.5 Library functions called from main() on page 1-50. • 1.8.6 Program exit and the assert macro on page 1-50. 1.8.1 Initialization of the execution environment and execution of the application You can customize execution initialization by defining your own __main that branches to __rt_entry. The entry point of a program is at __main in the C library where library code: 1. Copies non-root (RO and RW) execution regions from their load addresses to their execution addresses. Also, if any data sections are compressed, they are decompressed from the load address to the execution address. 2. Zeroes ZI regions. 3. Branches to __rt_entry. If you do not want the library to perform these actions, you can define your own __main that branches to __rt_entry. IMPORT __rt_entry EXPORT __main ENTRY __main B __rt_entry END The library function __rt_entry() runs the program as follows: 1. Sets up the stack and the heap by one of a number of means that include calling __user_setup_stackheap(), calling __rt_stackheap_init(), or loading the absolute addresses of scatter-loaded regions. 2. Calls __rt_lib_init() to initialize referenced library functions, initialize the locale and, if necessary, set up argc and argv for main(). For C++, calls the constructors for any top-level objects by way of __cpp_initialize__aeabi_. 3. Calls main(), the user-level root of the application. From main(), your program might call, among other things, library functions. 4. Calls exit() with the value returned by main(). Related concepts 1.8.5 Library functions called from main() on page 1-50. 1.8 Tailoring the C library to a new execution environment on page 1-47. 1.8.2 C++ initialization, construction and destruction on page 1-48. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-47 1 The ARM C and C++ Libraries 1.8 Tailoring the C library to a new execution environment 4.23 __rt_entry on page 4-163. 4.28 __rt_lib_init() on page 4-168. 1.8.2 C++ initialization, construction and destruction The C++ Standard places certain requirements on the construction and destruction of objects with static storage duration, and the ARM C++ compiler uses the .init_array area to achieve this. The .init_array area is a const data array of self-relative pointers to functions. For example, you might have the following C++ translation unit, contained in the file test.cpp: struct T { T(); ~T(); } t; int f() { return 4; } int i = f(); This translates into the following pseudocode: AREA ||.text||, CODE, READONLY int f() { return 4; } static void __sti___8_test_cpp { // construct 't' and register its destruction __aeabi_atexit(T::T(&t), &T::~T, &__dso_handle); i = f(); } AREA ||.init_array||, DATA, READONLY DCD __sti___8_test_cpp - {PC} AREA ||.data||, DATA t % 4 i % 4 This pseudocode is for illustration only. To see the code that is generated, compile the C++ source code with armcc -c --cpp -S. The linker collects each .init_array from the various translation units together. It is important that the .init_array is accumulated in the same order. The library routine __cpp_initialize__aeabi_ is called from the C library startup code, __rt_lib_init, before main. __cpp_initialize__aeabi_ walks through the .init_array calling each function in turn. On exit, __rt_lib_shutdown calls __cxa_finalize. Usually, there is at most one function call for T::T(), symbol reference _ZN1TC1Ev, one function call for T::~T(), symbol reference _ZN1TD1Ev, one __sti__ function, and four bytes of .init_array for each translation unit. The symbol reference for the function f() is _Z1fv. There is no way to determine the initialization order between translation units. Function-local static objects with destructors are also handled using __aeabi_atexit. .init_array sections must be placed contiguously within the same region for their base and limit symbols to be accessible. If they are not, the linker generates an error. Related concepts 1.8 Tailoring the C library to a new execution environment on page 1-47. 1.8.3 Exceptions system initialization The exceptions system can be initialized either on demand (that is, when first used), or before entering main(). ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-48 1 The ARM C and C++ Libraries 1.8 Tailoring the C library to a new execution environment Initialization on demand has the advantage of not allocating heap memory unless the exceptions system is used, but has the disadvantage that it becomes impossible to throw any exception (such as std::bad_alloc) if the heap is exhausted at the time of first use. The default behavior is to initialize on demand. To initialize the exceptions system before entering main(), include the following function in the link: extern "C" void __cxa_get_globals(void); extern "C" void __ARM_exceptions_init(void) { __cxa_get_globals(); } Although you can place the call to __cxa_get_globals() directly in your code, placing it in __ARM_exceptions_init() ensures that it is called as early as possible. That is, before any global variables are initialized and before main() is entered. __ARM_exceptions_init() is weakly referenced by the library initialization mechanism, and is called if it is present as part of __rt_lib_init(). Note The exception system is initialized by calls to various library functions, for example, std::set_terminate(). Therefore, you might not have to initialize before the entry to main(). Related concepts 1.8 Tailoring the C library to a new execution environment on page 1-47. 1.8.4 Emergency buffer memory for exceptions You can choose whether or not to allocate emergency memory that is to be used for throwing a std::bad_alloc exception when the heap is exhausted. To allocate emergency memory, you must include the symbol __ARM_exceptions_buffer_required in the link. A call is then made to __ARM_exceptions_buffer_init() as part of the exceptions system initialization. The symbol is not included by default. The following routines manage the exceptions emergency buffer: • extern "C" void *__ARM_exceptions_buffer_init() Called once during runtime to allocate the emergency buffer memory. It returns a pointer to the emergency buffer memory, or NULL if no memory is allocated. • extern "C" void *__ARM_exceptions_buffer_allocate(void *buffer, size_t size) Called when an exception is about to be thrown, but there is not enough heap memory available to allocate the exceptions object. buffer is the value previously returned by __ARM_exceptions_buffer_init(), or NULL if that routine was not called. __ARM_exceptions_buffer_allocate() returns a pointer to size bytes of memory that is aligned on an eight-byte boundary, or NULL if the allocation is not possible. • extern "C" void *__ARM_exceptions_buffer_free(void *buffer, void *addr) Called to free memory possibly allocated by __ARM_exceptions_buffer_allocate(). buffer is the value previously returned by __ARM_exceptions_buffer_init(), or NULL if that routine was not called. The routine determines whether the passed address has been allocated from the emergency memory buffer, and if so, frees it appropriately, then returns a non-NULL value. If the memory at addr was not allocated by __ARM_exceptions_buffer_allocate(), the routine must return NULL. Default definitions of these routines are present in the image, but you can supply your own versions to override the defaults supplied by the library. The default routines reserve enough space for a single std::bad_alloc exceptions object. If you do not require an emergency buffer, it is safe to redefine all these routines to return only NULL. Related concepts 1.8 Tailoring the C library to a new execution environment on page 1-47. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-49 1 The ARM C and C++ Libraries 1.8 Tailoring the C library to a new execution environment 1.8.5 Library functions called from main() The function main() can call a number of user-customizable functions in the C library. The function main() is the user-level root of the application. It requires the execution environment to be initialized and input/output functions to be capable of being called. While in main() the program might perform one of the following actions that calls user-customizable functions in the C library: • Extend the stack or heap. • Call library functions that require a callout to a user-defined function, for example __rt_fp_status_addr() or clock(). • Call library functions that use locale or CTYPE. • Perform floating-point calculations that require the floating-point unit or floating-point library. • Input or output directly through low-level functions, for example putc(), or indirectly through highlevel input/output functions and input/output support functions, for example, fprintf() or sys_open(). • Raise an error or other signal, for example ferror. Related concepts 1.8.1 Initialization of the execution environment and execution of the application on page 1-47. 1.8 Tailoring the C library to a new execution environment on page 1-47. 1.9 Assembler macros that tailor locale functions in the C library on page 1-52. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.21 Tailoring non-input/output C library functions on page 1-81. 1.10 Modification of C library functions for error signaling, error handling, and program exit on page 1-61. 1.8.6 Program exit and the assert macro A program can exit normally at the end of main() or it can exit prematurely because of an error. The behavior of the assert macro depends on a number of conditions: 1. If the NDEBUG macro is defined (on the command line or as part of a source file), the assert macro has no effect. 2. If the NDEBUG macro is not defined, the assert expression (the expression given to the assert macro) is evaluated. If the result is TRUE, that is != 0, the assert macro has no more effect. 3. If the assert expression evaluates to FALSE, the assert macro calls the __aeabi_assert() function if any of the following are true: • You are compiling with --strict. • You are using -O0 or -O1. • __ASSERT_MSG is defined. • _AEABI_PORTABILITY_LEVEL is defined and not 0. 4. If the assert expression evaluates to FALSE and the conditions specified in point 3 do not apply, the assert macro calls abort(). Then: a. abort() calls __rt_raise(). b. If __rt_raise() returns, abort() tries to finalize the library. If you are creating an application that does not use the library, __aeabi_assert() works if you reimplement abort() and the stdio functions. Another solution for retargeting is to re-implement the __aeabi_assert() function itself. The function prototype is: void __aeabi_assert(const char *expr, const char *file, int line); where: • expr points to the string representation of the expression that was not TRUE. • file and line identify the source location of the assertion. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-50 1 The ARM C and C++ Libraries 1.8 Tailoring the C library to a new execution environment The behavior for __aeabi_assert() supplied in the ARM C library is to print a message on stderr and call abort(). Related concepts 1.8 Tailoring the C library to a new execution environment on page 1-47. Related concepts 1.8.2 C++ initialization, construction and destruction on page 1-48. 1.8.5 Library functions called from main() on page 1-50. 1.8.1 Initialization of the execution environment and execution of the application on page 1-47. 1.8.4 Emergency buffer memory for exceptions on page 1-49. 1.8.6 Program exit and the assert macro on page 1-50. 1.8.3 Exceptions system initialization on page 1-48. Related references 4.23 __rt_entry on page 4-163. 4.25 __rt_exit() on page 4-165. 4.28 __rt_lib_init() on page 4-168. 4.29 __rt_lib_shutdown() on page 4-169. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-51 1 The ARM C and C++ Libraries 1.9 Assembler macros that tailor locale functions in the C library 1.9 Assembler macros that tailor locale functions in the C library Applications use locales when they display or process data that depends on the local language or region, for example, character set, monetary symbols, decimal point, time, and date. Locale-related functions are declared in the include file, rt_locale. This section contains the following subsections: • 1.9.1 Link time selection of the locale subsystem in the C library on page 1-52. • 1.9.2 Runtime selection of the locale subsystem in the C library on page 1-53. • 1.9.3 Definition of locale data blocks in the C library on page 1-53. • 1.9.4 LC_CTYPE data block on page 1-55. • 1.9.5 LC_COLLATE data block on page 1-57. • 1.9.6 LC_MONETARY data block on page 1-58. • 1.9.7 LC_NUMERIC data block on page 1-58. • 1.9.8 LC_TIME data block on page 1-59. 1.9.1 Link time selection of the locale subsystem in the C library The locale subsystem of the C library can be selected at link time or can be extended to be selectable at runtime. The following list describes the use of locale categories by the library: • The default implementation of each locale category is for the C locale. The library also provides an alternative, ISO8859-1 (Latin-1 alphabet) implementation of each locale category that you can select at link time. • Both the C and ISO8859-1 default implementations usually provide one locale for each category to select at runtime. • You can replace each locale category individually. • You can include as many locales in each category as you choose and you can name your locales as you choose. • Each locale category uses one word in the private static data of the library. • The locale category data is read-only and position independent. • scanf() forces the inclusion of the LC_CTYPE locale category, but in either of the default locales this adds only 260 bytes of read-only data to several kilobytes of code. ISO8859-1 implementation The default implementation of each locale category is for the C locale. The library also provides an alternative, ISO8859-1 (Latin-1 alphabet) implementation of each locale category that you can select at link time. The following table shows the ISO8859-1 (Latin-1 alphabet) locale categories. Table 1-6 Default ISO8859-1 locales Symbol Description __use_iso8859_ctype Selects the ISO8859-1 (Latin-1) classification of characters. This is essentially 7-bit ASCII, except that the character codes 160-255 represent a selection of useful European punctuation characters, letters, and accented letters. __use_iso8859_collate Selects the strcoll/strxfrm collation table appropriate to the Latin-1 alphabet. The default C locale does not require a collation table. __use_iso8859_monetary Selects the Sterling monetary category using Latin-1 coding. __use_iso8859_numeric Selects separation of thousands with commas in the printing of numeric values. __use_iso8859_locale Selects all the ISO8859-1 selections described in this table. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-52 1 The ARM C and C++ Libraries 1.9 Assembler macros that tailor locale functions in the C library There is no ISO8859-1 version of the LC_TIME category. Shift-JIS and UTF-8 implementation The Shift-JIS and UTF-8 locales let you use Japanese and Unicode characters. The following table shows the Shift-JIS (Japanese characters) or UTF-8 (Unicode characters) locale categories. Table 1-7 Default Shift-JIS and UTF-8 locales Function Description __use_sjis_ctype Sets the character set to the Shift-JIS multibyte encoding of Japanese characters __use_utf8_ctype Sets the character set to the UTF-8 multibyte encoding of all Unicode characters The following list describes the effects of Shift-JIS and UTF-8 encoding: • The ordinary ctype functions behave correctly on any byte value that is a self-contained character in Shift-JIS. For example, half-width katakana characters that Shift-JIS encodes as single bytes between 0xA6 and 0xDF are treated as alphabetic by isalpha(). • • UTF-8 encoding uses the same set of self-contained characters as the ASCII character set. The multibyte conversion functions such as mbrtowc(), mbsrtowcs(), and wcrtomb(), all convert between wide strings in Unicode and multibyte character strings in Shift-JIS or UTF-8. printf("%ls") converts a Unicode wide string into Shift-JIS or UTF-8 output, and scanf("%ls") converts Shift-JIS or UTF-8 input into a Unicode wide string. Related concepts Shift-JIS and UTF-8 implementation on page 1-53. Related references ISO8859-1 implementation on page 1-52. 1.9.2 Runtime selection of the locale subsystem in the C library The C library function setlocale() selects a locale at runtime for the locale category, or categories, specified in its arguments. It does this by selecting the requested locale separately in each locale category. In effect, each locale category is a small filing system containing an entry for each locale. The rt_locale.h and rt_locale.s header files describe what must be implemented and provide some useful support macros. Related references 4.32 setlocale() on page 4-172. 1.9.3 Definition of locale data blocks in the C library Locale data blocks let you customize your own locales. The locale data blocks are defined using a set of assembly language macros provided in rt_locale.s. Therefore, the recommended way to define locale blocks is by writing an assembly language source file. The ARM Compiler toolchain provides a set of macros for each type of locale data block, for example LC_CTYPE, LC_COLLATE, LC_MONETARY, LC_NUMERIC, and LC_TIME. You define each locale block in the same way with a _begin macro, some data macros, and an _end macro. Beginning the definition of a locale block To begin defining your locale block, call the _begin macro. This macro takes two arguments, a prefix and the textual name, as follows: ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-53 1 The ARM C and C++ Libraries 1.9 Assembler macros that tailor locale functions in the C library LC_TYPE_begin prefix, name where: TYPE is one of the following: • CTYPE • COLLATE • MONETARY • NUMERIC • TIME. prefix is the prefix for the assembler symbols defined within the locale data name is the textual name for the locale data. Specifying the data for a locale block To specify the data for your locale block, call the macros for that locale type in the order specified for that particular locale type. The syntax is as follows: LC_TYPE_function Where: TYPE is one of the following: • CTYPE • COLLATE • MONETARY • NUMERIC • TIME. function is a specific function, table(), full_wctype(), or multibyte(), related to your locale data. When specifying locale data, you must call the macro repeatedly for each respective function. Ending the definition of a locale block To complete the definition of your locale data block, you call the _end macro. This macro takes no arguments, as follows: LC_TYPE_end where: TYPE is one of the following: • CTYPE • COLLATE • MONETARY • NUMERIC • TIME. Example of a fixed locale block To write a fixed function that always returns the same locale, you can use the _start symbol name defined by the macros. The following shows how this is implemented for the CTYPE locale: GET rt_locale.s AREA my_locales, DATA, READONLY LC_CTYPE_begin my_ctype_locale, "MyLocale" ... ; include other LC_CTYPE_xxx macros here ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-54 1 The ARM C and C++ Libraries 1.9 Assembler macros that tailor locale functions in the C library LC_CTYPE_end AREA my_locale_func, CODE, READONLY _get_lc_ctype FUNCTION LDR r0, =my_ctype_locale_start BX lr ENDFUNC Example of multiple contiguous locale blocks Contiguous locale blocks suitable for passing to the _findlocale() function must be declared in sequence. You must call the macro LC_index_end to end the sequence of locale blocks. The following shows how this is implemented for the CTYPE locale: GET rt_locale.s AREA my_locales, DATA, READONLY my_ctype_locales LC_CTYPE_begin my_first_ctype_locale, "MyLocale1" ... ; include other LC_CTYPE_xxx macros here LC_CTYPE_end LC_CTYPE_begin my_second_ctype_locale, "MyLocale2" ... ; include other LC_CTYPE_xxx macros here LC_CTYPE_end LC_index_end AREA my_locale_func, CODE, READONLY IMPORT _findlocale _get_lc_ctype FUNCTION LDR r0, =my_ctype_locales B _findlocale ENDFUNC Related references 1.9.4 LC_CTYPE data block on page 1-55. 1.9.5 LC_COLLATE data block on page 1-57. 1.9.6 LC_MONETARY data block on page 1-58. 1.9.7 LC_NUMERIC data block on page 1-58. 1.9.8 LC_TIME data block on page 1-59. 1.9.4 LC_CTYPE data block The LC_CTYPE data block configures character classification and conversion. When defining a locale data block in the C library, the macros that define an LC_CTYPE data block are as follows: 1. Call LC_CTYPE_begin with a symbol name and a locale name. 2. Call LC_CTYPE_table repeatedly to specify 256 table entries. LC_CTYPE_table takes a single argument in quotes. This must be a comma-separated list of table entries. Each table entry describes one of the 256 possible characters, and can be either an illegal character (IL) or the bitwise OR of one or more of the following flags: __S whitespace characters __P punctuation characters __B printable space characters __L lowercase letters __U uppercase letters __N decimal digits __C control characters __X hexadecimal digit letters A-F and a-f ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-55 1 The ARM C and C++ Libraries 1.9 Assembler macros that tailor locale functions in the C library __A alphabetic but neither uppercase nor lowercase, such as Japanese katakana. Note A printable space character is defined as any character where the result of both isprint() and isspace() is true. __A must not be specified for the same character as either __N or __X. 3. If required, call one or both of the following optional macros: • LC_CTYPE_full_wctype. Calling this macro without arguments causes the C99 wide-character ctype functions (iswalpha(), iswupper(), ...) to return useful values across the full range of Unicode when this LC_CTYPE locale is active. If this macro is not specified, the wide ctype functions treat the first 256 wchar_t values as the same as the 256 char values, and the rest of the wchar_t range as containing illegal characters. • LC_CTYPE_multibyte defines this locale to be a multibyte character set. Call this macro with three arguments. The first two arguments are the names of functions that perform conversion between the multibyte character set and Unicode wide characters. The last argument is the value that must be taken by the C macro MB_CUR_MAX for the respective character set. The two function arguments have the following prototypes: size_t internal_mbrtowc(wchar_t *pwc, char c, mbstate_t *pstate); size_t internal_wcrtomb(char *s, wchar_t w, mbstate_t *pstate); internal_mbrtowc() takes one byte, c, as input, and updates the mbstate_t pointed to by pstate as a result of reading that byte. If the byte completes the encoding of a multibyte character, it writes the corresponding wide character into the location pointed to by pwc, and returns 1 to indicate that it has done so. If not, it returns -2 to indicate the state change of mbstate_t and that no character is output. Otherwise, it returns -1 to indicate that the encoded input is invalid. internal_wcrtomb() takes one wide character, w, as input, and writes some number of bytes into the memory pointed to by s. It returns the number of bytes output, or -1 to indicate that the input character has no valid representation in the multibyte character set. 4. Call LC_CTYPE_end, without arguments, to finish the locale block definition. Example LC_CTYPE data block LC_CTYPE_begin utf8_ctype, "UTF-8" ; ; Single-byte characters in the low half of UTF-8 are exactly ; the same as in the normal "C" locale. LC_CTYPE_table "__C, __C, __C, __C, __C, __C, __C, __C, __C" ; 0x00-0x08 LC_CTYPE_table "__C|__S, __C|__S, __C|__S, __C|__S, __C|__S" ; 0x09-0x0D(BS,LF,VT,FF,CR) LC_CTYPE_table "__C, __C, __C, __C, __C, __C, __C, __C, __C" ; 0x0E-0x16 LC_CTYPE_table "__C, __C, __C, __C, __C, __C, __C, __C, __C" ; 0x17-0x1F LC_CTYPE_table "__B|__S" ; space LC_CTYPE_table "__P, __P, __P, __P, __P, __P, __P, __P" ; !"#$%&'( LC_CTYPE_table "__P, __P, __P, __P, __P, __P, __P" ; )*+,-./ LC_CTYPE_table "__N, __N, __N, __N, __N, __N, __N, __N, __N, __N" ; 0-9 LC_CTYPE_table "__P, __P, __P, __P, __P, __P, __P" ; :;<=>?@ LC_CTYPE_table "__U|__X, __U|__X, __U|__X, __U|__X, __U|__X, __U|__X" ; A-F LC_CTYPE_table "__U, __U, __U, __U, __U, __U, __U, __U, __U, __U" ; G-P LC_CTYPE_table "__U, __U, __U, __U, __U, __U, __U, __U, __U, __U" ; Q-Z LC_CTYPE_table "__P, __P, __P, __P, __P, __P" ; [\]^_` LC_CTYPE_table "__L|__X, __L|__X, __L|__X, __L|__X, __L|__X, __L|__X" ; a-f LC_CTYPE_table "__L, __L, __L, __L, __L, __L, __L, __L, __L, __L" ; g-p LC_CTYPE_table "__L, __L, __L, __L, __L, __L, __L, __L, __L, __L" ; q-z LC_CTYPE_table "__P, __P, __P, __P" ; {|}~ LC_CTYPE_table "__C" ; 0x7F ; ; Nothing in the top half of UTF-8 is valid on its own as a ; single-byte character, so they are all illegal characters (IL). LC_CTYPE_table "IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL" LC_CTYPE_table "IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL" ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-56 1 The ARM C and C++ Libraries 1.9 Assembler macros that tailor locale functions in the C library LC_CTYPE_table "IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL" LC_CTYPE_table "IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL" LC_CTYPE_table "IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL" LC_CTYPE_table "IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL" LC_CTYPE_table "IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL" LC_CTYPE_table "IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL,IL" ; ; The UTF-8 ctype locale wants the full version of wctype. LC_CTYPE_full_wctype ; ; UTF-8 is a multibyte locale, so we must specify some ; conversion functions. MB_CUR_MAX is 6 for UTF-8 (the lead ; bytes 0xFC and 0xFD are each followed by five continuation ; bytes). ; ; The implementations of the conversion functions are not ; provided in this example. ; IMPORT utf8_mbrtowc IMPORT utf8_wcrtomb LC_CTYPE_multibyte utf8_mbrtowc, utf8_wcrtomb, 6 LC_CTYPE_end Related references 1.9.3 Definition of locale data blocks in the C library on page 1-53. 1.9.5 LC_COLLATE data block The LC_COLLATE data block configures collation of strings. When defining a locale data block in the C library, the macros that define an LC_COLLATE data block are as follows: 1. Call LC_COLLATE_begin with a symbol name and a locale name. 2. Call one of the following alternative macros: • Call LC_COLLATE_table repeatedly to specify 256 table entries. LC_COLLATE_table takes a single argument in quotes. This must be a comma-separated list of table entries. Each table entry describes one of the 256 possible characters, and can be a number indicating its position in the sorting order. For example, if character A is intended to sort before B, then entry 65 (corresponding to A) in the table, must be smaller than entry 66 (corresponding to B). • Call LC_COLLATE_no_table without arguments. This indicates that the collation order is the same as the string comparison order. Therefore, strcoll() and strcmp() are identical. 3. Call LC_COLLATE_end, without arguments, to finish the locale block definition. Example LC_COLLATE data block LC_COLLATE_begin LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table LC_COLLATE_table ARM DUI0378G_02 iso88591_collate, "ISO8859-1" "0x00, 0x01, 0x02, 0x03, 0x04, "0x08, 0x09, 0x0a, 0x0b, 0x0c, "0x10, 0x11, 0x12, 0x13, 0x14, "0x18, 0x19, 0x1a, 0x1b, 0x1c, "0x20, 0x21, 0x22, 0x23, 0x24, "0x28, 0x29, 0x2a, 0x2b, 0x2c, "0x30, 0x31, 0x32, 0x33, 0x34, "0x38, 0x39, 0x3a, 0x3b, 0x3c, "0x40, 0x41, 0x49, 0x4a, 0x4c, "0x54, 0x55, 0x5a, 0x5b, 0x5c, "0x67, 0x68, 0x69, 0x6a, 0x6b, "0x73, 0x74, 0x76, 0x79, 0x7a, "0x7e, 0x7f, 0x87, 0x88, 0x8a, "0x92, 0x93, 0x98, 0x99, 0x9a, "0xa5, 0xa6, 0xa7, 0xa8, 0xaa, "0xb2, 0xb3, 0xb6, 0xb9, 0xba, "0xbe, 0xbf, 0xc0, 0xc1, 0xc2, "0xc6, 0xc7, 0xc8, 0xc9, 0xca, "0xce, 0xcf, 0xd0, 0xd1, 0xd2, "0xd6, 0xd7, 0xd8, 0xd9, 0xda, "0xde, 0xdf, 0xe0, 0xe1, 0xe2, "0xe6, 0xe7, 0xe8, 0xe9, 0xea, "0xee, 0xef, 0xf0, 0xf1, 0xf2, "0xf6, 0xf7, 0xf8, 0xf9, 0xfa, "0x42, 0x43, 0x44, 0x45, 0x46, "0x4e, 0x4f, 0x50, 0x51, 0x56, "0x77, 0x5f, 0x61, 0x62, 0x63, "0x66, 0x6d, 0x6e, 0x6f, 0x70, "0x80, 0x81, 0x82, 0x83, 0x84, 0x05, 0x0d, 0x15, 0x1d, 0x25, 0x2d, 0x35, 0x3d, 0x4d, 0x5d, 0x6c, 0x7b, 0x8b, 0x9b, 0xab, 0xbb, 0xc3, 0xcb, 0xd3, 0xdb, 0xe3, 0xeb, 0xf3, 0xfb, 0x47, 0x57, 0x64, 0x75, 0x85, 0x06, 0x0e, 0x16, 0x1e, 0x26, 0x2e, 0x36, 0x3e, 0x52, 0x5e, 0x71, 0x7c, 0x90, 0x9c, 0xb0, 0xbc, 0xc4, 0xcc, 0xd4, 0xdc, 0xe4, 0xec, 0xf4, 0xfc, 0x48, 0x58, 0x65, 0x78, 0x86, Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 0x07" 0x0f" 0x17" 0x1f" 0x27" 0x2f" 0x37" 0x3f" 0x53" 0x60" 0x72" 0x7d" 0x91" 0x9e" 0xb1" 0xbd" 0xc5" 0xcd" 0xd5" 0xdd" 0xe5" 0xed" 0xf5" 0xfd" 0x4b" 0x59" 0xfe" 0xa9" 0x89" 1-57 1 The ARM C and C++ Libraries 1.9 Assembler macros that tailor locale functions in the C library LC_COLLATE_table "0x8c, 0x8d, 0x8e, 0x8f, 0x94, 0x95, 0x96, 0x97" LC_COLLATE_table "0xb7, 0x9d, 0x9f, 0xa0, 0xa1, 0xa2, 0xa3, 0xff" LC_COLLATE_table "0xa4, 0xac, 0xad, 0xae, 0xaf, 0xb4, 0xb8, 0xb5" LC_COLLATE_end Related references 1.9.3 Definition of locale data blocks in the C library on page 1-53. ISO8859-1 implementation on page 1-52. 1.9.6 LC_MONETARY data block The LC_MONETARY data block configures formatting of monetary values. When defining a locale data block in the C library, the macros that define an LC_MONETARY data block are as follows: 1. Call LC_MONETARY_begin with a symbol name and a locale name. 2. Call the LC_MONETARY data macros as follows: a. Call LC_MONETARY_fracdigits with two arguments: frac_digits and int_frac_digits from the lconv structure. b. Call LC_MONETARY_positive with four arguments: p_cs_precedes, p_sep_by_space, p_sign_posn and positive_sign. c. Call LC_MONETARY_negative with four arguments: n_cs_precedes, n_sep_by_space, n_sign_posn and negative_sign. d. Call LC_MONETARY_currsymbol with two arguments: currency_symbol and int_curr_symbol. e. Call LC_MONETARY_point with one argument: mon_decimal_point. f. Call LC_MONETARY_thousands with one argument: mon_thousands_sep. g. Call LC_MONETARY_grouping with one argument: mon_grouping. 3. Call LC_MONETARY_end, without arguments, to finish the locale block definition. Example LC_MONETARY data block LC_MONETARY_begin c_monetary, "C" LC_MONETARY_fracdigits 255, 255 LC_MONETARY_positive 255, 255, 255, "" LC_MONETARY_negative 255, 255, 255, "" LC_MONETARY_currsymbol "", "" LC_MONETARY_point "" LC_MONETARY_thousands "" LC_MONETARY_grouping "" LC_MONETARY_end Related references 1.9.3 Definition of locale data blocks in the C library on page 1-53. 1.9.7 LC_NUMERIC data block The LC_NUMERIC data block configures formatting of numeric values that are not monetary. When defining a locale data block in the C library, the macros that define an LC_NUMERIC data block are as follows: 1. Call LC_NUMERIC_begin with a symbol name and a locale name. 2. Call the LC_NUMERIC data macros as follows: a. Call LC_NUMERIC_point with one argument: decimal_point from lconv structure. b. Call LC_NUMERIC_thousands with one argument: thousands_sep. c. Call LC_NUMERIC_grouping with one argument: grouping. 3. Call LC_NUMERIC_end, without arguments, to finish the locale block definition. Example LC_NUMERIC data block LC_NUMERIC_begin c_numeric, "C" LC_NUMERIC_point "." LC_NUMERIC_thousands "" LC_NUMERIC_grouping "" LC_NUMERIC_end ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-58 1 The ARM C and C++ Libraries 1.9 Assembler macros that tailor locale functions in the C library Related references 1.9.3 Definition of locale data blocks in the C library on page 1-53. 1.9.8 LC_TIME data block The LC_TIME data block configures formatting of date and time values. When defining a locale data block in the C library, the macros that define an LC_TIME data block are as follows: 1. Call LC_TIME_begin with a symbol name and a locale name. 2. Call the LC_TIME data macros as follows: a. Call LC_TIME_week_short seven times to provide the short names for the days of the week. Sunday being the first day. Then call LC_TIME_week_long and repeat the process for long names. b. Call LC_TIME_month_short twelve times to provide the short names for the days of the month. Then call LC_TIME_month_long and repeat the process for long names. c. Call LC_TIME_am_pm with two arguments that are respectively the strings representing morning and afternoon. d. Call LC_TIME_formats with three arguments that are respectively the standard date/time format used in strftime("%c"), the standard date format strftime("%x"), and the standard time format strftime("%X"). These strings must define the standard formats in terms of other simpler strftime primitives. The example below shows that the standard date/time format is permitted to reference the other two formats. e. Call LC_TIME_c99format with a single string that is the standard 12-hour time format used in strftime("%r") as defined in C99. 3. Call LC_TIME_end, without arguments, to finish the locale block definition. Example LC_TIME data block LC_TIME_begin c_time, "C" LC_TIME_week_short "Sun" LC_TIME_week_short "Mon" LC_TIME_week_short "Tue" LC_TIME_week_short "Wed" LC_TIME_week_short "Thu" LC_TIME_week_short "Fri" LC_TIME_week_short "Sat" LC_TIME_week_long "Sunday" LC_TIME_week_long "Monday" LC_TIME_week_long "Tuesday" LC_TIME_week_long "Wednesday" LC_TIME_week_long "Thursday" LC_TIME_week_long "Friday" LC_TIME_week_long "Saturday" LC_TIME_month_short "Jan" LC_TIME_month_short "Feb" LC_TIME_month_short "Mar" LC_TIME_month_short "Apr" LC_TIME_month_short "May" LC_TIME_month_short "Jun" LC_TIME_month_short "Jul" LC_TIME_month_short "Aug" LC_TIME_month_short "Sep" LC_TIME_month_short "Oct" LC_TIME_month_short "Nov" LC_TIME_month_short "Dec" LC_TIME_month_long "January" LC_TIME_month_long "February" LC_TIME_month_long "March" LC_TIME_month_long "April" LC_TIME_month_long "May" LC_TIME_month_long "June" LC_TIME_month_long "July" LC_TIME_month_long "August" LC_TIME_month_long "September" LC_TIME_month_long "October" LC_TIME_month_long "November" LC_TIME_month_long "December" LC_TIME_am_pm "AM", "PM" LC_TIME_formats "%a %b %e %T %Y", "%m/%d/%y", "%H:%M:%S" LC_TIME_c99format "%I:%M:%S %p" LC_TIME_end ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-59 1 The ARM C and C++ Libraries 1.9 Assembler macros that tailor locale functions in the C library Related references 1.9.3 Definition of locale data blocks in the C library on page 1-53. Related references 1.6.8 C library API definitions for targeting a different environment on page 1-38. 1.7.1 Building an application without the C library on page 1-40. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-60 1 The ARM C and C++ Libraries 1.10 Modification of C library functions for error signaling, error handling, and program exit 1.10 Modification of C library functions for error signaling, error handling, and program exit All trap or error signals raised by the C library go through the __raise() function. You can reimplement this function or the lower-level functions that it uses. Caution The IEEE 754 standard for floating-point processing states that the default response to an exception is to proceed without a trap. You can modify floating-point error handling by tailoring the functions and definitions in fenv.h. The rt_misc.h header file contains more information on error-related functions. The following table shows the trap and error-handling functions. Table 1-8 Trap and error handling Function Description _sys_exit() Called, eventually, by all exits from the library. errno Is a static variable used with error handling. __rt_errno_addr() Is called to obtain the address of the variable errno. __raise() Raises a signal to indicate a runtime anomaly. __rt_raise() Raises a signal to indicate a runtime anomaly. __default_signal_handler() Displays an error indication to the user. _ttywrch() Writes a character to the console. The default implementation of _ttywrch() is semihosted and, therefore, uses semihosting calls. __rt_fp_status_addr() This function is called to obtain the address of the floating-point status word. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. 1.6.7 Indirect semihosting C library function dependencies on page 1-37. 1.7.1 Building an application without the C library on page 1-40. 4.41 _sys_exit() on page 4-182. 4.6 errno on page 4-145. 4.19 __raise() on page 4-159. 4.30 __rt_raise() on page 4-170. 4.5 __default_signal_handler() on page 4-144. 4.51 _ttywrch() on page 4-192. 4.26 __rt_fp_status_addr() on page 4-166. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-61 1 The ARM C and C++ Libraries 1.11 Stack and heap memory allocation and the ARM C and C++ libraries 1.11 Stack and heap memory allocation and the ARM C and C++ libraries The ARM C and C++ libraries require you to specify where the stack pointer begins, but specifying the heap is optional. However, some library functions use the heap, either explicitly (for example malloc) or implicitly (for example fopen). If you are providing a heap, you must: • • • Understand the heap usage requirements of the ARM C and C++ libraries. Configure the size and placement of the heap. Consider which heap implementation you want to use. If you are not providing a heap, you must: • Understand the heap usage requirements of the ARM C and C++ libraries. • Understand how to avoid or reimplement the heap-using functions. This section contains the following subsections: • 1.11.1 Library heap usage requirements of the ARM C and C++ libraries on page 1-62. • 1.11.2 Choosing a heap implementation for memory allocation functions on page 1-63. • 1.11.3 Stack pointer initialization and heap bounds on page 1-64. • 1.11.4 Legacy support for __user_initial_stackheap() on page 1-66. • 1.11.5 Avoiding the heap and heap-using library functions supplied by ARM on page 1-67. 1.11.1 Library heap usage requirements of the ARM C and C++ libraries Functions such as malloc() and other dynamic memory allocation functions explicitly allocate memory when used. However, some library functions and mechanisms implicitly allocate memory from the heap. If heap usage requirements are significant to your code development (for example, you might be developing code for an embedded system with a tiny memory footprint), you must be aware of both implicit and explicit heap requirements. In C standardlib, implicit heap usage occurs as a result of: • • Calling the library function fopen() and the first time that an I/O operation is applied to the resulting stream. Passing command-line arguments into the main() function. The size of heap memory allocated for fopen() is 80 bytes for the FILE structure. When the first I/O operation occurs, and not until the operation occurs, an additional default of 512 bytes of heap memory is allocated for a buffer associated with the operation. You can reconfigure the size of this buffer using setvbuf(). When fclose() is called, the default 80 bytes of memory is kept on a freelist for possible re-use. The 512-byte buffer is freed on fclose(). Declaring main() to take arguments requires 256 bytes of implicitly allocated memory from the heap. This memory is never freed because it is required for the duration of main(). In microlib, main() must not be declared to take arguments, so this heap usage requirement only applies to standardlib. In the standardlib context, it only applies if you have a heap. Note The memory sizes quoted might change in future releases. Related concepts 2.3 Library heap usage requirements of microlib on page 2-100. Related references 1.11.5 Avoiding the heap and heap-using library functions supplied by ARM on page 1-67. 1.11.2 Choosing a heap implementation for memory allocation functions on page 1-63. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-62 1 The ARM C and C++ Libraries 1.11 Stack and heap memory allocation and the ARM C and C++ libraries 1.11.2 Choosing a heap implementation for memory allocation functions malloc(), realloc(), calloc(), and free() are built on a heap abstract data type. You can choose between Heap1 or Heap2, the two provided heap implementations. The available heap implementations are: • Heap1, the default implementation, implements the smallest and simplest heap manager. • Heap2 provides an implementation with the performance cost of malloc() or free() growing logarithmically with the number of free blocks. Note The default implementations of malloc(), realloc(), and calloc() maintain an eight-byte aligned heap. Heap1 Heap1, the default implementation, implements the smallest and simplest heap manager. The heap is managed as a single-linked list of free blocks held in increasing address order. The allocation policy is first-fit by address. This implementation has low overheads, but the performance cost of malloc() or free() grows linearly with the number of free blocks. The smallest block that can be allocated is four bytes and there is an additional overhead of four bytes. If you expect more than 100 unallocated blocks it is recommended that you use Heap2. Heap2 Heap2 provides an implementation with the performance cost of malloc() or free() growing logarithmically with the number of free blocks. The allocation policy is first-fit by address. The smallest block that can be allocated is 12 bytes and there is an additional overhead of 4 bytes. Heap2 is recommended when you require near constant-time performance in the presence of hundreds of free blocks. To select the alternative standard implementation, use either of the following: • • IMPORT __use_realtime_heap from assembly language. #pragma import(__use_realtime_heap) from C. The Heap2 real-time heap implementation must know the maximum address space that the heap can span. The smaller the address range, the more efficient the algorithm is. By default, the heap extent is taken to be 16MB starting at the beginning of the heap (defined as the start of the first chunk of memory given to the heap manager by __rt_initial_stackheap() or __rt_heap_extend()). The heap bounds are given by: struct __heap_extent { unsigned base, range; }; __value_in_regs struct __heap_extent __user_heap_extent( unsigned defaultbase, unsigned defaultsize); The function prototype for __user_heap_extent() is in rt_misc.h. (The Heap1 algorithm does not require the bounds on the heap extent. Therefore, it never calls this function.) You must implement __user_heap_extent() if: • • ARM DUI0378G_02 You require a heap to span more than 16MB of address space. Your memory model can supply a block of memory at a lower address than the first one supplied. Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-63 1 The ARM C and C++ Libraries 1.11 Stack and heap memory allocation and the ARM C and C++ libraries If you know in advance that the address space bounds of your heap are small, you do not have to implement __user_heap_extent(), but it does speed up the heap algorithms if you do. The input parameters are the default values that are used if this routine is not defined. You can, for example, leave the default base value unchanged and only adjust the size. Note The size field returned must be a power of two. The library does not check this and fails in unexpected ways if this requirement is not met. If you return a size of zero, the extent of the heap is set to 4GB. Related references 1.11.5 Avoiding the heap and heap-using library functions supplied by ARM on page 1-67. 4.2 alloca() on page 4-141. 1.11.3 Stack pointer initialization and heap bounds The C library requires you to specify where the stack pointer begins. If you intend to use ARM library functions that use the heap, for example, malloc(), calloc(), or if you define argc and argv command-line arguments for main(), the C library also requires you to specify which region of memory the heap is initially expected to use. You must always specify where the stack pointer begins. The initial stack pointer must be aligned to a multiple of eight bytes. You might have to configure the heap if, for example: • • You intend to use ARM library functions that use the heap, for example, malloc(), calloc(). You define argc and argv command-line arguments for main() If you are using the C library's initialization code, use any of the following methods to configure the stack and heap: • • • Use the symbols __initial_sp, __heap_base, and __heap_limit. Use a scatter file to define ARM_LIB_STACKHEAP, ARM_LIB_STACK, or ARM_LIB_HEAP regions. Implement __user_setup_stackheap() or __user_initial_stackheap(). Note The first two methods are the only methods that microlib supports for defining where the stack pointer starts and for defining the heap bounds. If you are not using the C library's initialization code (see 1.7.1 Building an application without the C library on page 1-40), use the following method to configure the stack and heap: • Set up the stack pointer manually at your application's entry point. • Call _init_alloc() to set up an initial heap region, and implement __rt_heap_extend() if you need to add memory to it later. Configuring the stack and heap with symbols Define the symbol __initial_sp to point to the top of the stack. If using the heap, also define symbols __heap_base and __heap_limit. You can define these symbols in an assembly language file, or by using the embedded assembler in C. For example: __asm void dummy_function(void) { EXPORT __initial_sp EXPORT __heap_base EXPORT __heap_limit __initial_sp EQU STACK_BASE __heap_base EQU HEAP_BASE ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-64 1 The ARM C and C++ Libraries 1.11 Stack and heap memory allocation and the ARM C and C++ libraries __heap_limit EQU (HEAP_BASE + HEAP_SIZE) } The constants STACK_BASE, HEAP_BASE and HEAP_SIZE can be defined in a header file, for example stack.h, as follows: /* stack.h */ #define HEAP_BASE 0x20100000 /* Example memory addresses */ #define STACK_BASE 0x20200000 #define HEAP_SIZE ((STACK_BASE-HEAP_BASE)/2) #define STACK_SIZE ((STACK_BASE-HEAP_BASE)/2) Note This method of specifying the initial stack pointer and heap bounds is supported by both the standard C library (standardlib) and the micro C library (microlib). Configuring the stack and heap with a scatter file In a scatter file, either: • Define ARM_LIB_STACK and ARM_LIB_HEAP regions. • If you do not intend to use the heap, only define an ARM_LIB_STACK region. Define an ARM_LIB_STACKHEAP region. If you define an ARM_LIB_STACKHEAP region, the stack starts at the top of that region. The heap starts at the bottom. Configuring the stack and heap with __user_setup_stackheap() or __user_initial_stackheap() Implement __user_setup_stackheap() to set up the stack pointer and return the bounds of the initial heap region. If you are using legacy code that uses __user_initial_stackheap(), and you do not want to replace __user_initial_stackheap() with __user_setup_stackheap(), continue to use __user_initial_stackheap(). Note ARM recommends that you switch to using __user_setup_stackheap() if you are still using __user_initial_stackheap(), unless your implementation of __user_initial_stackheap() is: • Specialized in some way such that it is complex enough to require its own temporary stack to run on before it has created the proper stack. • Has some user-specific special requirement that means it has to be implemented in C rather than in assembly language. Configuring the heap from bare machine C using _init_alloc and __rt_heap_extend If you are using a heap implementation from bare machine C (that is an application that does not define main() and does not initialize the C library) you must define the base and top of the heap as well as providing a heap extension function. 1. Call _init_alloc(base, top) to define the base and top of the memory you want to manage as a heap. Note The parameters of _init_alloc(base, top) must be eight-byte aligned. 2. Define the function unsigned __rt_heap_extend(unsigned size, void **block) to handle calls to extend the heap when it becomes full. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-65 1 The ARM C and C++ Libraries 1.11 Stack and heap memory allocation and the ARM C and C++ libraries Stack and heap collision detection By default, if memory allocated for the heap is destined to overlap with memory that lies in close proximity with the stack, the potential collision of heap and stack is automatically detected and the requested heap allocation fails. If you do not require this automatic collision detection, you can save a small amount of code size by disabling it with #pragma import __use_two_region_memory. Note The memory allocation functions (malloc(), realloc(), calloc(), posix_memalign()) attempt to detect allocations that collide with the current stack pointer. Such detection cannot be guaranteed to always be successful. Although it is possible to automatically detect expansion of the heap into the stack, it is not possible to automatically detect expansion of the stack into heap memory. For legacy purposes, it is possible for you to bypass all of these methods and behavior. You can do this by defining the following functions to perform your own stack and heap memory management: • __rt_stackheap_init() • __rt_heap_extend() Extending heap size at runtime To enable the heap to extend into areas of memory other than the region of memory that is specified when the program starts, you can redefine the function __user_heap_extend(). __user_heap_extend() returns blocks of memory for heap usage in extending the size of the heap. Related concepts 1.11.4 Legacy support for __user_initial_stackheap() on page 1-66. Related references 4.52 __user_heap_extend() on page 4-193. 4.53 __user_heap_extent() on page 4-194. 4.61 Legacy function __user_initial_stackheap() on page 4-204. 4.27 __rt_heap_extend() on page 4-167. 4.31 __rt_stackheap_init() on page 4-171. 4.54 __user_setup_stackheap() on page 4-195. 4.55 __vectab_stack_and_reset on page 4-196. 4.54 __user_setup_stackheap() on page 4-195. Related information Specifying stack and heap using the scatter file. 1.11.4 Legacy support for __user_initial_stackheap() Defined in rt_misc.h, __user_initial_stackheap() is supported for backwards compatibility with earlier versions of the ARM C and C++ libraries. Note ARM recommends that you use __user_setup_stackheap() in preference to __user_initial_stackheap(). The differences between __user_initial_stackheap() and __user_setup_stackheap() are: • ARM DUI0378G_02 __user_initial_stackheap() receives the stack pointer (containing the same value it had on entry to __main()) in r1, and is expected to return the new stack base in r1. Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-66 1 The ARM C and C++ Libraries 1.11 Stack and heap memory allocation and the ARM C and C++ libraries • __user_setup_stackheap() receives the stack pointer in sp, and returns the stack base in sp. __user_initial_stackheap() is provided with a small temporary stack to run on. This temporary stack enables __user_initial_stackheap() to be implemented in C, providing that it uses no more than 88 bytes of stack space. __user_setup_stackheap() has no temporary stack and cannot usually be implemented in C. Using __user_setup_stackheap() instead of __user_initial_stackheap() reduces code size, because __user_setup_stackheap() has no requirement for a temporary stack. In the following circumstances you cannot use the provided __user_setup_stackheap() function, but you can use the __user_initial_stackheap() function: • Your implementation is sufficiently complex that it warrants the use of a temporary stack when setting up the initial heap and stack. • You have a requirement to implement the heap and stack creation code in C rather than in assembly language. Related concepts 1.11.3 Stack pointer initialization and heap bounds on page 1-64. Related references 4.61 Legacy function __user_initial_stackheap() on page 4-204. 4.54 __user_setup_stackheap() on page 4-195. 1.11.5 Avoiding the heap and heap-using library functions supplied by ARM If you are developing embedded systems that have limited RAM or that provide their own heap management (for example, an operating system), you might require a system that does not define a heap area. To avoid using the heap you can either: • • Re-implement the functions in your own application. Write the application so that it does not call any heap-using function. You can reference the __use_no_heap or __use_no_heap_region symbols in your code to guarantee that no heap-using functions are linked in from the ARM library. You are only required to import these symbols once in your application, for example, using either: • IMPORT __use_no_heap from assembly language. • #pragma import(__use_no_heap) from C. If you include a heap-using function and also reference __use_no_heap or __use_no_heap_region, the linker reports an error. For example, the following sample code results in the linker error shown: #include #include #pragma import(__use_no_heap) void main() { char *p = malloc(256); ... } Error: L6915E: Library reports error: __use_no_heap was requested, but malloc was referenced To find out which objects are using the heap, link with --verbose --list=out.txt, search the output for the relevant symbol (in this case malloc), and find out what object referenced it. __use_no_heap guards against the use of malloc(), realloc(), free(), and any function that uses those functions. For example, calloc() and other stdio functions. __use_no_heap_region has the same properties as __use_no_heap, but in addition, guards against other things that use the heap memory region. For example, if you declare main() as a function taking arguments, the heap region is used for collecting argc and argv. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-67 1 The ARM C and C++ Libraries 1.11 Stack and heap memory allocation and the ARM C and C++ libraries Related references 1.6.7 Indirect semihosting C library function dependencies on page 1-37. 1.11.2 Choosing a heap implementation for memory allocation functions on page 1-63. Related information --list=filename linker option. --verbose linker option. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-68 1 The ARM C and C++ Libraries 1.12 Tailoring input/output functions in the C and C++ libraries 1.12 Tailoring input/output functions in the C and C++ libraries The input/output library functions, such as the high-level fscanf() and fprintf(), and the low-level fputc() and ferror(), and the C++ object std::cout, are not target-dependent. However, the highlevel library functions perform input/output by calling the low-level ones. These low-level functions call system I/O functions that are target-dependent. To retarget input/output, you can: • Avoid the high-level library functions. • Redefine the low-level library functions. • Redefine the system I/O functions. Whether redefining the low-level library functions or redefining the system I/O functions is a better solution depends on your use. For example, UARTs write a single character at a time and the default fputc() uses buffering, so redefining this function without a buffer might suit a UART. However, where buffer operations are possible, redefining the system I/O functions would probably be more appropriate. Related concepts 1.14 The C library printf family of functions on page 1-72. 1.15 The C library scanf family of functions on page 1-73. 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library on page 1-74. 1.17 The C library functions fread(), fgets() and gets() on page 1-76. 1.18 Re-implementing __backspace() in the C library on page 1-77. 1.19 Re-implementing __backspacewc() in the C library on page 1-78. 1.20 Redefining target-dependent system I/O functions in the C library on page 1-79. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. 1.6.7 Indirect semihosting C library function dependencies on page 1-37. 1.7.1 Building an application without the C library on page 1-40. 1.13 Target dependencies on low-level functions in the C and C++ libraries on page 1-70. 4.38 _sys_close() on page 4-179. 4.39 _sys_command_string() on page 4-180. 4.40 _sys_ensure() on page 4-181. 4.42 _sys_flen() on page 4-183. 4.43 _sys_istty() on page 4-184. 4.44 _sys_open() on page 4-185. 4.45 _sys_read() on page 4-186. 4.46 _sys_seek() on page 4-187. 4.47 _sys_tmpnam() on page 4-188. 4.48 _sys_write() on page 4-189. 4.8 _fisatty() on page 4-147. 4.43 _sys_istty() on page 4-184. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-69 1 The ARM C and C++ Libraries 1.13 Target dependencies on low-level functions in the C and C++ libraries 1.13 Target dependencies on low-level functions in the C and C++ libraries Higher-level C and C++ library input/output functions are built upon lower-level functions. If you define your own versions of the lower-level functions, you can use the library versions of the higher-level functions directly. The following table shows the dependencies of the higher-level functions on lower-level functions. fgetc() uses __FILE, but fputc() uses __FILE and ferror(). Note You must provide definitions of __stdin and __stdout if you use any of their associated high-level functions. This applies even if your re-implementations of other functions, such as fgetc() and fputc(), do not reference any data stored in __stdin and __stdout. Table key: 1. __FILE, the file structure. 2. __stdin, the standard input object of type __FILE. 3. __stdout, the standard output object of type __FILE. 4. fputc(), outputs a character to a file. 5. ferror(), returns the error status accumulated during file I/O. 6. fgetc(), gets a character from a file. 7. fgetwc() 8. fputwc() 9. __backspace(), moves the file pointer to the previous character. 10. __backspacewc(). Table 1-9 Input/output dependencies High-level function Low-level object 1 2 3 4 5 6 7 8 9 10 ARM DUI0378G_02 fgets x - - - x x - - - - fgetws x - - - - - x - - - fprintf x - - x x - - - - - fputs x - - x - - - - - - fputws x - - - - - - x - - fread x - - - - x - - - - fscanf x - - - - x - - x - fwprintf x - - - x - - x - - fwrite x - - x - - - - - - fwscanf x - - - - - x - - x getchar x x - - - x - - - - gets x x - - x x - - - - getwchar x x - - - - x - - - perror x - x x - - - - - - printf x - x x x - - - - - putchar x - x x - - - - - - Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-70 1 The ARM C and C++ Libraries 1.13 Target dependencies on low-level functions in the C and C++ libraries Table 1-9 Input/output dependencies (continued) High-level function Low-level object 1 2 3 4 5 6 7 8 9 10 puts x - x x - - - - - - putwchar x - x - - - - x - - scanf x x - - - x - - x - vfprintf x - - x x - - - - - vfscanf x - - - - x - - x - vfwprintf x - - - x - - x - - vfwscanf x - - - - - x - - x vprintf x - x x x - - - - - vscanf x x - - - x - - x - vwprintf x - x - x - - x - - vwscanf x x - - - - x - - x wprintf x - x - x - - x - - wscanf x x - - - - x - - x Note If you choose to re-implement fgetc(), fputc(), and __backspace(), be aware that fopen() and related functions use the ARM layout for the __FILE structure. You might also have to re-implement fopen() and related functions if you define your own version of __FILE. Related concepts 1.14 The C library printf family of functions on page 1-72. 1.15 The C library scanf family of functions on page 1-73. 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library on page 1-74. 1.17 The C library functions fread(), fgets() and gets() on page 1-76. 1.18 Re-implementing __backspace() in the C library on page 1-77. 1.19 Re-implementing __backspacewc() in the C library on page 1-78. 1.20 Redefining target-dependent system I/O functions in the C library on page 1-79. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. Related information ISO C Reference. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-71 1 The ARM C and C++ Libraries 1.14 The C library printf family of functions 1.14 The C library printf family of functions The printf family consists of _printf(), printf(), _fprintf(), fprintf(), vprintf(), and vfprintf(). All these functions use __FILE opaquely and depend only on the functions fputc() and ferror(). The functions _printf() and _fprintf() are identical to printf() and fprintf() except that they cannot format floating-point values. The standard output functions of the form _printf(...) are equivalent to: fprintf(& __stdout, ...) where __stdout has type __FILE. Related concepts 1.15 The C library scanf family of functions on page 1-73. 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library on page 1-74. 1.17 The C library functions fread(), fgets() and gets() on page 1-76. 1.18 Re-implementing __backspace() in the C library on page 1-77. 1.19 Re-implementing __backspacewc() in the C library on page 1-78. 1.20 Redefining target-dependent system I/O functions in the C library on page 1-79. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.13 Target dependencies on low-level functions in the C and C++ libraries on page 1-70. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-72 1 The ARM C and C++ Libraries 1.15 The C library scanf family of functions 1.15 The C library scanf family of functions The scanf() family consists of scanf() and fscanf(). These functions depend only on the functions fgetc(), __FILE, and __backspace(). The standard input function of the form scanf(...) is equivalent to: fscanf(& __stdin, ...) where __stdin is of type __FILE. Related concepts 1.14 The C library printf family of functions on page 1-72. 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library on page 1-74. 1.17 The C library functions fread(), fgets() and gets() on page 1-76. 1.18 Re-implementing __backspace() in the C library on page 1-77. 1.19 Re-implementing __backspacewc() in the C library on page 1-78. 1.20 Redefining target-dependent system I/O functions in the C library on page 1-79. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.13 Target dependencies on low-level functions in the C and C++ libraries on page 1-70. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-73 1 The ARM C and C++ Libraries 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library If you define your own version of __FILE, your own fputc() and ferror() functions, and the __stdout object, you can use all of the printf() family, fwrite(), fputs(), puts() and the C++ object std::cout unchanged from the library. These examples show you how to do this. However, consider modifying the system I/O functions instead of these low-level library functions if you require real file handling. You are not required to re-implement every function shown in these examples. Only re-implement the functions that are used in your application. Retargeting printf() #include struct __FILE { int handle; /* Whatever you require here. If the only file you are using is */ /* standard output using printf() for debugging, no file handling */ /* is required. */ }; /* FILE is typedef’d in stdio.h. */ FILE __stdout; int fputc(int ch, FILE *f) { /* Your implementation of fputc(). */ return ch; } int ferror(FILE *f) { /* Your implementation of ferror(). */ return 0; } void test(void) { printf("Hello world\n"); } Note Be aware of endianness with fputc(). fputc() takes an int parameter, but contains only a character. Whether the character is in the first or the last byte of the integer variable depends on the endianness. The following code sample avoids problems with endianness: extern void sendchar(char *ch); int fputc(int ch, FILE *f) { /* example: write a character to an LCD */ char tempch = ch; // temp char avoids endianness issue sendchar(&tempch); // sendchar(&ch) would not work everywhere return ch; } Retargeting cout File 1: Re-implement any functions that require re-implementation. #include namespace std { struct __FILE { int handle; /* Whatever you require here. If the only file you are using is */ /* standard output using printf() for debugging, no file handling */ /* is required. */ }; FILE __stdout; FILE __stdin; FILE __stderr; int fgetc(FILE *f) ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-74 1 The ARM C and C++ Libraries 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library { } /* Your implementation of fgetc(). */ return 0; }; int fputc(int c, FILE *stream) { /* Your implementation of fputc(). */ } int ferror(FILE *stream) { /* Your implementation of ferror(). */ } long int ftell(FILE *stream) { /* Your implementation of ftell(). */ } int fclose(FILE *f) { /* Your implementation of fclose(). */ return 0; } int fseek(FILE *f, long nPos, int nMode) { /* Your implementation of fseek(). */ return 0; } int fflush(FILE *f) { /* Your implementation of fflush(). */ return 0; } File 2: Print "Hello world" using your re-implemented functions. #include #include using namespace std; int main() { cout << "Hello world\n"; return 0; } By default, fread() and fwrite() call fast block input/output functions that are part of the ARM stream implementation. If you define your own __FILE structure instead of using the ARM stream implementation, fread() and fwrite() call fgetc() instead of calling the block input/output functions. Related concepts 1.14 The C library printf family of functions on page 1-72. 1.15 The C library scanf family of functions on page 1-73. 1.17 The C library functions fread(), fgets() and gets() on page 1-76. 1.18 Re-implementing __backspace() in the C library on page 1-77. 1.19 Re-implementing __backspacewc() in the C library on page 1-78. 1.20 Redefining target-dependent system I/O functions in the C library on page 1-79. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.13 Target dependencies on low-level functions in the C and C++ libraries on page 1-70. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-75 1 The ARM C and C++ Libraries 1.17 The C library functions fread(), fgets() and gets() 1.17 The C library functions fread(), fgets() and gets() The functions fread(), fgets(), and gets() are implemented as fast block input/output functions where possible. These fast implementations are part of the ARM stream implementation and they bypass fgetc(). Where the fast implementation is not possible, they are implemented as a loop over fgetc() and ferror(). Each uses the FILE argument opaquely. If you provide your own implementation of __FILE, __stdin (for gets()), fgetc(), and ferror(), you can use these functions, and the C++ object std::cin directly from the library. Related concepts 1.14 The C library printf family of functions on page 1-72. 1.15 The C library scanf family of functions on page 1-73. 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library on page 1-74. 1.18 Re-implementing __backspace() in the C library on page 1-77. 1.19 Re-implementing __backspacewc() in the C library on page 1-78. 1.20 Redefining target-dependent system I/O functions in the C library on page 1-79. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.13 Target dependencies on low-level functions in the C and C++ libraries on page 1-70. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-76 1 The ARM C and C++ Libraries 1.18 Re-implementing __backspace() in the C library 1.18 Re-implementing __backspace() in the C library The function __backspace() is used by the scanf family of functions, and must be re-implemented if you retarget the stdio arrangements at the fgetc() level. Note Normally, you are not required to call __backspace() directly, unless you are implementing your own scanf-like function. The syntax is: int __backspace(FILE *stream); __backspace(stream) must only be called after reading a character from the stream. You must not call it after a write, a seek, or immediately after opening the file, for example. It returns to the stream the last character that was read from the stream, so that the same character can be read from the stream again by the next read operation. This means that a character that was read from the stream by scanf but that is not required (that is, it terminates the scanf operation) is read correctly by the next function that reads from the stream. __backspace is separate from ungetc(). This is to guarantee that a single character can be pushed back after the scanf family of functions has finished. The value returned by __backspace() is either 0 (success) or EOF (failure). It returns EOF only if used incorrectly, for example, if no characters have been read from the stream. When used correctly, __backspace() must always return 0, because the scanf family of functions do not check the error return. The interaction between __backspace() and ungetc() is: • If you apply __backspace() to a stream and then ungetc() a character into the same stream, subsequent calls to fgetc() must return first the character returned by ungetc(), and then the character returned by __backspace(). • If you ungetc() a character back to a stream, then read it with fgetc(), and then backspace it, the next character read by fgetc() must be the same character that was returned to the stream. That is the __backspace() operation must cancel the effect of the fgetc() operation. However, another call to ungetc() after the call to __backspace() is not required to succeed. • The situation where you ungetc() a character into a stream and then __backspace() another one immediately, with no intervening read, never arises. __backspace() must only be called after fgetc(), so this sequence of calls is illegal. If you are writing __backspace() implementations, you can assume that the ungetc() of a character into a stream followed immediately by a __backspace() with no intervening read, never occurs. Related concepts 1.14 The C library printf family of functions on page 1-72. 1.15 The C library scanf family of functions on page 1-73. 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library on page 1-74. 1.17 The C library functions fread(), fgets() and gets() on page 1-76. 1.19 Re-implementing __backspacewc() in the C library on page 1-78. 1.20 Redefining target-dependent system I/O functions in the C library on page 1-79. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.13 Target dependencies on low-level functions in the C and C++ libraries on page 1-70. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-77 1 The ARM C and C++ Libraries 1.19 Re-implementing __backspacewc() in the C library 1.19 Re-implementing __backspacewc() in the C library __backspacewc() is the wide-character equivalent of __backspace(). __backspacewc() behaves in the same way as __backspace() except that it pushes back the last wide character instead of a narrow character. Related concepts 1.14 The C library printf family of functions on page 1-72. 1.15 The C library scanf family of functions on page 1-73. 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library on page 1-74. 1.17 The C library functions fread(), fgets() and gets() on page 1-76. 1.18 Re-implementing __backspace() in the C library on page 1-77. 1.20 Redefining target-dependent system I/O functions in the C library on page 1-79. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.13 Target dependencies on low-level functions in the C and C++ libraries on page 1-70. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-78 1 The ARM C and C++ Libraries 1.20 Redefining target-dependent system I/O functions in the C library 1.20 Redefining target-dependent system I/O functions in the C library The default target-dependent I/O functions use semihosting. If any of these functions are redefined, then they must all be redefined. The function prototypes are contained in rt_sys.h. These functions are called by the C standard I/O library functions. For example, _sys_open() is called by fopen() and freopen(). _sys_open() uses the strings __stdin_name, __stdout_name, and __stderr_name during C library initialization to identify which standard I/O device handle to return. You can leave their values as the default (:tt) if _sys_open() does not use them. The following example shows you how to redefine the required functions for a device that supports writing but not reading. Example of retargeting the system I/O functions /* * These names are used during library initialization as the * file names opened for stdin, stdout, and stderr. * As we define _sys_open() to always return the same file handle, * these can be left as their default values. */ const char __stdin_name[] = ":tt"; const char __stdout_name[] = ":tt"; const char __stderr_name[] = ":tt"; FILEHANDLE _sys_open(const char *name, int openmode) { return 1; /* everything goes to the same output */ } int _sys_close(FILEHANDLE fh) { return 0; } int _sys_write(FILEHANDLE fh, const unsigned char *buf, unsigned len, int mode) { your_device_write(buf, len); return 0; } int _sys_read(FILEHANDLE fh, unsigned char *buf, unsigned len, int mode) { return -1; /* not supported */ } void _ttywrch(int ch) { char c = ch; your_device_write(&c, 1); } int _sys_istty(FILEHANDLE fh) { return 0; /* buffered output */ } int _sys_seek(FILEHANDLE fh, long pos) { return -1; /* not supported */ } long _sys_flen(FILEHANDLE fh) { return -1; /* not supported */ } rt_sys.h defines the type FILEHANDLE. The value of FILEHANDLE is returned by _sys_open() and identifies an open file on the host system. If the system I/O functions are redefined, both normal character I/O and wide character I/O work. That is, you are not required to do anything extra with these functions for wide character I/O to work. Related concepts 1.14 The C library printf family of functions on page 1-72. 1.15 The C library scanf family of functions on page 1-73. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-79 1 The ARM C and C++ Libraries 1.20 Redefining target-dependent system I/O functions in the C library 1.16 Redefining low-level library functions to enable direct use of high-level library functions in the C library on page 1-74. 1.17 The C library functions fread(), fgets() and gets() on page 1-76. 1.18 Re-implementing __backspace() in the C library on page 1-77. 1.19 Re-implementing __backspacewc() in the C library on page 1-78. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. 1.13 Target dependencies on low-level functions in the C and C++ libraries on page 1-70. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-80 1 The ARM C and C++ Libraries 1.21 Tailoring non-input/output C library functions 1.21 Tailoring non-input/output C library functions In addition to tailoring input/output C library functions, many C library functions that are not input/ output functions can also be tailored. Implementation of these ISO standard functions depends entirely on the target operating system. The default implementation of these functions is semihosted. That is, each function uses semihosting. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. 1.6.7 Indirect semihosting C library function dependencies on page 1-37. 4.3 clock() on page 4-142. 4.4 _clock_init() on page 4-143. 4.50 time() on page 4-191. 4.21 remove() on page 4-161. 4.22 rename() on page 4-162. 4.49 system() on page 4-190. 4.10 getenv() on page 4-149. 4.11 _getenv_init() on page 4-150. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-81 1 The ARM C and C++ Libraries 1.22 Real-time integer division in the ARM libraries 1.22 Real-time integer division in the ARM libraries The ARM library provides a real-time division routine and a standard division routine. The standard division routine supplied with the ARM libraries provides good overall performance. However, the amount of time required to perform a division depends on the input values. For example, a division that generates a four-bit quotient might require only 12 cycles while a 32-bit quotient might require 96 cycles. Depending on your target, some applications require a faster worst-case cycle count at the expense of lower average performance. For this reason, the ARM library provides two divide routines. The real-time routine: • • • • • Always executes in fewer than 45 cycles. Is faster than the standard division routine for larger quotients. Is slower than the standard division routine for typical quotients. Returns the same results. Does not require any change in the surrounding code. Note Real-time division is not available in the libraries for Cortex-M1 or Cortex-M0. Note The Cortex-R4 and Cortex-M3 processors support hardware floating-point divide, so they do not require the library divide routines. Select the real-time divide routine using either of the following methods: • IMPORT __use_realtime_division from assembly language. • #pragma import(__use_realtime_division) from C. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-82 1 The ARM C and C++ Libraries 1.23 ISO C library implementation definition 1.23 ISO C library implementation definition Describes how the libraries fulfill the requirements of the ISO specification. This section contains the following subsections: • 1.23.1 How the ARM C library fulfills ISO C specification requirements on page 1-83. • 1.23.2 mathlib error handling on page 1-84. • 1.23.3 ISO-compliant implementation of signals supported by the signal() function in the C library and additional type arguments on page 1-84. • 1.23.4 ISO-compliant C library input/output characteristics on page 1-85. • 1.23.5 Standard C++ library implementation definition on page 1-87. 1.23.1 How the ARM C library fulfills ISO C specification requirements The ISO specification leaves some features to implementors, but requires that implementation choices be documented. The implementation of the generic ARM C library in this respect is as follows: • The macro NULL expands to the integer constant 0. • If a program redefines a reserved external identifier, an error might occur when the program is linked with the standard libraries. If it is not linked with standard libraries, no error is diagnosed. • The __aeabi_assert() function prints information on the failing diagnostic on stderr and then calls the abort() function: *** assertion failed: expression, file name, line number Note The behavior of the assert macro depends on the conditions in operation at the most recent occurrence of #include . See 1.8.6 Program exit and the assert macro on page 1-50 for more information about the behavior of the assert macro. • • • The following functions test for character values in the range EOF (-1) to 255 inclusive: — isalnum() — isalpha() — iscntrl() — islower() — isprint() — isupper() — ispunct() The fully POSIX-compliant functions remquo(), remquof() and remquol() return the remainder of the division of x by y and store the quotient of the division in the pointer *quo. An implementationdefined integer value defines the number of bits of the quotient that are stored. In the ARM C library, this value is set to 4. C99 behavior, with respect to mathlib error handling, is enabled by default. Related concepts 1.23.4 ISO-compliant C library input/output characteristics on page 1-85. 1.8.6 Program exit and the assert macro on page 1-50. Related references 1.23.2 mathlib error handling on page 1-84. 1.23.3 ISO-compliant implementation of signals supported by the signal() function in the C library and additional type arguments on page 1-84. 1.23.5 Standard C++ library implementation definition on page 1-87. 1.26 C and C++ library naming conventions on page 1-91. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-83 1 The ARM C and C++ Libraries 1.23 ISO C library implementation definition 1.23.2 mathlib error handling The error handling of mathematical functions is consistent with Annex F of the ISO/IEC C99 standard. Related concepts 1.23.4 ISO-compliant C library input/output characteristics on page 1-85. 1.23.1 How the ARM C library fulfills ISO C specification requirements on page 1-83. Related references 1.23.3 ISO-compliant implementation of signals supported by the signal() function in the C library and additional type arguments on page 1-84. 1.23.5 Standard C++ library implementation definition on page 1-87. 1.23.3 ISO-compliant implementation of signals supported by the signal() function in the C library and additional type arguments The signal() function supports a number of signals. The following table shows the signals supported by the signal() function. It also shows which signals use an additional argument to give more information about the circumstance in which the signal was raised. The additional argument is given in the type parameter of __raise(). For example, division by floating-point zero results in a SIGFPE signal with a corresponding additional argument of FE_EX_DIVBYZERO. Table 1-10 Signals supported by the signal() function Signal Number Description SIGABRT 1 Returned when the abort() function is called. Additional argument None The abort() function is triggered when there is an untrapped C++ exception, or when an assertion fails. SIGFPE 2 Signals any arithmetic exception, for example, division by zero. Used by hard and soft floating-point and by integer division. A set of bits from FE_EX_INEXACT, FE_EX_UNDERFLOW, FE_EX_OVERFLOW, FE_EX_DIVBYZERO, FE_EX_INVALID, DIVBYZERO a SIGILL 3 Illegal instruction. None SIGINT b 4 Attention request from user. None SIGSEGV b 5 Bad memory access. None SIGTERM b 6 Termination request. None SIGSTAK 7 Obsolete. None SIGRTRED 8 Redirection failed on a runtime library input/output stream. Name of file or device being re-opened to redirect a standard stream SIGRTMEM 9 Out of heap space during initialization or after corruption. Size of failed request SIGUSR1 10 User-defined. User-defined SIGUSR2 11 User-defined. User-defined SIGPVFN 12 A pure virtual function was called from C++. - SIGCPPL 13 Not normally used. - a b These constants are defined in fenv.h. FE_EX_DIVBYZERO is for floating-point division while DIVBYZERO is for integer division. The library never generates this signal. It is available for you to raise manually, if required. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-84 1 The ARM C and C++ Libraries 1.23 ISO C library implementation definition Table 1-10 Signals supported by the signal() function (continued) Signal Number Description Additional argument reserved 15-31 Reserved. Reserved other > 31 User-defined. User-defined Although SIGSTAK exists in signal.h, this signal is not generated by the C library and is considered obsolete. A signal number greater than SIGUSR2 can be passed through __raise() and caught by the default signal handler, but it cannot be caught by a handler registered using signal(). signal() returns an error code if you try to register a handler for a signal number greater than SIGUSR2. The default handling of all recognized signals is to print a diagnostic message and call exit(). This default behavior applies at program startup and until you change it. Caution The IEEE 754 standard for floating-point processing states that the default action to an exception is to proceed without a trap. A raised exception in floating-point calculations does not, by default, generate SIGFPE. You can modify floating-point error handling by tailoring the functions and definitions in fenv.h. However, you must compile these functions with a non-default FP model, such as -fpmode=ieee_fixed and upwards. For all the signals in the above table, when a signal occurs, if the handler points to a function, the equivalent of signal(sig, SIG_DFL) is executed before the call to the handler. If the SIGILL signal is received by a handler specified to by the signal() function, the default handling is reset. Related concepts 1.23.4 ISO-compliant C library input/output characteristics on page 1-85. 1.23.1 How the ARM C library fulfills ISO C specification requirements on page 1-83. 3.6.8 Exception types recognized by the ARM floating-point environment on page 3-134. Related references 1.23.2 mathlib error handling on page 1-84. 1.23.5 Standard C++ library implementation definition on page 1-87. 1.10 Modification of C library functions for error signaling, error handling, and program exit on page 1-61. 4.19 __raise() on page 4-159. 4.30 __rt_raise() on page 4-170. Related information --fpmode=model compiler option. --exceptions, --no_exceptions compiler option. IEEE Standard for Floating-Point Arithmetic (IEEE 754), 1985 version. 1.23.4 ISO-compliant C library input/output characteristics The generic ARM C library has defined input/output characteristics. These input/output characteristics are as follows: ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-85 1 The ARM C and C++ Libraries 1.23 ISO C library implementation definition • • • • • • • • • • • • • • • • • • • The last line of a text stream does not require a terminating newline character. Space characters written out to a text stream immediately before a newline character do appear when read back in. No NUL characters are appended to a binary output stream. The file position indicator of an append mode stream is initially placed at the end of the file. A write to a text stream causes the associated file to be truncated beyond the point where the write occurred if this is the behavior of the device category of the file. If semihosting is used, the maximum number of open files is limited by the available target memory. A zero-length file exists, that is, where no characters have been written by an output stream. A file can be opened many times for reading, but only once for writing or updating. A file cannot simultaneously be open for reading on one stream, and open for writing or updating on another. localtime() is implemented and returns the local time. gmtime() is not implemented and returns NULL. Therefore converting between time-zones is not supported. The status returned by exit() is the same value that was passed to it. For definitions of EXIT_SUCCESS and EXIT_FAILURE, see the header file stdlib.h. Semihosting, however, does not pass the status back to the execution environment. The error messages returned by the strerror() function are identical to those given by the perror() function. If the size of area requested is zero, calloc() and realloc() return NULL. If the size of area requested is zero, malloc() returns a pointer to a zero-size block. abort() closes all open files and deletes all temporary files. fprintf() prints %p arguments in lowercase hexadecimal format as if a precision of 8 had been specified. If the variant form (%#p) is used, the number is preceded by the character @. fscanf() treats %p arguments exactly the same as %x arguments. fscanf() always treats the character "-" in a %...[...] argument as a literal character. ftell(), fsetpos() and fgetpos() set errno to the value of EDOM on failure. perror() generates the messages shown in the following table. Table 1-11 perror() messages Error Message 0 No error (errno = 0) EDOM EDOM - function argument out of range ERANGE ERANGE - function result not representable ESIGNUM ESIGNUM - illegal signal number Others Unknown error The following characteristics are unspecified in the ARM C library. They must be specified in an ISOcompliant implementation: • The validity of a filename. • Whether remove() can remove an open file. • The effect of calling the rename() function when the new name already exists. • The effect of calling getenv() (the default is to return NULL, no value available). • The effect of calling system(). • The value returned by clock(). Related concepts 1.23.1 How the ARM C library fulfills ISO C specification requirements on page 1-83. Related references 1.23.2 mathlib error handling on page 1-84. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-86 1 The ARM C and C++ Libraries 1.23 ISO C library implementation definition 1.23.3 ISO-compliant implementation of signals supported by the signal() function in the C library and additional type arguments on page 1-84. 1.23.5 Standard C++ library implementation definition on page 1-87. 1.23.5 Standard C++ library implementation definition The ARM C++ library provides all of the library defined in the ISO/IEC 14882 :1998(E) C++ Standard, aside from some limitations. For information on implementation-defined behavior that is defined in the Rogue Wave C++ library, see the Rogue Wave HTML documentation. The Standard C++ library is distributed in binary form only. The following table describes the most important features missing from the current release. Table 1-12 Standard C++ library differences Standard Implementation differences locale The locale message facet is not supported. It fails to open catalogs at runtime because the ARM C library does not support catopen() and catclose() through nl_types.h. One of two locale definitions can be selected at link time. Other locales can be created by user-redefinable functions. Timezone Not supported by the ARM C library. Thread safety The following points summarize thread safety in the Rogue Wave C++ library: • The function std::set_new_handler() is not thread-safe. This means that some forms of ::operator new and ::operator delete are not thread-safe with respect to std::set_new_handler(): — The default C++ runtime library implementations of the following use malloc() and free() and are thread-safe with respect to each other. They are not thread-safe with respect to std::set_new_handler(). You are permitted to replace them: ::operator new(std::size_t) ::operator new[](std::size_t) ::operator new(std::size_t, const std::nothrow_t&) ::operator new[](std::size_t, const std::nothrow_t) ::operator delete(void*) ::operator delete[](void*) ::operator delete(void*, const std::nothrow_t&) ::operator delete[](void*, const std::nothrow_t&) — The following placement forms are also thread-safe. You are not permitted to replace them: ::operator new(std::size_t, void*) ::operator new[](std::size_t, void*) ::operator delete(void*, void*) ::operator delete[](void*, void*) • • ARM DUI0378G_02 Construction and destruction of global objects are not thread-safe. Construction of local static objects can be made thread-safe if you re-implement the functions __cxa_guard_acquire(), __cxa_guard_release(), __cxa_guard_abort(), __cxa_atexit() and Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-87 1 The ARM C and C++ Libraries 1.23 ISO C library implementation definition __aeabi_atexit() appropriately. For example, with appropriate re-implementation, the following construction of lsobj can be made thread-safe: struct T { T(); }; void f() { static T lsobj; } • • Throwing an exception is thread-safe if any user constructors and destructors that get called are also thread-safe. The ARM C++ library uses the ARM C library. To use the ARM C++ library in a multithreaded environment, you must provide the same functions that you would be required to provide when using the ARM C library in a multithreaded environment. Related information Rogue Wave Standard C++ Library Documentation. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-88 1 The ARM C and C++ Libraries 1.24 C library functions and extensions 1.24 C library functions and extensions The ARM C library is fully compliant with the ISO C99 library standard and provides a number of GNU, POSIX, BSD-derived, and ARM Compiler-specific extensions. The following table describes these extensions. Table 1-13 C library extensions Function Header file definition Extension wcscasecmp() wchar.h GNU extension with ARM library support wcsncasecmp() wchar.h GNU extension with ARM library support wcstombs() stdlib.h POSIX extended functionality posix_memalign() stdlib.h POSIX extended functionality alloca() alloca.h Common nonstandard extension to many C libraries strlcpy() string.h Common BSD-derived extension to many C libraries strlcat() string.h Common BSD-derived extension to many C libraries strcasecmp() string.h Standardized by POSIX strncasecmp() string.h Standardized by POSIX _fisatty() stdio.h Specific to ARM Compiler __heapstats() stdlib.h Specific to ARM Compiler __heapvalid() stdlib.h Specific to ARM Compiler Related references 4.56 wcscasecmp() on page 4-197. 4.57 wcsncasecmp() on page 4-198. 4.58 wcstombs() on page 4-199. 4.2 alloca() on page 4-141. 4.36 strlcat() on page 4-177. 4.37 strlcpy() on page 4-178. 4.34 strcasecmp() on page 4-175. 4.35 strncasecmp() on page 4-176. 4.8 _fisatty() on page 4-147. 4.12 __heapstats() on page 4-151. 4.13 __heapvalid() on page 4-152. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-89 1 The ARM C and C++ Libraries 1.25 Compiler generated and library-resident helper functions 1.25 Compiler generated and library-resident helper functions Compiler support or helper functions specific to the compilation tools are typically used when the compiler cannot easily produce a suitable code sequence itself. In RVCT 5.06 and later, the ARM Compiler options --common_functions and --no_common_functions control whether the compiler generates and embeds helper functions in the resulting object files, or whether the helper functions reside in libraries. In RVCT v4.0 and later, the compiler generates and embeds helper functions in the resulting object files. In RVCT v3.1 and earlier, the helper functions reside in libraries. Because these libraries are specific to the ARM Compiler, they are intended to be redistributed as necessary with your own code. For example, if you are distributing a library to a third party, they might also require the appropriate helper library to link their final application successfully. Be aware of redistribution rights of the libraries, as specified in your End User License Agreement. Related references 1.26 C and C++ library naming conventions on page 1-91. Related information --common_functions, --no_common_functions compiler option. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-90 1 The ARM C and C++ Libraries 1.26 C and C++ library naming conventions 1.26 C and C++ library naming conventions The library filename identifies how the variant was built. Note The library naming convention described in this documentation applies to the current release of the ARM compilation tools. Do not rely on C and C++ library names. They might change in future releases. Normally, you do not have to list any of the C and C++ libraries explicitly on the linker command line. The ARM linker automatically selects the correct C or C++ libraries to use, and it might use several, based on the accumulation of the object attributes. The values for the fields of the filename, and the relevant build options are: root/prefix_arch[fpu][entrant][enum][wchar].endian root cpplib An ARM C++ library. prefix c ISO C and C++ basic runtime support. cpp Rogue Wave C++ library. cpprt The ARM C++ runtime libraries. f --fpmode=ieee_fixed. IEEE-compliant library with a fixed rounding mode (round to nearest) and no inexact exceptions. fj --fpmode=ieee_no_fenv. IEEE-compliant library with a fixed rounding mode (round to nearest) and no exceptions. fz --fpmode=fast or --fpmode=std. Behaves like the fj library, but additionally flushes denormals and infinities to zero. This library behaves like the ARM VFP in Fast mode. This is the default. g --fpmode=ieee_full. IEEE-compliant library with configurable rounding mode and all IEEE exceptions. h Compiler support (helper function) library. m Transcendental math functions. mc Non ISO C-compliant ISO C micro-library basic runtime support. mf Non IEEE 754 floating-point compliant micro-library support. arch ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-91 1 The ARM C and C++ Libraries 1.26 C and C++ library naming conventions 4 An ARM only library for use with ARMv4. t An ARM/Thumb interworking library for use with ARMv4T. 5 An ARM/Thumb interworking library for use with ARMv5T and later. w A Thumb only library for use with ARMv6-M. p A Thumb only library for use with ARMv7-M. 2 A combined ARM and Thumb library for use with Cortex-R series processors. You can prevent this library being selected using the linker option --no_thumb2_library. fpu m A variant of the library for processors that have single-precision hardware floatingpoint only, such as Cortex-M4. v Uses VFP instruction set. s Soft VFP. Note If none of v, m, or s are present in a library name, the library provides no floating-point support. entrant e Position-independent access to static data. f FPIC addressing is enabled. Note If neither e nor f is present in a library name, the library either: • Uses position-dependent access to static data. This is the case for the main C libraries with prefixes c or mc. • Does not access static data, or does so only with the help of the main C library. This is the case for fplib and mathlib libraries with prefixes fz, fj, f, g, mf, or m. enum n Compatible with the compiler option, --enum_is_int. wchar u Indicates the size of wchar_t.When present, the library is compatible with the compiler option, --wchar32. Otherwise, it is compatible with --wchar16. endian l Little-endian. b Big-endian. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-92 1 The ARM C and C++ Libraries 1.26 C and C++ library naming conventions For example: armlib/c_4.b cpplib/cpprt_5f.l Note Not all variant/name combinations are valid. See the armlib and cpplib directories for the libraries that are supplied with the ARM Compiler. The linker command-line option --info libraries provides information on every library that is automatically selected for the link stage. Related concepts 1.25 Compiler generated and library-resident helper functions on page 1-90. Related information --enum_is_int compiler option. --wchar16 compiler option. --wchar32 compiler option. --info=topic[,topic,...] linker option. --thumb2_library linker option. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-93 1 The ARM C and C++ Libraries 1.27 Using macro__ARM_WCHAR_NO_IO to disable FILE declaration and wide I/O function prototypes 1.27 Using macro__ARM_WCHAR_NO_IO to disable FILE declaration and wide I/O function prototypes In strict C/C++ mode, the header files wchar.h and cwchar do not declare the FILE type. You can also define the macro __ARM_WCHAR_NO_IO to cause these header files not to declare FILE or the wide I/O function prototypes. Declaring the FILE type can lead to better consistency in debug information. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-94 1 The ARM C and C++ Libraries 1.28 Using library functions with execute-only memory 1.28 Using library functions with execute-only memory The ARM Compiler lets you build applications for execute-only memory. However, the ARM C and C++ libraries are not execute-only compliant. If your application calls library functions, the library objects included in the image are not execute-only compliant. You must ensure these objects are not assigned to an execute-only memory region. Note ARM does not provide libraries that are built without literal pools. The libraries still use literal pools, even when you use the various --no_*_literal_pools options. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 1-95 Chapter 2 The ARM C Micro-library Describes microlib, the C micro-library. It contains the following sections: • 2.1 About microlib on page 2-97. • 2.2 Differences between microlib and the default C library on page 2-98. • 2.3 Library heap usage requirements of microlib on page 2-100. • 2.4 ISO C features missing from microlib on page 2-101. • 2.5 Building an application with microlib on page 2-103. • 2.6 Configuring the stack and heap for use with microlib on page 2-104. • 2.7 Entering and exiting programs linked with microlib on page 2-105. • 2.8 Tailoring the microlib input/output functions on page 2-106. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-96 2 The ARM C Micro-library 2.1 About microlib 2.1 About microlib Microlib is an alternative library to the default C library. It is intended for use with deeply embedded applications that must fit into extremely small memory footprints. These applications do not run under an operating system. Note Microlib does not attempt to be an ISO C-compliant library. Microlib is highly optimized for small code size. It has less functionality than the default C library and some ISO C features are completely missing. Some library functions are also slower. Functions in microlib are responsible for: • Creating an environment that a C program can execute in. This includes: — Creating a stack. — Creating a heap, if required. — Initializing the parts of the library the program uses. • Starting execution by calling main(). Related concepts 2.2 Differences between microlib and the default C library on page 2-98. 2.3 Library heap usage requirements of microlib on page 2-100. 2.4 ISO C features missing from microlib on page 2-101. 2.5 Building an application with microlib on page 2-103. 2.7 Entering and exiting programs linked with microlib on page 2-105. Related tasks 2.6 Configuring the stack and heap for use with microlib on page 2-104. Related references 2.8 Tailoring the microlib input/output functions on page 2-106. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-97 2 The ARM C Micro-library 2.2 Differences between microlib and the default C library 2.2 Differences between microlib and the default C library There are a number of differences between microlib and the default C library. The main differences are: • Microlib is not compliant with the ISO C library standard. Some ISO features are not supported and others have less functionality. • Microlib is not compliant with the IEEE 754 standard for binary floating-point arithmetic. • Microlib is highly optimized for small code size. • Locales are not configurable. The default C locale is the only one available. • main() must not be declared to take arguments and must not return. In main, argc and argv parameters are undefined and cannot be used to access command-line arguments. • Microlib provides limited support for C99 functions. Specifically, microlib does not support the following C99 functions: — functions: feclearexcept fegetround fesetenv fetestexcept fegetenv feholdexcept fesetexceptflag feupdateenv fegetexceptflag feraiseexcept fesetround — Wide characters in general: btowc fputws getwc iswblank iswgraph iswspace mbrlen mbtowc swscanf ungetwc vswscanf wcschr wcsftime wcsncpy wcsspn wcstoimax wcstoll wcstoumax wctrans wmemcpy wscanf fgetwc fwide getwchar iswcntrl iswlower iswupper mbsinit putwc towctrans vfwprintf vwprintf wcscmp wcslen wcspbrk wcsstr wcstok wcstombs wcsxfrm wctype wmemmove fgetws fwprintf iswalnum iswctype iswprint iswxdigit mbsrtowcs putwchar towlower vfwscanf vwscanf wcscoll wcsncat wcsrchr wcstod wcstol wcstoul wctob wmemchr wmemset fputwc fwscanf iswalpha iswdigit iswpunct mblen mbstowcs swprintf towupper vswprintf wcscat wcscspn wcsncmp wcsrtombs wcstof wcstold wcstoull wctomb wmemcmp wprintf — Auxiliary functions: ilogb lgamma logb nextafter nexttoward ilogbf lgammaf logbf nextafterf nexttowardf ilogbl lgammal logbl nextafterl nexttowardl — Functions relating to program startup and shutdown and other OS interaction: _Exit system • • • • • • • • • ARM DUI0378G_02 atexit time exit Microlib does not support C++. Microlib does not support operating system functions. Microlib does not support position-independent code. Microlib does not provide mutex locks to guard against code that is not thread safe. Microlib does not support wide characters or multibyte strings. Microlib does not support selectable one or two region memory models as the standard library (stdlib) does. Microlib provides only the two region memory model with separate stack and heap regions. Microlib does not support the bit-aligned memory functions _membitcpy[b|h|w][b|l]() and membitmove[b|h|w][b|l](). Microlib can be used with either --fpmode=std or --fpmode=fast. The level of ANSI C stdio support that is provided can be controlled with #pragma import(__use_full_stdio). Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-98 2 The ARM C Micro-library 2.2 Differences between microlib and the default C library • • • #pragma import(__use_smaller_memcpy) selects a smaller, but slower, version of memcpy(). setvbuf() and setbuf() always fail because all streams are unbuffered. feof() and ferror() always return 0 because the error and EOF indicators are not supported. Related concepts 2.1 About microlib on page 2-97. 2.3 Library heap usage requirements of microlib on page 2-100. 2.4 ISO C features missing from microlib on page 2-101. 2.5 Building an application with microlib on page 2-103. 2.7 Entering and exiting programs linked with microlib on page 2-105. Related tasks 2.6 Configuring the stack and heap for use with microlib on page 2-104. Related references 2.8 Tailoring the microlib input/output functions on page 2-106. Related information --fpmode=model compiler option. #pragma import(__use_full_stdio). #pragma import(__use_smaller_memcpy). ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-99 2 The ARM C Micro-library 2.3 Library heap usage requirements of microlib 2.3 Library heap usage requirements of microlib Library heap usage requirements for microlib differ to those of standardlib. The differences are: • • The size of heap memory allocated for fopen() is 20 bytes for the FILE structure. No buffer is ever allocated. You must not declare main() to take arguments if you are using microlib. Note The size of heap memory allocated for fopen() might change in future releases. Related concepts 1.11.1 Library heap usage requirements of the ARM C and C++ libraries on page 1-62. 2.1 About microlib on page 2-97. 2.2 Differences between microlib and the default C library on page 2-98. 2.4 ISO C features missing from microlib on page 2-101. 2.5 Building an application with microlib on page 2-103. 2.7 Entering and exiting programs linked with microlib on page 2-105. Related tasks 2.6 Configuring the stack and heap for use with microlib on page 2-104. Related references 2.8 Tailoring the microlib input/output functions on page 2-106. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-100 2 The ARM C Micro-library 2.4 ISO C features missing from microlib 2.4 ISO C features missing from microlib Microlib does not support all ISO C90 features. Major ISO C90 features not supported by microlib are: Wide character and multibyte support All functions dealing with wide characters or multibyte strings are not supported by microlib. A link error is generated if these are used. For example, mbtowc(), wctomb(), mbstowcs() and wcstombs(). All functions defined in Normative Addendum 1 are not supported by microlib. Operating system interaction Almost all functions that interact with an operating system are not supported by microlib. For example, abort(), exit(), atexit(), assert(), time(), system() and getenv(). An exception is clock(). A minimal implementation of clock() has been provided, which returns only –1, not the elapsed time. You may reimplement clock() (and _clock_init(), which it needs), if required. File I/O By default, all the stdio functions that interact with a file pointer return an error if called. The only exceptions to this are the three standard streams stdin, stdout and stderr. You can change this behavior using #pragma import(__use_full_stdio). Use of this pragma provides a microlib version of stdio that supports ANSI C, with only the following exceptions: • The error and EOF indicators are not supported, so feof() and ferror() return 0. • All streams are unbuffered, so setvbuf() and setbuf() fail. Configurable locale The default C locale is the only one available. Signals The functions signal() and raise() are provided but microlib does not generate signals. The only exception to this is if the program explicitly calls raise(). Floating-point support Floating-point support diverges from IEEE 754 in the following ways, but uses the same data formats and matches IEEE 754 in operations involving only normalized numbers: • Operations involving NaNs, infinities or input denormals produce indeterminate results. Operations that produce a result that is nonzero but very small in value, return zero. • IEEE exceptions cannot be flagged by microlib, and there is no fp_status() register in microlib. • The sign of zero is not treated as significant by microlib, and zeroes that are output from microlib floating-point arithmetic have an unknown sign bit. • Only the default rounding mode is supported. Position independent and thread safe code Microlib has no reentrant variant. Microlib does not provide mutex locks to guard against code that is not thread safe. Use of microlib is not compatible with FPIC or RWPI compilation modes, and although ROPI code can be linked with microlib, the resulting binary is not ROPIcompliant overall. Command-line arguments In main, argc and argv parameters are undefined and cannot be used to access command-line arguments. Related concepts 2.1 About microlib on page 2-97. 2.2 Differences between microlib and the default C library on page 2-98. 2.3 Library heap usage requirements of microlib on page 2-100. 2.5 Building an application with microlib on page 2-103. 2.7 Entering and exiting programs linked with microlib on page 2-105. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-101 2 The ARM C Micro-library 2.4 ISO C features missing from microlib Related tasks 2.6 Configuring the stack and heap for use with microlib on page 2-104. Related references 2.8 Tailoring the microlib input/output functions on page 2-106. 1.6.8 C library API definitions for targeting a different environment on page 1-38. 1.7.1 Building an application without the C library on page 1-40. 4.3 clock() on page 4-142. 4.4 _clock_init() on page 4-143. Related information #pragma import(__use_full_stdio). ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-102 2 The ARM C Micro-library 2.5 Building an application with microlib 2.5 Building an application with microlib To build a program using microlib, you must use the command-line option --library_type=microlib. This option can be used by the compiler, assembler or linker. Use --library_type=microlib with the linker to override all other options. Compiler option armcc --library_type=microlib -c main.c armcc -c extra.c armlink -o image.axf main.o extra.o Specifying --library_type=microlib when compiling main.c results in an object file containing an attribute that asks the linker to use microlib. Compiling extra.c with --library_type=microlib is unnecessary, because the request to link against microlib exists in the object file generated by compiling main.c. Assembler option armcc -c main.c armcc -c extra.c armasm --library_type=microlib more.s armlink -o image.axf main.o extra.o more.o The request to the linker to use microlib is made as a result of assembling more.s with -library_type=microlib. Linker option armcc -c main.c armcc -c extra.c armlink --library_type=microlib -o image.axf main.o extra.o Neither object file contains the attribute requesting that the linker link against microlib, so the linker selects microlib as a result of being explicitly asked to do so on the command line. Related concepts 2.1 About microlib on page 2-97. 2.2 Differences between microlib and the default C library on page 2-98. 2.3 Library heap usage requirements of microlib on page 2-100. 2.4 ISO C features missing from microlib on page 2-101. 2.7 Entering and exiting programs linked with microlib on page 2-105. Related tasks 2.6 Configuring the stack and heap for use with microlib on page 2-104. Related references 2.8 Tailoring the microlib input/output functions on page 2-106. Related information --library_type=lib compiler option. --library_type=lib assembler option. input-file-list linker option. --library_type=lib linker option. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-103 2 The ARM C Micro-library 2.6 Configuring the stack and heap for use with microlib 2.6 Configuring the stack and heap for use with microlib To use microlib, you must specify an initial pointer for the stack. You can specify the initial pointer in a scatter file or using the __initial_sp symbol. To use the heap functions, for example, malloc(), calloc(), realloc() and free(), you must specify the location and size of the heap region. To configure the stack and heap for use with microlib, use either of the following methods: • Define the symbol __initial_sp to point to the top of the stack. If using the heap, also define symbols __heap_base and __heap_limit. __initial_sp must be aligned to a multiple of eight bytes. __heap_limit must point to the byte beyond the last byte in the heap region. • In a scatter file, either: — Define ARM_LIB_STACK and ARM_LIB_HEAP regions. If you do not intend to use the heap, only define an ARM_LIB_STACK region. — Define an ARM_LIB_STACKHEAP region. If you define an ARM_LIB_STACKHEAP region, the stack starts at the top of that region. The heap starts at the bottom. Examples To set up the initial stack and heap pointers using armasm assembly language: EXPORT __initial_sp __initial_sp EQU 0x100000 EXPORT __heap_base __heap_base EQU 0x400000 EXPORT __heap_limit __heap_limit EQU 0x800000 ; top of the stack ; start of the heap ; end of the heap To set up the initial stack and heap pointers using embedded assembler in C: __asm void dummy_function(void) { EXPORT __initial_sp __initial_sp EQU 0x100000 EXPORT __heap_base __heap_base EQU 0x400000 EXPORT __heap_limit __heap_limit EQU 0x800000 } ; top of the stack ; start of the heap ; end of the heap Related concepts 2.1 About microlib on page 2-97. 2.2 Differences between microlib and the default C library on page 2-98. 2.3 Library heap usage requirements of microlib on page 2-100. 2.4 ISO C features missing from microlib on page 2-101. 2.5 Building an application with microlib on page 2-103. 2.7 Entering and exiting programs linked with microlib on page 2-105. 1.11.3 Stack pointer initialization and heap bounds on page 1-64. Related references 2.8 Tailoring the microlib input/output functions on page 2-106. Related information About scatter-loading. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-104 2 The ARM C Micro-library 2.7 Entering and exiting programs linked with microlib 2.7 Entering and exiting programs linked with microlib Microlib requires a main() function that takes no arguments and never returns. Use main() to begin your program. Do not declare main() to take arguments. Microlib does not support command-line arguments from an operating system. Your program must not return from main(). This is because microlib does not contain any code to handle exit from main(). Microlib does not support programs that call exit(). You can ensure that your main() function does not return, by inserting an endless loop at the end of the function. For example: void main() { ... while (1); // endless loop to prevent return from main() } Related concepts 2.1 About microlib on page 2-97. 2.2 Differences between microlib and the default C library on page 2-98. 2.3 Library heap usage requirements of microlib on page 2-100. 2.4 ISO C features missing from microlib on page 2-101. 2.5 Building an application with microlib on page 2-103. Related tasks 2.6 Configuring the stack and heap for use with microlib on page 2-104. Related references 2.8 Tailoring the microlib input/output functions on page 2-106. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-105 2 The ARM C Micro-library 2.8 Tailoring the microlib input/output functions 2.8 Tailoring the microlib input/output functions Microlib provides a limited stdio subsystem. To use high-level I/O functions you must reimplement the base I/O functions. Microlib provides a limited stdio subsystem that supports unbuffered stdin, stdout and stderr only. This enables you to use printf() for displaying diagnostic messages from your application. To use high-level I/O functions you must provide your own implementation of the following base functions so that they work with your own I/O device. fputc() Implement this base function for all output functions. For example, fprintf(), printf(), fwrite(), fputs(), puts(), putc() and putchar(). fgetc() Implement this base function for all input functions. For example, fscanf(), scanf(), fread(), read(), fgets(), gets(), getc() and getchar(). __backspace() Implement this base function if your input functions use scanf() or fscanf(). Note Conversions that are not supported in microlib are %lc, %ls and %a. Related concepts 2.1 About microlib on page 2-97. 2.2 Differences between microlib and the default C library on page 2-98. 2.3 Library heap usage requirements of microlib on page 2-100. 2.4 ISO C features missing from microlib on page 2-101. 2.5 Building an application with microlib on page 2-103. 2.7 Entering and exiting programs linked with microlib on page 2-105. 1.18 Re-implementing __backspace() in the C library on page 1-77. Related tasks 2.6 Configuring the stack and heap for use with microlib on page 2-104. Related references 1.12 Tailoring input/output functions in the C and C++ libraries on page 1-69. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 2-106 Chapter 3 Floating-point Support Describes ARM support for floating-point computations. It contains the following sections: • 3.1 About floating-point support on page 3-108. • 3.2 The software floating-point library, fplib on page 3-109. • 3.3 Controlling the ARM floating-point environment on page 3-115. • 3.4 Using C99 signaling NaNs provided by mathlib (_WANT_SNAN) on page 3-127. • 3.5 mathlib double and single-precision floating-point functions on page 3-128. • 3.6 IEEE 754 arithmetic on page 3-129. • 3.7 Using the Vector Floating-Point (VFP) support libraries on page 3-137. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-107 3 Floating-point Support 3.1 About floating-point support 3.1 About floating-point support The ARM floating-point environment is an implementation of the IEEE 754-1985 standard for binary floating-point arithmetic. An ARM system might have: • A VFP coprocessor. • No floating-point hardware. If you compile for a system with a hardware VFP coprocessor, the ARM compiler makes use of it. If you compile for a system without a coprocessor, the compiler implements the computations in software. For example, the compiler option --fpu=vfp selects a hardware VFP coprocessor and the option --fpu=softvfp specifies that arithmetic operations are to be performed in software, without the use of any coprocessor instructions. Related concepts 3.2 The software floating-point library, fplib on page 3-109. 3.6 IEEE 754 arithmetic on page 3-129. Related information --fpu=name compiler option. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-108 3 Floating-point Support 3.2 The software floating-point library, fplib 3.2 The software floating-point library, fplib The software floating-point library, fplib, provides software implementations of floating-point operations. When programs are compiled to use a floating-point coprocessor, they perform basic floating-point arithmetic by means of floating-point machine instructions for the target coprocessor. When programs are compiled to use software floating-point, there is no floating-point instruction set available, so the ARM libraries provide a set of procedure calls to do floating-point arithmetic. These procedures are in the software floating-point library, fplib. This section contains the following subsections: • 3.2.1 Calling fplib routines on page 3-109. • 3.2.2 fplib arithmetic on numbers in a particular format on page 3-110. • 3.2.3 fplib conversions between floats, long longs, doubles, and ints on page 3-111. • 3.2.4 fplib comparisons between floats and doubles on page 3-112. • 3.2.5 fplib C99 functions on page 3-113. 3.2.1 Calling fplib routines Floating-point routines have names like __aeabi_dadd (add two doubles) and __aeabi_fdiv (divide two floats). User programs can call these routines directly. Even in environments with a coprocessor, the routines are provided. They are typically only a few instructions long because all they do is execute the appropriate coprocessor instruction. All the fplib routines are called using a software floating-point variant of the calling standard. This means that floating-point arguments are passed and returned in integer registers. By contrast, if the program is compiled for a coprocessor, floating-point data is passed in its floating-point registers. So, for example, __aeabi_dadd takes a double in registers r0 and r1, and another double in registers r2 and r3, and returns the sum in r0 and r1. Note For a double in registers r0 and r1, the register that holds the high 32 bits of the double depends on whether your program is little-endian or big-endian. Software floating-point library routines are declared in one of two header files: • A small number of fplib routines that implement C99 functionality are declared in the standard header file math.h. • All other fplib routines are declared in the header file rt_fp.h. You can include this file if you want to call an fplib routine directly. To call a function from assembler, the software floating-point function is named fn. For example, to call the nextafter() function, implement the following code: IMPORT nextafter BL nextafter Related concepts 3.2 The software floating-point library, fplib on page 3-109. Related references 3.2.2 fplib arithmetic on numbers in a particular format on page 3-110. 3.2.3 fplib conversions between floats, long longs, doubles, and ints on page 3-111. 3.2.4 fplib comparisons between floats and doubles on page 3-112. 3.2.5 fplib C99 functions on page 3-113. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-109 3 Floating-point Support 3.2 The software floating-point library, fplib Related information Application Binary Interface (ABI) for the ARM Architecture. Compiler support for floating-point computations and linkage. 3.2.2 fplib arithmetic on numbers in a particular format fplib provides a number of routines to perform arithmetic on numbers in a particular format. The following table describes these routines. Arguments and return types are always in the same format. Table 3-1 Arithmetic routines Function Argument types Return type Operation __aeabi_fadd 2 float float Return x plus y __aeabi_fsub 2 float float Return x minus y __aeabi_frsub 2 float float Return y minus x __aeabi_fmul 2 float float Return x times y __aeabi_fdiv 2 float float Return x divided by y _frdiv 2 float float Return y divided by x _frem 2 float float Return remainder of x by y (see note a) _frnd float float Return x rounded to an integer (see note b) _fsqrt float float Return square root of x __aeabi_dadd 2 double double Return x plus y __aeabi_dsub 2 double double Return x minus y __aeabi_drsub 2 double double Return y minus x __aeabi_dmul 2 double double Return x times y __aeabi_ddiv 2 double double Return x divided by y _drdiv 2 double double Return y divided by x _drem 2 double double Return remainder of x by y (see notes a and c)ce _drnd double double Return x rounded to an integer (see note b)d _dsqrt double double Return square root of x Related concepts 3.2 The software floating-point library, fplib on page 3-109. 3.2.1 Calling fplib routines on page 3-109. Related references 3.2.3 fplib conversions between floats, long longs, doubles, and ints on page 3-111. 3.2.4 fplib comparisons between floats and doubles on page 3-112. 3.2.5 fplib C99 functions on page 3-113. c d e Functions that perform the IEEE 754 remainder operation. This is defined to take two numbers, x and y, and return a number z so that z = x – ny, where n is an integer. To return an exactly correct result, n is chosen so that z is no bigger than half of x (so that z might be negative even if both x and y are positive). The IEEE 754 remainder function is not the same as the operation performed by the C library function fmod, where z always has the same sign as x. Where the IEEE 754 specification gives two acceptable choices of n, the even one is chosen. This behavior occurs independently of the current rounding mode. Functions that perform the IEEE 754 round-to-integer operation. This takes a number and rounds it to an integer (in accordance with the current rounding mode), but returns that integer in the floating-point number format rather than as a C int variable. To convert a number to an int variable, you must use the _ffix routines. The IEEE 754 remainder() function is a synonym for _drem. remainder() is defined in math.h. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-110 3 Floating-point Support 3.2 The software floating-point library, fplib Related information Application Binary Interface (ABI) for the ARM Architecture. 3.2.3 fplib conversions between floats, long longs, doubles, and ints fplib provides a number of routines to perform conversions between number formats. The following table describes these routines. Table 3-2 Number format conversion routines f Function Argument types Return type __aeabi_f2d float double __aeabi_d2f double float _fflt int float _ffltu unsigned int float _dflt int double _dfltu unsigned int double _ffix float int _ffix_r float int _ffixu float unsigned int f _ffixu_r float unsigned int _dfix double int f _dfix_r double int _dfixu double unsigned int _dfixu_r double unsigned int _ll_sto_f long long float _ll_uto_f unsigned long long float _ll_sto_d long long _ll_uto_d unsigned long long double _ll_sfrom_f float long long _ll_sfrom_f_r float long long _ll_ufrom_f float unsigned long long _ll_ufrom_f_r float unsigned long long _ll_sfrom_d long long f double f double f _ll_sfrom_d_r double long long _ll_ufrom_d double unsigned long long _ll_ufrom_d_r double unsigned long long f f Rounded toward zero, independently of the current rounding mode. This is because the C standard requires implicit conversions to integers to round this way, so it is convenient not to have to change the rounding mode to do so. Each function has a corresponding function with _r on the end of its name, that performs the same operation but rounds according to the current mode. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-111 3 Floating-point Support 3.2 The software floating-point library, fplib Related concepts 3.2 The software floating-point library, fplib on page 3-109. 3.2.1 Calling fplib routines on page 3-109. Related references 3.2.2 fplib arithmetic on numbers in a particular format on page 3-110. 3.2.4 fplib comparisons between floats and doubles on page 3-112. 3.2.5 fplib C99 functions on page 3-113. Related information Application Binary Interface (ABI) for the ARM Architecture. 3.2.4 fplib comparisons between floats and doubles fplib provides a number of routines to perform comparisons between floating-point numbers. The following table describes these routines. Table 3-3 Floating-point comparison routines ARM DUI0378G_02 Function Argument types Return type Condition tested Notes _fcmpeq 2 float Flags, EQ/NE x equal to y a _fcmpge 2 float Flags, HS/LO x greater than or equal to y a, b _fcmple 2 float Flags, HI/LS x less than or equal to y a, b _feq 2 float Boolean x equal to y - _fneq 2 float Boolean x not equal to y - _fgeq 2 float Boolean x greater than or equal to y b _fgr 2 float Boolean x greater than y b _fleq 2 float Boolean x less than or equal to y b _fls 2 float Boolean x less than y b _dcmpeq 2 double Flags, EQ/NE x equal to y a _dcmpge 2 double Flags, HS/LO x greater than or equal to y a, b _dcmple 2 double Flags, HI/LS x less than or equal to y a, b _deq 2 double Boolean x equal to y - _dneq 2 double Boolean x not equal to y - _dgeq 2 double Boolean x greater than or equal to y b _dgr 2 double Boolean x greater than y b _dleq 2 double Boolean x less than or equal to y b _dls 2 double Boolean x less than y b _fcmp4 2 float Flags, VFP x less than or equal to y c _fcmp4e 2 float Flags, VFP x less than or equal to y b, c _fdcmp4 float, double Flags, VFP x less than or equal to y c _fdcmp4e float, double Flags, VFP x less than or equal to y b, c _dcmp4 Flags, VFP x less than or equal to y c 2 double Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-112 3 Floating-point Support 3.2 The software floating-point library, fplib Table 3-3 Floating-point comparison routines (continued) Function Argument types Return type Condition tested Notes _dcmp4e 2 double Flags, VFP x less than or equal to y b, c _dfcmp4 double, float Flags, VFP x less than or equal to y c _dfcmp4e double, float Flags, VFP x less than or equal to y b, c Notes on floating-point comparison routines a Returns results in the ARM condition flags. This is efficient in assembly language, because you can directly follow a call to the function with a conditional instruction, but it means there is no way to use this function from C. This function is not declared in rt_fp.h. b Causes an Invalid Operation exception if either argument is a NaN, even a quiet NaN. Other functions only cause Invalid Operation if an argument is an SNaN. QNaNs return not equal when compared to anything, including other QNaNs (so comparing a QNaN to the same QNaN still returns not equal). c Returns VFP-type status flags in the PSR. Also returns VFP-type status flags in the top four bits of r0, meaning that it is possible to use this function from C. This function is declared in rt_fp.h. Related concepts 3.2 The software floating-point library, fplib on page 3-109. 3.2.1 Calling fplib routines on page 3-109. Related references 3.2.2 fplib arithmetic on numbers in a particular format on page 3-110. 3.2.3 fplib conversions between floats, long longs, doubles, and ints on page 3-111. 3.2.5 fplib C99 functions on page 3-113. 3.2.5 fplib C99 functions fplib provides a number of routines that implement C99 functionality. The following table describes these functions. Table 3-4 fplib C99 functions Function Argument types Return type Returns section Standard ilogb double int Exponent of argument x 7.12.6.5 ilogbf float int Exponent of argument x 7.12.6.5 ilogbl long double int Exponent of argument x 7.12.6.5 logb double double Exponent of argument x 7.12.6.11 logbf float float Exponent of argument x 7.12.6.11 logbl long double long double Exponent of argument x 7.12.6.11 scalbn double, int double x * (FLT_RADIX ** n) 7.12.6.13 scalbnf float, int float x * (FLT_RADIX ** n) 7.12.6.13 scalbnl long double, int long double x * (FLT_RADIX ** n) 7.12.6.13 ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-113 3 Floating-point Support 3.2 The software floating-point library, fplib Table 3-4 fplib C99 functions (continued) Function Argument types Return type Returns section Standard scalbln double, long int double x * (FLT_RADIX ** n) 7.12.6.13 scalblnf float, long int float x * (FLT_RADIX ** n) 7.12.6.13 scalblnl long double, long int long double x * (FLT_RADIX ** n) 7.12.6.13 nextafter 2 double double Next representable value after x towards y 7.12.11.3 nextafterf 2 float float Next representable value after x towards y 7.12.11.3 nextafterl 2 long double long double Next representable value after x towards y 7.12.11.3 nexttoward double, long double double Next representable value after x towards y 7.12.11.4 nexttowardf float, long double float Next representable value after x towards y 7.12.11.4 nexttowardl 2 long double long double Next representable value after x towards y 7.12.11.4 Related concepts 3.2 The software floating-point library, fplib on page 3-109. 3.2.1 Calling fplib routines on page 3-109. Related references 3.2.2 fplib arithmetic on numbers in a particular format on page 3-110. 3.2.3 fplib conversions between floats, long longs, doubles, and ints on page 3-111. 3.2.4 fplib comparisons between floats and doubles on page 3-112. Related concepts 3.2.1 Calling fplib routines on page 3-109. Related references 3.2.2 fplib arithmetic on numbers in a particular format on page 3-110. 3.2.3 fplib conversions between floats, long longs, doubles, and ints on page 3-111. 3.2.4 fplib comparisons between floats and doubles on page 3-112. 3.2.5 fplib C99 functions on page 3-113. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-114 3 Floating-point Support 3.3 Controlling the ARM floating-point environment 3.3 Controlling the ARM floating-point environment The ARM compilation tools supply several different interfaces to the floating-point environment, for compatibility and porting ease. These interfaces enable you to change the rounding mode, enable and disable trapping of exceptions, and install your own custom exception trap handlers. This section contains the following subsections: • 3.3.1 Floating-point functions for compatibility with Microsoft products on page 3-115. • 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. • 3.3.3 C99 rounding mode and floating-point exception macros on page 3-116. • 3.3.4 Exception flag handling on page 3-116. • 3.3.5 Functions for handling rounding modes on page 3-117. • 3.3.6 Functions for saving and restoring the whole floating-point environment on page 3-118. • 3.3.7 Functions for temporarily disabling exceptions on page 3-118. • 3.3.8 ARM floating-point compiler extensions to the C99 interface on page 3-119. • 3.3.9 Writing a custom exception trap handler on page 3-120. • 3.3.10 Example of a custom exception handler on page 3-124. • 3.3.11 Exception trap handling by signals on page 3-125. 3.3.1 Floating-point functions for compatibility with Microsoft products Functions defined in float.h give compatibility with Microsoft products to ease porting of floatingpoint code to the ARM architecture. These functions require you to select a floating-point model that supports exceptions. For example, --fpmode=ieee_full or --fpmode=ieee_fixed. Related concepts 3.3 Controlling the ARM floating-point environment on page 3-115. Related references 5.1 _clearfp() on page 5-206. 5.2 _controlfp() on page 5-207. 5.8 _statusfp() on page 5-217. 3.3.2 C99-compatible functions for controlling the ARM floating-point environment The compiler supports all functions defined in the C99 standard, and functions that are not C99-standard. Note The following functionality requires you to select a floating-point model that supports exceptions, such as --fpmode=ieee_full or --fpmode=ieee_fixed. The C99-compatible functions are the only interface that enables you to install custom exception trap handlers with the ability to define your own return value. All the function prototypes, data types, and macros for this functionality are defined in fenv.h. C99 defines two data types, fenv_t and fexcept_t. The C99 standard does not give information about these types, so for portable code you must treat them as opaque. The compiler defines them to be structure types. The type fenv_t is defined to hold all the information about the current floating-point environment. This comprises: • The rounding mode. • The exception sticky flags. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-115 3 Floating-point Support 3.3 Controlling the ARM floating-point environment • • Whether each exception is masked. What handlers are installed, if any. The type fexcept_t is defined to hold all the information relevant to a given set of exceptions. Related concepts 3.3 Controlling the ARM floating-point environment on page 3-115. 3.3.4 Exception flag handling on page 3-116. 3.3.5 Functions for handling rounding modes on page 3-117. 3.3.6 Functions for saving and restoring the whole floating-point environment on page 3-118. 3.3.7 Functions for temporarily disabling exceptions on page 3-118. Related references 3.3.3 C99 rounding mode and floating-point exception macros on page 3-116. 3.3.8 ARM floating-point compiler extensions to the C99 interface on page 3-119. Related information --fpmode=model compiler option. 3.3.3 C99 rounding mode and floating-point exception macros C99 defines a macro for each rounding mode and each exception Note The following functionality requires you to select a floating-point model that supports exceptions, such as --fpmode=ieee_full or --fpmode=ieee_fixed. The C99 rounding mode and exception macros are: • FE_DIVBYZERO • FE_INEXACT • FE_INVALID • FE_OVERFLOW • FE_UNDERFLOW • FE_ALL_EXCEPT • FE_DOWNWARD • FE_TONEAREST • FE_TOWARDZERO • FE_UPWARD The exception macros are bit fields. The macro FE_ALL_EXCEPT is the bitwise OR of all of them. Related concepts 3.3.5 Functions for handling rounding modes on page 3-117. Related references 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. Related information --fpmode=model compiler option. --fpmode=model compiler option. 3.3.4 Exception flag handling The feclearexcept(), fetestexcept(), and feraiseexcept() functions let you clear, test and raise exceptions. The fegetexceptflag() and fesetexceptflag() functions let you save and restore information about a given exception. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-116 3 Floating-point Support 3.3 Controlling the ARM floating-point environment Note The following functionality requires you to select a floating-point model that supports exceptions, such as --fpmode=ieee_full or --fpmode=ieee_fixed. C99 defines these functions as follows: void feclearexcept(int excepts); int fetestexcept(int excepts); void feraiseexcept(int excepts); The feclearexcept() function clears the sticky flags for the given exceptions. The fetestexcept() function returns the bitwise OR of the sticky flags for the given exceptions, so that if the Overflow flag was set but the Underflow flag was not, then calling fetestexcept(FE_OVERFLOW|FE_UNDERFLOW) would return FE_OVERFLOW. The feraiseexcept() function raises the given exceptions, in unspecified order. If an exception trap is enabled for an exception raised this way, it is called. C99 also provides functions to save and restore all information about a given exception. This includes the sticky flag, whether the exception is trapped, and the address of the trap handler, if any. These functions are: void fegetexceptflag(fexcept_t *flagp, int excepts); void fesetexceptflag(const fexcept_t *flagp, int excepts); The fegetexceptflag() function copies all the information relating to the given exceptions into the fexcept_t variable provided. The fesetexceptflag() function copies all the information relating to the given exceptions from the fexcept_t variable into the current floating-point environment. Note You can use fesetexceptflag() to set the sticky flag of a trapped exception to 1 without calling the trap handler, whereas feraiseexcept() calls the trap handler for any trapped exception. Related references 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. Related information --fpmode=model compiler option. --fpmode=model compiler option. 3.3.5 Functions for handling rounding modes The fegetround() and fesetround functions let you get and set the current rounding mode. Note The following functionality requires you to select a floating-point model that supports exceptions, such as --fpmode=ieee_full or --fpmode=ieee_fixed. C99 defines these functions as follows: int fegetround(void); int fesetround(int round); The fegetround() function returns the current rounding mode. The current rounding mode has a value equal to one of the C99 rounding mode macros or exceptions. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-117 3 Floating-point Support 3.3 Controlling the ARM floating-point environment The fesetround() function sets the current rounding mode to the value provided. fesetround() returns zero for success, or nonzero if its argument is not a valid rounding mode. Related references 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. 3.3.3 C99 rounding mode and floating-point exception macros on page 3-116. Related information --fpmode=model compiler option. --fpmode=model compiler option. 3.3.6 Functions for saving and restoring the whole floating-point environment The fegetenv and fesetenv functions let you save and restore the entire floating-point environment. Note The following functionality requires you to select a floating-point model that supports exceptions, such as --fpmode=ieee_full or --fpmode=ieee_fixed. C99 defines these functions as follows: void fegetenv(fenv_t *envp); void fesetenv(const fenv_t *envp); The fegetenv() function stores the current state of the floating-point environment into the fenv_t variable provided. The fesetenv() function restores the environment from the variable provided. Like fesetexceptflag(), fesetenv() does not call trap handlers when it sets the sticky flags for trapped exceptions. Related references 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. Related information --fpmode=model compiler option. --fpmode=model compiler option. 3.3.7 Functions for temporarily disabling exceptions The feholdexcept and feupdateenv functions let you temporarily disable exception trapping. Note The following functionality requires you to select a floating-point model that supports exceptions, such as --fpmode=ieee_full or --fpmode=ieee_fixed. These functions let you avoid risking exception traps when executing code that might cause exceptions. This is useful when, for example, trapped exceptions are using the ARM default behavior. The default is to cause SIGFPE and terminate the application. int feholdexcept(fenv_t *envp); void feupdateenv(const fenv_t *envp); The feholdexcept() function saves the current floating-point environment in the fenv_t variable provided, sets all exceptions to be untrapped, and clears all the exception sticky flags. You can then execute code that might cause unwanted exceptions, and make sure the sticky flags for those exceptions are cleared. Then you can call feupdateenv(). This restores any exception traps and calls them if ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-118 3 Floating-point Support 3.3 Controlling the ARM floating-point environment necessary. For example, suppose you have a function, frob(), that might cause the Underflow or Invalid Operation exceptions (assuming both exceptions are trapped). You are not interested in Underflow, but you want to know if an invalid operation is attempted. You can implement the following code to do this: fenv_t env; feholdexcept(&env); frob(); feclearexcept(FE_UNDERFLOW); feupdateenv(&env); Then, if the frob() function raises Underflow, it is cleared again by feclearexcept(), so no trap occurs when feupdateenv() is called. However, if frob() raises Invalid Operation, the sticky flag is set when feupdateenv() is called, so the trap handler is invoked. This mechanism is provided by C99 because C99 specifies no way to change exception trapping for individual exceptions. A better method is to use __ieee_status() to disable the Underflow trap while leaving the Invalid Operation trap enabled. This has the advantage that the Invalid Operation trap handler is provided with all the information about the invalid operation (that is, what operation was being performed, and on what data), and can invent a result for the operation. Using the C99 method, the Invalid Operation trap handler is called after the fact, receives no information about the cause of the exception, and is called too late to provide a substitute result. Related references 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. 5.5 __ieee_status() on page 5-212. Related information --fpmode=model compiler option. --fpmode=model compiler option. 3.3.8 ARM floating-point compiler extensions to the C99 interface The ARM C library provides some extensions to the C99 interface to enable it to do everything that the ARM floating-point environment is capable of. This includes trapping and untrapping individual exception types, and installing custom trap handlers. Note The following functionality requires you to select a floating-point model that supports exceptions, such as --fpmode=ieee_full or --fpmode=ieee_fixed. The types fenv_t and fexcept_t are not defined by C99 to be anything in particular. The ARM compiler defines them both to be the same structure type. fenv_t and fexcept_t have the following structure: typedef struct{ unsigned __statusword; __ieee_handler_t __invalid_handler; __ieee_handler_t __divbyzero_handler; __ieee_handler_t __overflow_handler; __ieee_handler_t __underflow_handler; __ieee_handler_t __inexact_handler; } fenv_t, fexcept_t; ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-119 3 Floating-point Support 3.3 Controlling the ARM floating-point environment The members of this structure are: • __statusword, the same status variable that the function __ieee_status() sees, laid out in the same format. • Five function pointers giving the address of the trap handler for each exception. By default, each is NULL. This means that if the exception is trapped, the default exception trap action happens. The default is to cause a SIGFPE signal. typedef struct{ unsigned __statusword; } fenv_t, fexcept_t; Related concepts 3.3 Controlling the ARM floating-point environment on page 3-115. 3.3.9 Writing a custom exception trap handler on page 3-120. 3.3.10 Example of a custom exception handler on page 3-124. Related references 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. 5.5 __ieee_status() on page 5-212. Related information --fpmode=model compiler option. --fpmode=model compiler option. 3.3.9 Writing a custom exception trap handler Custom exception trap handlers let you override the default exception handling behavior. For example, when converting Fortran code you might want to override the division by zero exception to return 1 rather than an invalid operation exception. Note The following functionality requires you to select a floating-point model that supports exceptions, such as --fpmode=ieee_full or --fpmode=ieee_fixed. If you want to install a custom exception trap handler, declare it as a function like this: __softfp __ieee_value_t myhandler(__ieee_value_t op1, __ieee_value_t op2, __ieee_edata_t edata); The value returned from this function is of type __ieee_value_t and is used as the result of the operation that caused the exception. The function must be declared __softfp in order to be usable as a handler. The parameters to this function are: ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-120 3 Floating-point Support 3.3 Controlling the ARM floating-point environment op1, op2 These specify the operands, or the intermediate result, for the operation that caused the exception: • For the Invalid Operation and Divide by Zero exceptions, the original operands are supplied. • For the Inexact Result exception, all that is supplied is the ordinary result that would have been returned anyway. This is provided in op1. • For the Overflow exception, an intermediate result is provided. This result is calculated by working out what the operation would have returned if the exponent range had been big enough, and then adjusting the exponent so that it fits in the format. The exponent is adjusted by 192 (0xC0) in single-precision, and by 1536 (0x600) in double-precision. • If Overflow happens when converting a double to a float, the result is supplied in double format, rounded to single-precision, with the exponent biased by 192. For the Underflow exception, a similar intermediate result is produced, but the bias value is added to the exponent instead of being subtracted. The edata parameter also contains a flag to show whether the intermediate result has had to be rounded up, down, or not at all. The type __ieee_value_t is defined as a union of all the possible types that an operand can be passed as: typedef union{ float __f; float __s; double __d; short __h; unsigned short __uh; int __i; unsigned int __ui; long long __l; unsigned long long __ul; ... /* __STRICT_ANSI__ */ struct { int __word1, __word2; } __str; } __ieee_value_t; /* in and out values passed to traps */ Note If you do not compile with --strict, and you have code that used the older definition of __ieee_value_t which named the fields differently, your older code still works. See the file fenv.h for more information. edata This contains flags that give information about the exception that occurred, and what operation was being performed. (The type __ieee_edata_t is a synonym for unsigned int.) edata flags for exception trap handler The flags contained in edata are: edata & FE_EX_RDIR This is nonzero if the intermediate result in Underflow was rounded down, and 0 if it was rounded up or not rounded. (The difference between the last two is given in the Inexact Result bit.) This bit is meaningless for any other type of exception. edata & FE_EX_exception This is nonzero if the given exception (INVALID, DIVBYZERO, OVERFLOW, UNDERFLOW, or INEXACT) occurred. This enables you to: • Use the same handler function for more than one exception type (the function can test these bits to tell what exception it is supposed to handle). • Determine whether Overflow and Underflow intermediate results have been rounded or are exact. Because the FE_EX_INEXACT bit can be set in combination with either FE_EX_OVERFLOW or FE_EX_UNDERFLOW, you must determine the type of exception that actually occurred by testing Overflow and Underflow before testing Inexact. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-121 3 Floating-point Support 3.3 Controlling the ARM floating-point environment edata & FE_EX_FLUSHZERO This is nonzero if the FZ bit was set when the operation was performed. edata & FE_EX_ROUND_MASK This gives the rounding mode that applies to the operation. This is normally the same as the current rounding mode, unless the operation that caused the exception was a routine such as _ffix, that always rounds toward zero. The available rounding mode values are FE_EX_ROUND_NEAREST, FE_EX_ROUND_PLUSINF, FE_EX_ROUND_MINUSINF and FE_EX_ROUND_ZERO. edata & FE_EX_INTYPE_MASK This gives the type of the operands to the function, as one of the type values shown in the following table. Table 3-5 FE_EX_INTYPE_MASK operand type flags Flag Operand type FE_EX_INTYPE_FLOAT float FE_EX_INTYPE_DOUBLE double FE_EX_INTYPE_FD float double FE_EX_INTYPE_DF double float FE_EX_INTYPE_HALF short FE_EX_INTYPE_INT int FE_EX_INTYPE_UINT unsigned int FE_EX_INTYPE_LONGLONG long long FE_EX_INTYPE_ULONGLONG unsigned long long edata & FE_EX_OUTTYPE_MASK This gives the type of the operands to the function, as one of the type values shown in the following table. Table 3-6 FE_EX_OUTTYPE_MASK operand type flags Flag Operand type FE_EX_OUTTYPE_FLOAT float FE_EX_OUTTYPE_DOUBLE double FE_EX_OUTTYPE_HALF short FE_EX_OUTTYPE_INT int FE_EX_OUTTYPE_UINT unsigned int FE_EX_OUTTYPE_LONGLONG long long FE_EX_OUTTYPE_ULONGLONG unsigned long long ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-122 3 Floating-point Support 3.3 Controlling the ARM floating-point environment edata & FE_EX_FN_MASK This gives the nature of the operation that caused the exception, as one of the operation codes shown in the following table. Table 3-7 FE_EX_FN_MASK operation type flags Flag Operation type FE_EX_FN_ADD Addition. FE_EX_FN_SUB Subtraction. FE_EX_FN_MUL Multiplication. FE_EX_FN_DIV Division. FE_EX_FN_REM Remainder. FE_EX_FN_RND Round to integer. FE_EX_FN_SQRT Square root. FE_EX_FN_CMP Compare. FE_EX_FN_CVT Convert between formats. FE_EX_FN_LOGB Exponent fetching. FE_EX_FN_SCALBN Scaling. Note The FE_EX_INTYPE_MASK flag only specifies the type of the first operand. The second operand is always an int. FE_EX_FN_NEXTAFTER Next representable number. Note Both operands are the same type. Calls to nexttoward cause the value of the second operand to change to a value that is of the same type as the first operand. This does not affect the result. FE_EX_FN_RAISE The exception was raised explicitly, by feraiseexcept() or feupdateenv(). In this case, almost nothing in the edata word is valid. When the operation is a comparison, the result must be returned as if it were an int, and must be one of the four values shown in the following table. Input and output types are the same for all operations except Compare and Convert. Table 3-8 FE_EX_CMPRET_MASK comparison type flags Flag Comparison FE_EX_CMPRET_LESS op1 is less than op2 FE_EX_CMPRET_EQUAL op1 is equal to op2 FE_EX_CMPRET_GREATER op1 is greater than op2 FE_EX_CMPRET_UNORDERED op1 and op2 are not comparable Related concepts 3.3.10 Example of a custom exception handler on page 3-124. 3.3.11 Exception trap handling by signals on page 3-125. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-123 3 Floating-point Support 3.3 Controlling the ARM floating-point environment 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. 3.3 Controlling the ARM floating-point environment on page 3-115. Related references 3.3.8 ARM floating-point compiler extensions to the C99 interface on page 3-119. 3.3.8 ARM floating-point compiler extensions to the C99 interface on page 3-119. 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. 4.30 __rt_raise() on page 4-170. 5.5 __ieee_status() on page 5-212. Related information --fpmode=model compiler option. --strict, --no_strict compiler option. 3.3.10 Example of a custom exception handler This example exception trap handler overrides the division by zero exception to return 1 rather than an invalid operation exception. Note The following functionality requires you to select a floating-point model that supports exceptions, such as --fpmode=ieee_full or --fpmode=ieee_fixed. Suppose you are converting some Fortran code into C. The Fortran numerical standard requires 0 divided by 0 to be 1, whereas IEEE 754 defines 0 divided by 0 to be an Invalid Operation and so by default it returns a quiet NaN. The Fortran code is likely to rely on this behavior, and rather than modifying the code, it is probably easier to make 0 divided by 0 return 1. After the handler is installed, dividing 0.0 by 0.0 returns 1.0. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-124 3 Floating-point Support 3.3 Controlling the ARM floating-point environment Custom exception handler #include #include #include __softfp { __ieee_value_t myhandler(__ieee_value_t op1, __ieee_value_t op2, __ieee_edata_t edata) __ieee_value_t ret; if ((edata & FE_EX_FN_MASK) == FE_EX_FN_DIV) { if ((edata & FE_EX_INTYPE_MASK) == FE_EX_INTYPE_FLOAT) { if (op1.f == 0.0 && op2.f == 0.0) { ret.f = 1.0; return ret; } } if ((edata & FE_EX_INTYPE_MASK) == FE_EX_INTYPE_DOUBLE) { if (op1.d == 0.0 && op2.d == 0.0) { ret.d = 1.0; return ret; } } } /* For all other invalid operations, raise SIGFPE as usual */ raise(SIGFPE); } int main(void) { float i, j, k; fenv_t env; fegetenv(&env); env.statusword |= FE_IEEE_MASK_INVALID; env.invalid_handler = myhandler; fesetenv(&env); i = 0.0; j = 0.0; k = i/j; printf("k is %f\n", k); } Related concepts 3.3.9 Writing a custom exception trap handler on page 3-120. 3.3.11 Exception trap handling by signals on page 3-125. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. 3.3 Controlling the ARM floating-point environment on page 3-115. Related references 3.3.8 ARM floating-point compiler extensions to the C99 interface on page 3-119. 3.3.8 ARM floating-point compiler extensions to the C99 interface on page 3-119. 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. 4.30 __rt_raise() on page 4-170. Related information --fpmode=model compiler option. 3.3.11 Exception trap handling by signals You can use the SIGFPE signal to handle exceptions. Note The following functionality requires you to select a floating-point model that supports exceptions, such as --fpmode=ieee_full or --fpmode=ieee_fixed. If an exception is trapped but the trap handler address is set to NULL, a default trap handler is used. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-125 3 Floating-point Support 3.3 Controlling the ARM floating-point environment The default trap handler raises a SIGFPE signal. The default handler for SIGFPE prints an error message and terminates the program. If you trap SIGFPE, you can declare your signal handler function to have a second parameter that tells you the type of floating-point exception that occurred. This feature is provided for compatibility with Microsoft products. The values are _FPE_INVALID, _FPE_ZERODIVIDE, _FPE_OVERFLOW, _FPE_UNDERFLOW and _FPE_INEXACT. They are defined in float.h. For example: void sigfpe(int sig, int etype){ printf("SIGFPE (%s)\n", etype == _FPE_INVALID ? "Invalid Operation" : etype == _FPE_ZERODIVIDE ? "Divide by Zero" : etype == _FPE_OVERFLOW ? "Overflow" : etype == _FPE_UNDERFLOW ? "Underflow" : etype == _FPE_INEXACT ? "Inexact Result" : "Unknown"); } signal(SIGFPE, (void(*)(int))sigfpe); To generate your own SIGFPE signals with this extra information, you can call the function __rt_raise() instead of the ISO function raise(). For example: __rt_raise(SIGFPE, _FPE_INVALID); __rt_raise() is declared in rt_misc.h. Related concepts 3.3.9 Writing a custom exception trap handler on page 3-120. 3.3.10 Example of a custom exception handler on page 3-124. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. 3.3 Controlling the ARM floating-point environment on page 3-115. Related references 3.3.8 ARM floating-point compiler extensions to the C99 interface on page 3-119. 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. 4.30 __rt_raise() on page 4-170. Related information --fpmode=model compiler option. Related concepts 3.3.1 Floating-point functions for compatibility with Microsoft products on page 3-115. Related references 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. 3.3.8 ARM floating-point compiler extensions to the C99 interface on page 3-119. 5.5 __ieee_status() on page 5-212. 5.3 __fp_status() on page 5-209. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-126 3 Floating-point Support 3.4 Using C99 signaling NaNs provided by mathlib (_WANT_SNAN) 3.4 Using C99 signaling NaNs provided by mathlib (_WANT_SNAN) If you want to use signaling NaNs, you must indicate this to the compiler by defining the macro _WANT_SNAN in your application. This macro must be defined before you include any standard C headers. If your application is comprised of two or more translation units, either all or none of them must define _WANT_SNAN. That is, the definition must be consistent for any given application. You must also use the relevant command-line option when you compile your source code. This is associated with the predefined macro __SUPPORT_SNAN__. Related information Predefined macros. WG14 - C N965, Optional support for Signaling NaNs. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-127 3 Floating-point Support 3.5 mathlib double and single-precision floating-point functions 3.5 mathlib double and single-precision floating-point functions The math library, mathlib, provides double and single-precision functions for mathematical calculations. For example, to calculate a cube root, you can use cbrt() (double-precision) or cbrtf() (singleprecision). ISO/IEC 14882 specifies that in addition to the double versions of the math functions in , C++ adds float (and long double) overloaded versions of these functions. The ARM implementation extends this in scope to include the additional math functions that do not exist in C90, but that do exist in C99. In C++, std::cbrt() on a float argument selects the single-precision version of the function, and the same type of selection applies to other floating-point functions in C++. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-128 3 Floating-point Support 3.6 IEEE 754 arithmetic 3.6 IEEE 754 arithmetic The ARM floating-point environment is an implementation of the IEEE 754 standard for binary floatingpoint arithmetic. This section contains the following subsections: • 3.6.1 Basic data types for IEEE 754 arithmetic on page 3-129. • 3.6.2 Single precision data type for IEEE 754 arithmetic on page 3-129. • 3.6.3 Double precision data type for IEEE 754 arithmetic on page 3-130. • 3.6.4 Sample single precision floating-point values for IEEE 754 arithmetic on page 3-131. • 3.6.5 Sample double precision floating-point values for IEEE 754 arithmetic on page 3-132. • 3.6.6 IEEE 754 arithmetic and rounding on page 3-133. • 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. • 3.6.8 Exception types recognized by the ARM floating-point environment on page 3-134. 3.6.1 Basic data types for IEEE 754 arithmetic ARM floating-point values are stored in one of two data types, single-precision and double-precision. In this documentation, they are called float and double, these being the corresponding C data types. Related concepts 3.6 IEEE 754 arithmetic on page 3-129. 3.6.2 Single precision data type for IEEE 754 arithmetic on page 3-129. 3.6.3 Double precision data type for IEEE 754 arithmetic on page 3-130. 3.6.6 IEEE 754 arithmetic and rounding on page 3-133. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. Related references 3.6.4 Sample single precision floating-point values for IEEE 754 arithmetic on page 3-131. 3.6.5 Sample double precision floating-point values for IEEE 754 arithmetic on page 3-132. Related information IEEE Standard for Floating-Point Arithmetic (IEEE 754), 1985 version. 3.6.2 Single precision data type for IEEE 754 arithmetic A float value is 32 bits wide. The structure is: 31 30 S 23 22 Exp 0 Frac Figure 3-1 IEEE 754 single-precision floating-point format The S field gives the sign of the number. It is 0 for positive, or 1 for negative. The Exp field gives the exponent of the number, as a power of two. It is biased by 0x7F (127), so that very small numbers have exponents near zero and very large numbers have exponents near 0xFF (255). For example: • • • • • ARM DUI0378G_02 If Exp = 0x7D (125), the number is between 0.25 and 0.5 (not including 0.5). If Exp = 0x7E (126), the number is between 0.5 and 1.0 (not including 1.0). If Exp = 0x7F (127), the number is between 1.0 and 2.0 (not including 2.0). If Exp = 0x80 (128), the number is between 2.0 and 4.0 (not including 4.0). If Exp = 0x81 (129), the number is between 4.0 and 8.0 (not including 8.0). Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-129 3 Floating-point Support 3.6 IEEE 754 arithmetic The Frac field gives the fractional part of the number. It usually has an implicit 1 bit on the front that is not stored to save space. For example, if Exp is 0x7F: • If Frac = 00000000000000000000000 (binary), the number is 1.0. • If Frac = 10000000000000000000000 (binary), the number is 1.5. • If Frac = 01000000000000000000000 (binary), the number is 1.25. • If Frac = 11000000000000000000000 (binary), the number is 1.75. In general, the numeric value of a bit pattern in this format is given by the formula: (–1)S * 2(Exp–0x7F) * (1 + Frac * 2–23) Numbers stored in this form are called normalized numbers. The maximum and minimum exponent values, 0 and 255, are special cases. Exponent 255 can represent infinity and store Not a Number (NaN) values. Infinity can occur as a result of dividing by zero, or as a result of computing a value that is too large to store in this format. NaN values are used for special purposes. Infinity is stored by setting Exp to 255 and Frac to all zeros. If Exp is 255 and Frac is nonzero, the bit pattern represents a NaN. Exponent 0 can represent very small numbers in a special way. If Exp is zero, then the Frac field has no implicit 1 on the front. This means that the format can store 0.0, by setting both Exp and Frac to all 0 bits. It also means that numbers that are too small to store using Exp >= 1 are stored with less precision than the ordinary 23 bits. These are called denormals. Related concepts 3.6 IEEE 754 arithmetic on page 3-129. 3.6.3 Double precision data type for IEEE 754 arithmetic on page 3-130. 3.6.6 IEEE 754 arithmetic and rounding on page 3-133. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. Related references 3.6.1 Basic data types for IEEE 754 arithmetic on page 3-129. 3.6.4 Sample single precision floating-point values for IEEE 754 arithmetic on page 3-131. 3.6.5 Sample double precision floating-point values for IEEE 754 arithmetic on page 3-132. Related information IEEE Standard for Floating-Point Arithmetic (IEEE 754), 1985 version. 3.6.3 Double precision data type for IEEE 754 arithmetic A double value is 64 bits wide. The structure is: 63 62 S 52 Exp 51 0 Frac Figure 3-2 IEEE 754 double-precision floating-point format As with single-precision float data types, S is the sign, Exp the exponent, and Frac the fraction. Most of the detail of float values remains true for double values, except that: • The Exp field is biased by 0x3FF (1023) instead of 0x7F, so numbers between 1.0 and 2.0 have an Exp field of 0x3FF. • The Exp value representing infinity and NaNs is 0x7FF (2047) instead of 0xFF. Related concepts 3.6 IEEE 754 arithmetic on page 3-129. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-130 3 Floating-point Support 3.6 IEEE 754 arithmetic 3.6.2 Single precision data type for IEEE 754 arithmetic on page 3-129. 3.6.6 IEEE 754 arithmetic and rounding on page 3-133. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. Related references 3.6.1 Basic data types for IEEE 754 arithmetic on page 3-129. 3.6.4 Sample single precision floating-point values for IEEE 754 arithmetic on page 3-131. 3.6.5 Sample double precision floating-point values for IEEE 754 arithmetic on page 3-132. Related information IEEE Standard for Floating-Point Arithmetic (IEEE 754), 1985 version. 3.6.4 Sample single precision floating-point values for IEEE 754 arithmetic Sample float bit patterns, together with their mathematical values. Table 3-9 Sample single-precision floating-point values Float value S Exp Frac Mathematical value 0x3F800000 0 0x7F 000...000 1.0 0xBF800000 1 0x7F 000...000 -1.0 0x3F800001g 0 0x7F 000...001 1.000 000 119 0x3F400000 0 0x7E 100...000 0.75 0x00800000h 0 0x01 000...000 1.18*10-38 0x00000001i 0 0x00 000...001 1.40*10-45 0x7F7FFFFFj 0 0xFE 111...111 3.40*1038 0x7F800000 0 0xFF 000...000 Plus infinity 0xFF800000 1 0xFF 000...000 Minus infinity 0x00000000k 0 0x00 000...000 0.0 0x7F800001 0 0xFF 000...001 Signaling NaN 0x7FC00000l 0 0xFF 100...000 Quiet NaN Related concepts 3.6 IEEE 754 arithmetic on page 3-129. 3.6.2 Single precision data type for IEEE 754 arithmetic on page 3-129. 3.6.3 Double precision data type for IEEE 754 arithmetic on page 3-130. 3.6.6 IEEE 754 arithmetic and rounding on page 3-133. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. g h i j k l The smallest representable number that can be seen to be greater than 1.0. The amount that it differs from 1.0 is known as the machine epsilon. This is 0.000 000 119 in float, and 0.000 000 000 000 000 222 in double. The machine epsilon gives a rough idea of the number of significant figures the format can keep track of. float can do six or seven places. double can do fifteen or sixteen. The smallest value that can be represented as a normalized number in each format. Numbers smaller than this can be stored as denormals, but are not held with as much precision. The smallest positive number that can be distinguished from zero. This is the absolute lower limit of the format. The largest finite number that can be stored. Attempting to increase this number by addition or multiplication causes overflow and generates infinity (in general). Zero. Strictly speaking, they show plus zero. Zero with a sign bit of 1, minus zero, is treated differently by some operations, although the comparison operations (for example == and !=) report that the two types of zero are equal. There are two types of NaNs, signaling NaNs and quiet NaNs. Quiet NaNs have a 1 in the first bit of Frac, and signaling NaNs have a zero there. The difference is that signaling NaNs cause an exception when used, whereas quiet NaNs do not. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-131 3 Floating-point Support 3.6 IEEE 754 arithmetic Related references 3.6.1 Basic data types for IEEE 754 arithmetic on page 3-129. 3.6.5 Sample double precision floating-point values for IEEE 754 arithmetic on page 3-132. Related information IEEE Standard for Floating-Point Arithmetic (IEEE 754), 1985 version. 3.6.5 Sample double precision floating-point values for IEEE 754 arithmetic Sample double bit patterns, together with their mathematical values. Table 3-10 Sample double-precision floating-point values Double value S Exp Frac Mathematical value 0x3FF0000000000000 0 0x3FF 000...000 1.0 0xBFF0000000000000 1 0x3FF 000...000 -1.0 0x3FF0000000000001m 0 0x3FF 000...001 1.000 000 000 000 000 222 0x3FE8000000000000 0 0x3FE 100...000 0.75 0x0010000000000000n 0 0x001 000...000 2.23*10-308 0x0000000000000001o 0 0x000 000...001 4.94*10-324 0x7FEFFFFFFFFFFFFFp 0 0x7FE 111...111 1.80*10308 0x7FF0000000000000 0 0x7FF 000...000 Plus infinity 0xFFF0000000000000 1 0x7FF 000...000 Minus infinity 0x0000000000000000q 0 0x000 000...000 0.0 0x7FF0000000000001 0 0x7FF 000...001 Signaling NaN 0x7FF8000000000000r 0 0x7FF 100...000 Quiet NaN Related concepts 3.6 IEEE 754 arithmetic on page 3-129. 3.6.2 Single precision data type for IEEE 754 arithmetic on page 3-129. 3.6.3 Double precision data type for IEEE 754 arithmetic on page 3-130. 3.6.6 IEEE 754 arithmetic and rounding on page 3-133. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. Related references 3.6.1 Basic data types for IEEE 754 arithmetic on page 3-129. 3.6.4 Sample single precision floating-point values for IEEE 754 arithmetic on page 3-131. Related information IEEE Standard for Floating-Point Arithmetic (IEEE 754), 1985 version. m n o p q r The smallest representable number that can be seen to be greater than 1.0. The amount that it differs from 1.0 is known as the machine epsilon. This is 0.000 000 119 in float, and 0.000 000 000 000 000 222 in double. The machine epsilon gives a rough idea of the number of significant figures the format can keep track of. float can do six or seven places. double can do fifteen or sixteen. The smallest value that can be represented as a normalized number in each format. Numbers smaller than this can be stored as denormals, but are not held with as much precision. The smallest positive number that can be distinguished from zero. This is the absolute lower limit of the format. The largest finite number that can be stored. Attempting to increase this number by addition or multiplication causes overflow and generates infinity (in general). Zero. Strictly speaking, they show plus zero. Zero with a sign bit of 1, minus zero, is treated differently by some operations, although the comparison operations (for example == and !=) report that the two types of zero are equal. There are two types of NaNs, signaling NaNs and quiet NaNs. Quiet NaNs have a 1 in the first bit of Frac, and signaling NaNs have a zero there. The difference is that signaling NaNs cause an exception when used, whereas quiet NaNs do not. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-132 3 Floating-point Support 3.6 IEEE 754 arithmetic 3.6.6 IEEE 754 arithmetic and rounding IEEE 754 defines different rounding rules to use when calculating arithmetic results. Arithmetic is generally performed by computing the result of an operation as if it were stored exactly (to infinite precision), and then rounding it to fit in the format. Apart from operations whose result already fits exactly into the format (such as adding 1.0 to 1.0), the correct answer is generally somewhere between two representable numbers in the format. The system then chooses one of these two numbers as the rounded result. It uses one of the following methods: Round to nearest The system chooses the nearer of the two possible outputs. If the correct answer is exactly halfway between the two, the system chooses the output where the least significant bit of Frac is zero. This behavior (round-to-even) prevents various undesirable effects. This is the default mode when an application starts up. It is the only mode supported by the ordinary floating-point libraries. Hardware floating-point environments and the enhanced floating-point libraries support all four rounding modes. Round up, or round toward plus infinity The system chooses the larger of the two possible outputs (that is, the one further from zero if they are positive, and the one closer to zero if they are negative). Round down, or round toward minus infinity The system chooses the smaller of the two possible outputs (that is, the one closer to zero if they are positive, and the one further from zero if they are negative). Round toward zero, or chop, or truncate The system chooses the output that is closer to zero, in all cases. Related concepts 3.6 IEEE 754 arithmetic on page 3-129. 3.6.2 Single precision data type for IEEE 754 arithmetic on page 3-129. 3.6.3 Double precision data type for IEEE 754 arithmetic on page 3-130. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. Related references 3.6.1 Basic data types for IEEE 754 arithmetic on page 3-129. 3.6.4 Sample single precision floating-point values for IEEE 754 arithmetic on page 3-131. 3.6.5 Sample double precision floating-point values for IEEE 754 arithmetic on page 3-132. Related information IEEE Standard for Floating-Point Arithmetic (IEEE 754), 1985 version. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic Floating-point arithmetic operations can run into various problems. These are known as exceptions, because they indicate unusual or exceptional situations. For example, the result computed might be either too big or too small to fit into the format, or there might be no way to calculate the result (as in trying to take the square root of a negative number, or trying to divide zero by zero). The ARM floating-point environment can handle an exception by inventing a plausible result for the operation and returning that result, or by trapping the exception. For example, the square root of a negative number can produce a NaN, and trying to compute a value too big to fit in the format can produce infinity. If an exception occurs and is ignored, a flag is set in the floating-point status word to tell you that something went wrong at some time in the past. When an exception occurs, a piece of code called a trap handler is run. The system provides a default trap handler that prints an error message and terminates the application. However, you can supply your ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-133 3 Floating-point Support 3.6 IEEE 754 arithmetic own trap handlers to clean up the exceptional condition in whatever way you choose. Trap handlers can even supply a result to be returned from the operation. For example, if you had an algorithm where it was convenient to assume that 0 divided by 0 was 1, you could supply a custom trap handler for the Invalid Operation exception to identify that particular case and substitute the answer you required. Related concepts 3.3.9 Writing a custom exception trap handler on page 3-120. 3.3.10 Example of a custom exception handler on page 3-124. 3.3.11 Exception trap handling by signals on page 3-125. 3.3 Controlling the ARM floating-point environment on page 3-115. 3.6 IEEE 754 arithmetic on page 3-129. 3.6.2 Single precision data type for IEEE 754 arithmetic on page 3-129. 3.6.3 Double precision data type for IEEE 754 arithmetic on page 3-130. 3.6.6 IEEE 754 arithmetic and rounding on page 3-133. Related references 3.3.8 ARM floating-point compiler extensions to the C99 interface on page 3-119. 3.3.2 C99-compatible functions for controlling the ARM floating-point environment on page 3-115. 4.30 __rt_raise() on page 4-170. 3.6.1 Basic data types for IEEE 754 arithmetic on page 3-129. 3.6.4 Sample single precision floating-point values for IEEE 754 arithmetic on page 3-131. 3.6.5 Sample double precision floating-point values for IEEE 754 arithmetic on page 3-132. Related information IEEE Standard for Floating-Point Arithmetic (IEEE 754), 1985 version. 3.6.8 Exception types recognized by the ARM floating-point environment The ARM floating-point environment recognizes a number of different types of exception. The following types of exception are recognized: Invalid Operation exception This occurs when there is no sensible result for an operation. This can happen for any of the following reasons: • Performing any operation on a signaling NaN, except the simplest operations (copying and changing the sign). • Adding plus infinity to minus infinity, or subtracting an infinity from itself. • Multiplying infinity by zero. • Dividing 0 by 0, or dividing infinity by infinity. • Taking the remainder from dividing anything by 0, or infinity by anything. • Taking the square root of a negative number (not including minus zero). • Converting a floating-point number to an integer if the result does not fit. • Comparing two numbers if one of them is a NaN. If the Invalid Operation exception is not trapped, these operations return a quiet NaN. The exception is conversion to an integer. This returns zero because there are no quiet NaNs in integers. Divide by Zero exception This occurs if you divide a finite nonzero number by zero. Be aware that: • Dividing zero by zero gives an Invalid Operation exception. • Dividing infinity by zero is valid and returns infinity. If Divide by Zero is not trapped, the operation returns infinity. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-134 3 Floating-point Support 3.6 IEEE 754 arithmetic Overflow exception This occurs when the result of an operation is too big to fit into the format. This happens, for example, if you add the largest representable number to itself. The largest float value is 0x7F7FFFFF. If Overflow is not trapped, the operation returns infinity, or the largest finite number, depending on the rounding mode. Underflow exception This can occur when the result of an operation is too small to be represented as a normalized number (with Exp at least 1). The situations that cause Underflow depend on whether it is trapped or not: • If Underflow is trapped, it occurs whenever a result is too small to be represented as a normalized number. • If Underflow is not trapped, it only occurs if the result requires rounding. So, for example, dividing the float number 0x00800000 by 2 does not signal Underflow, because the result 0x00400000 is exact. However, trying to multiply the float number 0x00000001 by 1.5 does signal Underflow. Note For readers familiar with the IEEE 754 specification, the chosen implementation options in the ARM compiler are to detect tininess before rounding, and to detect loss of accuracy as an inexact result. • If Underflow is not trapped, the result is rounded to one of the two nearest representable denormal numbers, according to the current rounding mode. The loss of precision is ignored and the system returns the best result it can. The Inexact Result exception happens whenever the result of an operation requires rounding. This would cause significant loss of speed if it had to be detected on every operation in software, so the ordinary floating-point libraries do not support the Inexact Result exception. The enhanced floating-point libraries, and hardware floating-point systems, all support Inexact Result. If Inexact Result is not trapped, the system rounds the result in the usual way. The flag for Inexact Result is also set by Overflow and Underflow if either one of those is not trapped. All exceptions are untrapped by default. Related concepts 3.3.9 Writing a custom exception trap handler on page 3-120. 3.3.4 Exception flag handling on page 3-116. 3.3.10 Example of a custom exception handler on page 3-124. 3.3.11 Exception trap handling by signals on page 3-125. 3.6 IEEE 754 arithmetic on page 3-129. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. Related references 1.23.3 ISO-compliant implementation of signals supported by the signal() function in the C library and additional type arguments on page 1-84. 3.6.4 Sample single precision floating-point values for IEEE 754 arithmetic on page 3-131. Related information IEEE Standard for Floating-Point Arithmetic (IEEE 754), 1985 version. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-135 3 Floating-point Support 3.6 IEEE 754 arithmetic Related concepts 3.6.2 Single precision data type for IEEE 754 arithmetic on page 3-129. 3.6.3 Double precision data type for IEEE 754 arithmetic on page 3-130. 3.6.6 IEEE 754 arithmetic and rounding on page 3-133. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. Related references 3.6.1 Basic data types for IEEE 754 arithmetic on page 3-129. 3.6.4 Sample single precision floating-point values for IEEE 754 arithmetic on page 3-131. 3.6.5 Sample double precision floating-point values for IEEE 754 arithmetic on page 3-132. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-136 3 Floating-point Support 3.7 Using the Vector Floating-Point (VFP) support libraries 3.7 Using the Vector Floating-Point (VFP) support libraries The VFP support libraries are used by the VFP Support Code. The VFP Support Code is executed from an undefined instruction trap that is triggered when an exceptional floating-point condition occurs. Related information Limitations on hardware handling of floating-point arithmetic. Implementation of Vector Floating-Point (VFP) support code. Using VFP with RVDS, Application Note 133. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 3-137 Chapter 4 The C and C++ Library Functions reference Describes the standard C and C++ library functions that are extensions to the C Standard or that differ in some way to the standard. Some of the standard functions interact with the ARM retargetable semihosting environment. Such functions are also documented. It contains the following sections: • 4.1 __aeabi_errno_addr() on page 4-140. • 4.2 alloca() on page 4-141. • 4.3 clock() on page 4-142. • 4.4 _clock_init() on page 4-143. • 4.5 __default_signal_handler() on page 4-144. • 4.6 errno on page 4-145. • 4.7 _findlocale() on page 4-146. • 4.8 _fisatty() on page 4-147. • 4.9 _get_lconv() on page 4-148. • 4.10 getenv() on page 4-149. • 4.11 _getenv_init() on page 4-150. • 4.12 __heapstats() on page 4-151. • 4.13 __heapvalid() on page 4-152. • 4.14 lconv structure on page 4-153. • 4.15 localeconv() on page 4-155. • 4.16 _membitcpybl(), _membitcpybb(), _membitcpyhl(), _membitcpyhb(), _membitcpywl(), _membitcpywb(), _membitmovebl(), _membitmovebb(), _membitmovehl(), _membitmovehb(), _membitmovewl(), _membitmovewb() on page 4-156. • 4.17 posix_memalign() on page 4-157. • 4.18 #pragma import(_main_redirection) on page 4-158. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-138 4 The C and C++ Library Functions reference • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • • ARM DUI0378G_02 4.19 __raise() on page 4-159. 4.20 _rand_r() on page 4-160. 4.21 remove() on page 4-161. 4.22 rename() on page 4-162. 4.23 __rt_entry on page 4-163. 4.24 __rt_errno_addr() on page 4-164. 4.25 __rt_exit() on page 4-165. 4.26 __rt_fp_status_addr() on page 4-166. 4.27 __rt_heap_extend() on page 4-167. 4.28 __rt_lib_init() on page 4-168. 4.29 __rt_lib_shutdown() on page 4-169. 4.30 __rt_raise() on page 4-170. 4.31 __rt_stackheap_init() on page 4-171. 4.32 setlocale() on page 4-172. 4.33 _srand_r() on page 4-174. 4.34 strcasecmp() on page 4-175. 4.35 strncasecmp() on page 4-176. 4.36 strlcat() on page 4-177. 4.37 strlcpy() on page 4-178. 4.38 _sys_close() on page 4-179. 4.39 _sys_command_string() on page 4-180. 4.40 _sys_ensure() on page 4-181. 4.41 _sys_exit() on page 4-182. 4.42 _sys_flen() on page 4-183. 4.43 _sys_istty() on page 4-184. 4.44 _sys_open() on page 4-185. 4.45 _sys_read() on page 4-186. 4.46 _sys_seek() on page 4-187. 4.47 _sys_tmpnam() on page 4-188. 4.48 _sys_write() on page 4-189. 4.49 system() on page 4-190. 4.50 time() on page 4-191. 4.51 _ttywrch() on page 4-192. 4.52 __user_heap_extend() on page 4-193. 4.53 __user_heap_extent() on page 4-194. 4.54 __user_setup_stackheap() on page 4-195. 4.55 __vectab_stack_and_reset on page 4-196. 4.56 wcscasecmp() on page 4-197. 4.57 wcsncasecmp() on page 4-198. 4.58 wcstombs() on page 4-199. 4.59 Thread-safe C library functions on page 4-200. 4.60 C library functions that are not thread-safe on page 4-202. 4.61 Legacy function __user_initial_stackheap() on page 4-204. Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-139 4 The C and C++ Library Functions reference 4.1 __aeabi_errno_addr() 4.1 __aeabi_errno_addr() The __aeabi_errno_addr() returns the address of the C library errno variable when the C library attempts to read or write errno. Syntax volatile int *__aeabi_errno_addr(void); Usage The library provides a default implementation. It is unlikely that you have to re-implement this function. This function is not part of the C library standard, but the ARM C library supports it as an extension. Related references 4.6 errno on page 4-145. 4.24 __rt_errno_addr() on page 4-164. Related information C Library ABI for the ARM Architecture. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-140 4 The C and C++ Library Functions reference 4.2 alloca() 4.2 alloca() Defined in alloca.h, the alloca() function allocates local storage in a function. It returns a pointer to the number of bytes of memory allocated. Syntax void *alloca(size_t size); Usage The default implementation returns an eight-byte aligned block of memory on the stack. Memory returned from alloca() must never be passed to free(). Instead, the memory is de-allocated automatically when the function that called alloca() returns. Note alloca() must not be called through a function pointer. You must take care when using alloca() and setjmp() in the same function, because memory allocated by alloca() between calling setjmp() and longjmp() is de-allocated by the call to longjmp(). This function is a common nonstandard extension to many C libraries. Returns Returns in size a pointer to the number of bytes of memory allocated. Related concepts 1.5.3 ARM C libraries and thread-safe functions on page 1-25. Related references 1.7.1 Building an application without the C library on page 1-40. 4.59 Thread-safe C library functions on page 4-200. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-141 4 The C and C++ Library Functions reference 4.3 clock() 4.3 clock() This is the standard C library clock function from time.h. Syntax clock_t clock(void); Usage The default implementation of this function uses semihosting. If the units of clock_t differ from the default of centiseconds, you must define __CLK_TCK on the compiler command line or in your own header file. The value in the definition is used for CLK_TCK and CLOCKS_PER_SEC. The default value is 100 for centiseconds. Note If you re-implement clock() you must also re-implement _clock_init(). Returns The returned value is an unsigned integer. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-142 4 The C and C++ Library Functions reference 4.4 _clock_init() 4.4 _clock_init() Defined in rt_misc.h, the _clock_init() function is an initialization function for clock(). It is not part of the C library standard, but the ARM C library supports it as an extension. Syntax void _clock_init(void); Usage This is a function that you can re-implement in an implementation-specific way. It is called from the library initialization code, so you do not have to call it from your application code. Note You must re-implement this function if you re-implement clock(). The initialization that _clock_init() applies enables clock() to return the time that has elapsed since the program was started. An example of how you might re-implement _clock_init() might be to set the timer to zero. However, if your implementation of clock() relies on a system timer that cannot be reset, then _clock_init() could instead read the time at startup (when called from the library initialization code), with clock() subsequently subtracting the time that was read at initialization, from the current value of the timer. In both cases, some form of initialization is required of _clock_init(). Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-143 4 The C and C++ Library Functions reference 4.5 __default_signal_handler() 4.5 __default_signal_handler() Defined in rt_misc.h, the __default_signal_handler() function handles a raised signal. The default action is to print an error message and exit. This function is not part of the C library standard, but the ARM C library supports it as an extension. Syntax int __default_signal_handler(int signal, int type); Usage The default signal handler returns a nonzero value to indicate that the caller has to arrange for the program to exit. You can replace the default signal handler by defining: int __default_signal_handler(int signal, int type); The interface is the same as __raise(), but this function is only called after the C signal handling mechanism has declined to process the signal. A complete list of the defined signals is in signal.h. Note The signals used by the libraries might change in future releases of ARM Compiler. Related references 1.23.3 ISO-compliant implementation of signals supported by the signal() function in the C library and additional type arguments on page 1-84. 1.6.7 Indirect semihosting C library function dependencies on page 1-37. 4.19 __raise() on page 4-159. 4.51 _ttywrch() on page 4-192. 4.41 _sys_exit() on page 4-182. 4.30 __rt_raise() on page 4-170. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-144 4 The C and C++ Library Functions reference 4.6 errno 4.6 errno The C library errno variable is defined in the implicit static data area of the library. This area is identified by __user_libspace(). The function that returns the address of errno is: (*(volatile int *) __aeabi_errno_addr()) You can define __aeabi_errno_addr() if you want to place errno at a user-defined location instead of the default location identified by __user_libspace(). Note Legacy versions of errno.h might define errno in terms of __rt_errno_addr() rather than __aeabi_errno_addr(). The function name __rt_errno_addr() is a legacy from pre-ABI versions of the tools, and is still supported to ensure that object files generated with those tools link successfully. Returns The return value is a pointer to a variable of type int, containing the currently applicable instance of errno. Related concepts 1.5.4 Use of static data in the C libraries on page 1-25. Related references 4.1 __aeabi_errno_addr() on page 4-140. 4.24 __rt_errno_addr() on page 4-164. 1.5.5 Use of the __user_libspace static data area by the C libraries on page 1-26. Related information Application Binary Interface for the ARM Architecture. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-145 4 The C and C++ Library Functions reference 4.7 _findlocale() 4.7 _findlocale() Defined in rt_locale.h, _findlocale() searches a set of contiguous locale data blocks for the requested locale, and returns a pointer to that locale. This function is not part of the C library standard, but the ARM C library supports it as an extension. Syntax void const *_findlocale(void const *index, const char *name); Where: index is a pointer to a set of locale data blocks that are contiguous in memory and that end with a terminating value (set by the LC_index_end macro). name is the name of the locale to find. Usage You can use _findlocale() as an optional helper function when defining your own locale setup. The _get_lc_*() functions, for example, _get_lc_ctype(), are expected to return a pointer to a locale definition created using the assembler macros. If you only want to write one locale definition, you can write an implementation of _get_lc_ctype() that always returns the same pointer. However, if you want to use different locale definitions at runtime, then the _get_lc_*() functions have to be able to return a different data block depending on the name passed to them as an argument. _findlocale() provides an easy way to do this. Returns Returns a pointer to the requested data block. Related concepts 1.9 Assembler macros that tailor locale functions in the C library on page 1-52. 1.9.2 Runtime selection of the locale subsystem in the C library on page 1-53. Related references 1.9.1 Link time selection of the locale subsystem in the C library on page 1-52. 1.9.3 Definition of locale data blocks in the C library on page 1-53. 4.14 lconv structure on page 4-153. 4.9 _get_lconv() on page 4-148. 4.15 localeconv() on page 4-155. 4.32 setlocale() on page 4-172. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-146 4 The C and C++ Library Functions reference 4.8 _fisatty() 4.8 _fisatty() Defined in stdio.h, the _fisatty() function determines whether the given stdio stream is attached to a terminal device or a normal file. It calls the _sys_istty() low-level function on the underlying file handle. This function is not part of the C library standard, but the ARM C library supports it as an extension. Syntax int _fisatty(FILE *stream); The return value indicates the stream destination: 0 A file. 1 A terminal. Negative An error. Related references 4.43 _sys_istty() on page 4-184. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-147 4 The C and C++ Library Functions reference 4.9 _get_lconv() 4.9 _get_lconv() Defined in locale.h, _get_lconv() performs the same function as the standard C library function, localeconv(), except that it delivers the result in user-provided memory instead of an internal static variable. _get_lconv() sets the components of an lconv structure with values appropriate for the formatting of numeric quantities. Syntax void _get_lconv(struct lconv *lc); Usage This extension to the ISO C library does not use any static data. If you are building an application that must conform strictly to the ISO C standard, use localeconv() instead. Returns The existing lconv structure lc is filled with formatting data. Related references 4.7 _findlocale() on page 4-146. 4.14 lconv structure on page 4-153. 4.15 localeconv() on page 4-155. 4.32 setlocale() on page 4-172. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-148 4 The C and C++ Library Functions reference 4.10 getenv() 4.10 getenv() This is the standard C library getenv() function from stdlib.h. It gets the value of a specified environment variable. Syntax char *getenv(const char *name); Usage The default implementation returns NULL, indicating that no environment information is available. If you re-implement getenv(), ARM recommends that you re-implement it in such a way that it searches some form of environment list for the input string, name. The set of environment names and the method for altering the environment list are implementation-defined. getenv() does not depend on any other function, and no other function depends on getenv(). A function closely associated with getenv() is _getenv_init(). _getenv_init() is called during startup if it is defined, to enable a user re-implementation of getenv() to initialize itself. Returns The return value is a pointer to a string associated with the matched list member. The array pointed to must not be modified by the program, but might be overwritten by a subsequent call to getenv(). ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-149 4 The C and C++ Library Functions reference 4.11 _getenv_init() 4.11 _getenv_init() Defined in rt_misc.h, the _getenv_init() function enables a user version of getenv() to initialize itself. This function is not part of the C library standard, but the ARM C library supports it as an extension. Syntax void _getenv_init(void); Usage If this function is defined, the C library initialization code calls it when the library is initialized, that is, before main() is entered. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-150 4 The C and C++ Library Functions reference 4.12 __heapstats() 4.12 __heapstats() Defined in stdlib.h, the __heapstats() function displays statistics on the state of the storage allocation heap. Syntax void __heapstats(int (*dprint)(void *param, char const *format,...), void *param); Usage The default implementation in the compiler gives information on how many free blocks exist, and estimates their size ranges. The __heapstats() function generates output as follows: 32272 bytes in 2 free blocks (avge size 16136) 1 blocks 2^12+1 to 2^13 1 blocks 2^13+1 to 2^14 Line 1 of the output displays the total number of bytes, the number of free blocks, and the average size. The following lines give an estimate of the size of each block in bytes, expressed as a range. __heapstats() does not give information on the number of used blocks. The function outputs its results by calling the output function dprint(), that must work like fprintf(). The first parameter passed to dprint() is the supplied pointer param. You can pass fprintf() itself, provided you cast it to the right function pointer type. This type is defined as a typedef for convenience. It is called __heapprt. For example: __heapstats((__heapprt)fprintf, stderr); Note If you call fprintf() on a stream that you have not already sent output to, the library calls malloc() internally to create a buffer for the stream. If this happens in the middle of a call to __heapstats(), the heap might be corrupted. Therefore, you must ensure you have already sent some output to stderr. If you are using the default one-region memory model, heap memory is allocated only as it is required. This means that the amount of free heap changes as you allocate and deallocate memory. For example, the sequence: int *ip; __heapstats((__heapprt)fprintf,stderr); ip = malloc(200000); free(ip); __heapstats((__heapprt)fprintf,stderr); // print initial free heap size // print heap size after freeing gives output such as: 4076 bytes in 1 free blocks (avge size 4076) 1 blocks 2^10+1 to 2^11 2008180 bytes in 1 free blocks (avge size 2008180) 1 blocks 2^19+1 to 2^20 This function is not part of the C library standard, but the ARM C library supports it as an extension. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-151 4 The C and C++ Library Functions reference 4.13 __heapvalid() 4.13 __heapvalid() Defined in stdlib.h, the __heapvalid() function performs a consistency check on the heap. Syntax int __heapvalid(int (*dprint)(void *param, char const *format,...), void *param, int verbose); Usage __heapvalid() outputs full information about every free block if the verbose parameter is nonzero. Otherwise, it only outputs errors. The function outputs its results by calling the output function dprint(), that must work like fprintf(). The first parameter passed to dprint() is the supplied pointer param. You can pass fprintf() itself, provided you cast it to the right function pointer type. This type is defined as a typedef for convenience. It is called __heapprt. For example: __heapvalid((__heapprt) fprintf, stderr, 0); Note If you call fprintf() on a stream that you have not already sent output to, the library calls malloc() internally to create a buffer for the stream. If this happens in the middle of a call to __heapvalid(), the heap might be corrupted. You must therefore ensure you have already sent some output to stderr. The example code fails if you have not already written to the stream. This function is not part of the C library standard, but the ARM C library supports it as an extension. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-152 4 The C and C++ Library Functions reference 4.14 lconv structure 4.14 lconv structure Defined in locale.h, the lconv structure contains numeric formatting information The structure is filled by the functions _get_lconv() and localeconv(). The definition of lconv from locale.h is as follows. struct lconv { char *decimal_point; /* The decimal point character used to format non monetary quantities */ char *thousands_sep; /* The character used to separate groups of digits to the left of the */ /* decimal point character in formatted non monetary quantities. */ char *grouping; /* A string whose elements indicate the size of each group of digits */ /* in formatted non monetary quantities. See below for more details. */ char *int_curr_symbol; /* The international currency symbol applicable to the current locale.*/ /* The first three characters contain the alphabetic international */ /* currency symbol in accordance with those specified in ISO 4217. */ /* Codes for the representation of Currency and Funds. The fourth */ /* character (immediately preceding the null character) is the */ /* character used to separate the international currency symbol from */ /* the monetary quantity. */ char *currency_symbol; /* The local currency symbol applicable to the current locale. */ char *mon_decimal_point; /* The decimal point used to format monetary quantities. */ char *mon_thousands_sep; /* The separator for groups of digits to the left of the decimal point*/ /* in formatted monetary quantities. */ char *mon_grouping; /* A string whose elements indicate the size of each group of digits */ /* in formatted monetary quantities. See below for more details. */ char *positive_sign; /* The string used to indicate a non negative-valued formatted */ /* monetary quantity. */ char *negative_sign; /* The string used to indicate a negative-valued formatted monetary */ /* quantity. */ char int_frac_digits; /* The number of fractional digits (those to the right of the */ /* decimal point) to be displayed in an internationally formatted */ /* monetary quantities. */ char frac_digits; /* The number of fractional digits (those to the right of the */ /* decimal point) to be displayed in a formatted monetary quantity. */ char p_cs_precedes; /* Set to 1 or 0 if the currency_symbol respectively precedes or */ /* succeeds the value for a non negative formatted monetary quantity. */ char p_sep_by_space; /* Set to 1 or 0 if the currency_symbol respectively is or is not */ /* separated by a space from the value for a non negative formatted */ /* monetary quantity. */ char n_cs_precedes; /* Set to 1 or 0 if the currency_symbol respectively precedes or */ /* succeeds the value for a negative formatted monetary quantity. */ char n_sep_by_space; /* Set to 1 or 0 if the currency_symbol respectively is or is not */ /* separated by a space from the value for a negative formatted */ /* monetary quantity. */ char p_sign_posn; /* Set to a value indicating the position of the positive_sign for a */ /* non negative formatted monetary quantity. See below for more details*/ char n_sign_posn; /* Set to a value indicating the position of the negative_sign for a */ /* negative formatted monetary quantity. */ }; The elements of grouping and mon_grouping are interpreted as follows: CHAR_MAX No additional grouping is to be performed. 0 The previous element is repeated for the remainder of the digits. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-153 4 The C and C++ Library Functions reference 4.14 lconv structure other The value is the number of digits that comprise the current group. The next element is examined to determine the size of the next group of digits to the left of the current group. The value of p_sign_posn and n_sign_posn are interpreted as follows: 0 Parentheses surround the quantity and currency symbol. 1 The sign string precedes the quantity and currency symbol. 2 The sign string is after the quantity and currency symbol. 3 The sign string immediately precedes the currency symbol. 4 The sign string immediately succeeds the currency symbol. Related references 4.7 _findlocale() on page 4-146. 4.9 _get_lconv() on page 4-148. 4.15 localeconv() on page 4-155. 4.32 setlocale() on page 4-172. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-154 4 The C and C++ Library Functions reference 4.15 localeconv() 4.15 localeconv() Defined in stdlib.h, localeconv() creates and sets the components of an lconv structure with values appropriate for the formatting of numeric quantities according to the rules of the current locale. Syntax struct lconv *localeconv(void); Usage The members of the structure with type char * are strings. Any of these, except for decimal_point, can point to an empty string, "", to indicate that the value is not available in the current locale or is of zero length. The members with type char are non-negative numbers. Any of the members can be CHAR_MAX to indicate that the value is not available in the current locale. This function is not thread-safe, because it uses an internal static buffer. _get_lconv() provides a thread-safe alternative. Returns The function returns a pointer to the filled-in object. The structure pointed to by the return value is not modified by the program, but might be overwritten by a subsequent call to the localeconv() function. In addition, calls to the setlocale() function with categories LC_ALL, LC_MONETARY, or LC_NUMERIC might overwrite the contents of the structure. Related references 4.7 _findlocale() on page 4-146. 4.14 lconv structure on page 4-153. 4.9 _get_lconv() on page 4-148. 4.32 setlocale() on page 4-172. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-155 4 The C and C++ Library Functions reference 4.16 _membitcpybl(), _membitcpybb(), _membitcpyhl(), _membitcpyhb(), _membitcpywl(), _membitcpywb(), _membitmovebl(), _membitmovebb(), _membitmovehl(), _membitmovehb(), _membitmovewl(), _membitmovewb() 4.16 _membitcpybl(), _membitcpybb(), _membitcpyhl(), _membitcpyhb(), _membitcpywl(), _membitcpywb(), _membitmovebl(), _membitmovebb(), _membitmovehl(), _membitmovehb(), _membitmovewl(), _membitmovewb() Similar to the standard C library memcpy() and memmove() functions, these nonstandard C library functions provide bit-aligned memory operations. They are defined in string.h. Syntax void _membitcpy[b|h|w][b|l](void *dest, const void *src, int dest_offset, int src_offset, size_t nbits); void _membitmove[b|h|w][b|l](void *dest, const void *src, int dest_offset, int src_offset, size_t nbits); Usage The number of contiguous bits specified by nbits is copied, or moved (depending on the function being used), from a memory location starting src_offset bits after (or before if a negative offset) the address pointed to by src, to a location starting dest_offset bits after (or before if a negative offset) the address pointed to by dest. To define a contiguous sequence of bits, a form of ordering is required. The variants of each function define this order, as follows: • Functions whose second-last character is b, for example _membitcpybl(), are byte-oriented. Byteoriented functions consider all of the bits in one byte to come before the bits in the next byte. • Functions whose second-last character is h are halfword-oriented. • Functions whose second-last character is w are word-oriented. Within each byte, halfword, or word, the bits can be considered to go in different order depending on the endianness. Functions ending in b, for example _membitmovewb(), are bitwise big-endian. This means that the Most Significant Bit (MSB) of each byte, halfword, or word (as appropriate) is considered to be the first bit in the word, and the Least Significant Bit (LSB) is considered to be the last. Functions ending in l are bitwise little-endian. They consider the LSB to come first and the MSB to come last. As with memcpy() and memmove(), the bitwise memory copying functions copy as fast as they can in their assumption that source and destination memory regions do not overlap, whereas the bitwise memory move functions ensure that source data in overlapping regions is copied before being overwritten. On a little-endian platform, the bitwise big-endian functions are distinct, but the bitwise little-endian functions use the same bit ordering, so they are synonymous symbols that refer to the same function. On a big-endian platform, the bitwise big-endian functions are all effectively the same, but the bitwise littleendian functions are distinct. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-156 4 The C and C++ Library Functions reference 4.17 posix_memalign() 4.17 posix_memalign() Defined in stdlib.h, the posix_memalign() function provides aligned memory allocation. This function is fully POSIX-compliant. Syntax int posix_memalign(void **memptr, size_t alignment, size_t size); Usage This function allocates size bytes of memory at an address that is a multiple of alignment. The value of alignment must be a power of two and a multiple of sizeof(void *). You can free memory allocated by posix_memalign() using the standard C library free() function. Returns The returned address is written to the void * variable pointed to by memptr. The integer return value from the function is zero on success, or an error code on failure. If no block of memory can be found with the requested size and alignment, the function returns ENOMEM and the value of *memptr is undefined. Related information The Open Group Base Specifications, IEEE Std 1003.1. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-157 4 The C and C++ Library Functions reference 4.18 #pragma import(_main_redirection) 4.18 #pragma import(_main_redirection) This pragma enables automatic command-line redirection. Defining this pragma lets you use the < and > command-line operators to redirect the standard input, output, and error streams at program startup. If you do not define this pragma and attempt to use redirection operators on the command-line, the redirection operators and associated filenames are passed to the program as ordinary argument strings. Syntax #pragma import(_main_redirection) Related information Environment. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-158 4 The C and C++ Library Functions reference 4.19 __raise() 4.19 __raise() Defined in rt_misc.h, the __raise() function raises a signal to indicate a runtime anomaly. It is not part of the C library standard, but the ARM C library supports it as an extension. Syntax int __raise(int signal, int type); where: signal is an integer that holds the signal number. type is an integer, string constant or variable that provides additional information about the circumstances that the signal was raised in, for some kinds of signal. Usage If the user has configured the handling of the signal by calling signal() then __raise() takes the action specified by the user. That is, either to ignore the signal or to call the user-provided handler function. Otherwise, __raise() calls __default_signal_handler(), which provides the default signal handling behavior. You can replace the __raise() function by defining: int __raise(int signal, int type); This enables you to bypass the C signal mechanism and its data-consuming signal handler vector, but otherwise gives essentially the same interface as: int __default_signal_handler(int signal, int type); The default signal handler of the library uses the type parameter of __raise() to vary the messages it outputs. Returns There are three possibilities for a __raise() return condition: no return The handler performs a long jump or restart. 0 The signal was handled. nonzero The calling code must pass that return value to the exit code. The default library implementation calls _sys_exit(rc) if __raise() returns a nonzero return code rc. Related concepts 1.5.11 Thread safety in the ARM C library on page 1-31. Related references 1.23.3 ISO-compliant implementation of signals supported by the signal() function in the C library and additional type arguments on page 1-84. 1.6.7 Indirect semihosting C library function dependencies on page 1-37. 4.5 __default_signal_handler() on page 4-144. 4.51 _ttywrch() on page 4-192. 4.41 _sys_exit() on page 4-182. 4.30 __rt_raise() on page 4-170. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-159 4 The C and C++ Library Functions reference 4.20 _rand_r() 4.20 _rand_r() Defined in stdlib.h, the _rand_r() function is a reentrant version of the rand() function. Syntax int _rand_r(struct _rand_state * buffer); where: buffer is a pointer to a user-supplied buffer storing the state of the random number generator. Usage This function enables you to explicitly supply your own buffer in thread-local storage. Related references 4.33 _srand_r() on page 4-174. 4.60 C library functions that are not thread-safe on page 4-202. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-160 4 The C and C++ Library Functions reference 4.21 remove() 4.21 remove() This is the standard C library remove() function from stdio.h. Syntax int remove(const char *filename); Usage The default implementation of this function uses semihosting. remove() causes the file whose name is the string pointed to by filename to be removed. Subsequent attempts to open the file result in failure, unless it is created again. If the file is open, the behavior of the remove() function is implementation-defined. Returns Returns zero if the operation succeeds or nonzero if it fails. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-161 4 The C and C++ Library Functions reference 4.22 rename() 4.22 rename() This is the standard C library rename() function from stdio.h. Syntax int rename(const char *old, const char *new); Usage The default implementation of this function uses semihosting. rename() causes the file whose name is the string pointed to by old to be subsequently known by the name given by the string pointed to by new. The file named old is effectively removed. If a file named by the string pointed to by new exists prior to the call of the rename() function, the behavior is implementation-defined. Returns Returns zero if the operation succeeds or nonzero if it fails. If the operation returns nonzero and the file existed previously it is still known by its original name. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-162 4 The C and C++ Library Functions reference 4.23 __rt_entry 4.23 __rt_entry The symbol __rt_entry is the starting point for a program using the ARM C library. Control passes to __rt_entry after all scatter-loaded regions have been relocated to their execution addresses. Usage The default implementation of __rt_entry: 1. Sets up the heap and stack. 2. Initializes the C library by calling __rt_lib_init. 3. Calls main(). 4. Shuts down the C library, by calling __rt_lib_shutdown. 5. Exits. __rt_entry must end with a call to one of the following functions: exit() Calls atexit()-registered functions and shuts down the library. __rt_exit() Shuts down the library but does not call atexit() functions. _sys_exit() Exits directly to the execution environment. It does not shut down the library and does not call atexit() functions. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-163 4 The C and C++ Library Functions reference 4.24 __rt_errno_addr() 4.24 __rt_errno_addr() The __rt_errno_addr() function is called to get the address of the C library errno variable when the C library attempts to read or write errno. Syntax volatile int *__rt_errno_addr(void); Usage The library provides a default implementation. It is unlikely that you have to reimplement this function. This function is not part of the C library standard, but the ARM C library supports it as an extension. Note This function is associated with pre-ABI versions of the compilation tools. However, it remains supported to ensure that object files compiled with those tools link successfully. Unless you are working with object files compiled with pre-ABI versions of the tools, use __aeabi_errno_addr() instead of __rt_errno_addr(). Related references 4.1 __aeabi_errno_addr() on page 4-140. 4.6 errno on page 4-145. Related information Application Binary Interface for the ARM Architecture. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-164 4 The C and C++ Library Functions reference 4.25 __rt_exit() 4.25 __rt_exit() Defined in rt_misc.h, the __rt_exit() function shuts down the library but does not call functions registered with atexit(). atexit()-registered functions are called by exit(). The __rt_exit() function is not part of the C library standard, but the ARM C library supports it as an extension. Syntax void __rt_exit(int code); Where code is not used by the standard function. Usage Shuts down the C library by calling __rt_lib_shutdown(), and then calls _sys_exit() to terminate the application. Reimplement _sys_exit() rather than __rt_exit(). Returns This function does not return. Related references 4.41 _sys_exit() on page 4-182. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-165 4 The C and C++ Library Functions reference 4.26 __rt_fp_status_addr() 4.26 __rt_fp_status_addr() Defined in rt_fp.h, the __rt_fp_status_addr() function returns the address of the floating-point status word. By default, the floating-point status word resides in __user_libspace. This function is not part of the C library standard, but the ARM C library supports it as an extension. Syntax unsigned *__rt_fp_status_addr(void); Usage If __rt_fp_status_addr() is not defined, the default implementation from the C library is used. The value is initialized when __rt_lib_init() calls _fp_init(). The constants for the status word are listed in fenv.h. The default floating-point status is 0. Returns The address of the floating-point status word. Related concepts 1.5.11 Thread safety in the ARM C library on page 1-31. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-166 4 The C and C++ Library Functions reference 4.27 __rt_heap_extend() 4.27 __rt_heap_extend() Defined in rt_heap.h, the __rt_heap_extend() function returns a new eight-byte aligned block of memory to add to the heap, if possible. If you reimplement __rt_stackheap_init(), you must reimplement this function. An incomplete prototype implementation is in rt_memory.s. This function is not part of the C library standard, but the ARM C library supports it as an extension. Syntax extern unsigned __rt_heap_extend(unsigned size, void **block); Usage The calling convention is ordinary AAPCS. On entry, r0 is the minimum size of the block to add, and r1 holds a pointer to a location to store the base address. The default implementation has the following characteristics: • The returned size must be either: — A multiple of eight bytes of at least the requested size. — 0, denoting that the request cannot be honored. • The returned base address is aligned on an eight-byte boundary. • Size is measured in bytes. • The function is subject only to ARM Architecture Procedure Call Standard (AAPCS) constraints. Returns The default implementation extends the heap if there is sufficient free heap memory. If it cannot, it calls __user_heap_extend() if it is implemented. On exit, r0 is the size of the block acquired, or 0 if nothing could be obtained, and the memory location r1 pointed to on entry contains the base address of the block. Related concepts 1.11.3 Stack pointer initialization and heap bounds on page 1-64. Related references 4.31 __rt_stackheap_init() on page 4-171. 4.52 __user_heap_extend() on page 4-193. 4.53 __user_heap_extent() on page 4-194. 4.54 __user_setup_stackheap() on page 4-195. Related information Procedure Call Standard for the ARM Architecture. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-167 4 The C and C++ Library Functions reference 4.28 __rt_lib_init() 4.28 __rt_lib_init() Defined in rt_misc.h, this is the library initialization function and is the companion to __rt_lib_shutdown(). Syntax extern value_in_regs struct __argc_argv __rt_lib_init(unsigned heapbase, unsigned heaptop); where: heapbase is the start of the heap memory block. heaptop is the end of the heap memory block. Usage This function is called immediately after __rt_stackheap_init() and is passed an initial chunk of memory to use as a heap. This function is the standard ARM C library initialization function and it must not be reimplemented. Returns This function returns argc and argv ready to be passed to main(). The structure is returned in the registers as: struct __argc_argv { int argc; char **argv; int r2, r3; // optional extra arguments that on entry to main() are }; // found in registers R2 and R3. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-168 4 The C and C++ Library Functions reference 4.29 __rt_lib_shutdown() 4.29 __rt_lib_shutdown() Defined in rt_misc.h, __rt_lib_shutdown() is the library shutdown function and is the companion to __rt_lib_init(). Syntax void __rt_lib_shutdown(void); Usage This function is provided in case a user must call it directly. This is the standard ARM C library shutdown function and it must not be reimplemented. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-169 4 The C and C++ Library Functions reference 4.30 __rt_raise() 4.30 __rt_raise() Defined in rt_misc.h, the __rt_raise() function raises a signal to indicate a runtime anomaly. This function is not part of the C library standard, but the ARM C library supports it as an extension. Syntax void __rt_raise(int signal, int type); where: signal is an integer that holds the signal number. type is an integer, string constant or variable that provides additional information about the circumstances that the signal was raised in, for some kinds of signal. Usage Redefine this function to replace the entire signal handling mechanism for the library. The default implementation calls __raise(). Depending on the value returned from __raise(): no return The handler performed a long jump or restart and __rt_raise() does not regain control. 0 The signal was handled and __rt_raise() exits. nonzero The default library implementation calls _sys_exit(rc) if __raise() returns a nonzero return code rc. Related references 1.23.3 ISO-compliant implementation of signals supported by the signal() function in the C library and additional type arguments on page 1-84. 4.5 __default_signal_handler() on page 4-144. 4.19 __raise() on page 4-159. 4.51 _ttywrch() on page 4-192. 4.41 _sys_exit() on page 4-182. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-170 4 The C and C++ Library Functions reference 4.31 __rt_stackheap_init() 4.31 __rt_stackheap_init() Defined in rt_misc.h, the __rt_stackheap_init() function sets up the stack pointer and returns a region of memory for use as the initial heap. It is called from the library initialization code. On return from this function, SP must point to the top of the stack region, r0 must point to the base of the heap region, and r1 must point to the limit of the heap region. A user-defined memory model (that is, __rt_stackheap_init() and __rt_heap_extend()) is allocated 16 bytes of storage from the __user_perproc_libspace area if wanted. It accesses this storage by calling __rt_stackheap_storage() to return a pointer to its 16-byte region. This function is not part of the C library standard, but the ARM C library supports it as an extension. Related concepts 1.11.3 Stack pointer initialization and heap bounds on page 1-64. Related references 4.27 __rt_heap_extend() on page 4-167. 4.52 __user_heap_extend() on page 4-193. 4.53 __user_heap_extent() on page 4-194. 4.54 __user_setup_stackheap() on page 4-195. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-171 4 The C and C++ Library Functions reference 4.32 setlocale() 4.32 setlocale() Defined in locale.h, the setlocale() function selects the appropriate locale as specified by the category and locale arguments. Syntax char *setlocale(int category, const char *locale); Usage Use the setlocale() function to change or query part or all of the current locale. The effect of the category argument for each value is: LC_COLLATE Affects the behavior of strcoll(). LC_CTYPE Affects the behavior of the character handling functions. LC_MONETARY Affects the monetary formatting information returned by localeconv(). LC_NUMERIC Affects the decimal-point character for the formatted input/output functions and the string conversion functions and the numeric formatting information returned by localeconv(). LC_TIME Can affect the behavior of strftime(). For currently supported locales, the option has no effect. LC_ALL Affects all locale categories. This is the bitwise OR of all the locale categories. A value of "C" for locale specifies the minimal environment for C translation. An empty string, "", for locale specifies the implementation-defined native environment. At program startup, the equivalent of setlocale(LC_ALL, "C") is executed. Valid locale values depend on which __use_X_ctype symbol is imported (__use_iso8859_ctype, __use_sjis_ctype, or __use_utf8_ctype), and on user-defined locales. Note Only one __use_X_ctype symbol can be imported. Returns If a pointer to a string is given for locale and the selection is valid, the string associated with the specified category for the new locale is returned. If the selection cannot be honored, a null pointer is returned and the locale is not changed. A null pointer for locale causes the string associated with the category for the current locale to be returned and the locale is not changed. If category is LC_ALL and the most recent successful locale-setting call uses a category other than LC_ALL, a composite string might be returned. The string returned when used in a subsequent call with its associated category restores that part of the program locale. The string returned is not modified by the program, but might be overwritten by a subsequent call to setlocale(). Related concepts Shift-JIS and UTF-8 implementation on page 1-53. Related references ISO8859-1 implementation on page 1-52. 1.9.3 Definition of locale data blocks in the C library on page 1-53. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-172 4 The C and C++ Library Functions reference 4.32 setlocale() 4.7 _findlocale() on page 4-146. 4.14 lconv structure on page 4-153. 4.9 _get_lconv() on page 4-148. 4.15 localeconv() on page 4-155. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-173 4 The C and C++ Library Functions reference 4.33 _srand_r() 4.33 _srand_r() Defined in stdlib.h, this is a reentrant version of the srand() function. Syntax int _srand_r(struct _rand_state * buffer, unsigned int seed); where: buffer is a pointer to a user-supplied buffer storing the state of the random number generator. seed is a seed for a new sequence of pseudo-random numbers to be returned by subsequent calls to _rand_r(). Usage This function enables you to explicitly supply your own buffer that can be used for thread-local storage. If _srand_r() is repeatedly called with the same seed value, the same sequence of pseudo-random numbers is repeated. If _rand_r() is called before any calls to _srand_r() have been made with the same buffer, undefined behavior occurs because the buffer is not initialized. Related references 4.20 _rand_r() on page 4-160. 4.60 C library functions that are not thread-safe on page 4-202. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-174 4 The C and C++ Library Functions reference 4.34 strcasecmp() 4.34 strcasecmp() Defined in string.h, the strcasecmp() function performs a case-insensitive string comparison test. Syntax extern _ARMABI int strcasecmp(const char *s1, const char *s2); Related information Application Binary Interface for the ARM Architecture. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-175 4 The C and C++ Library Functions reference 4.35 strncasecmp() 4.35 strncasecmp() Defined in string.h, the strncasecmp() function performs a case-insensitive string comparison test of not more than a specified number of characters. Syntax extern _ARMABI int strncasecmp(const char *s1, const char *s2, size_t n); Related information Application Binary Interface for the ARM Architecture. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-176 4 The C and C++ Library Functions reference 4.36 strlcat() 4.36 strlcat() Defined in string.h, the strlcat() function concatenates two strings. Syntax extern size_t strlcat(char *dst, const char *src, size_t size); Usage strlcat() appends up to size-strlen(dst)-1 bytes from the NUL-terminated string src to the end of dst. It takes the full size of the buffer, not only the length, and terminates the result with NUL as long as size is greater than 0. Include a byte for the NUL in your size value. The strlcat() function returns the total length of the string that would have been created if there was unlimited space. This might or might not be equal to the length of the string actually created, depending on whether there was enough space. This means that you can call strlcat() once to find out how much space is required, then allocate it if you do not have enough, and finally call strlcat() a second time to create the required string. This function is a common BSD-derived extension to many C libraries. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-177 4 The C and C++ Library Functions reference 4.37 strlcpy() 4.37 strlcpy() Defined in string.h, the strlcpy() function copies up to size-1 characters from the NUL-terminated string src to dst. Syntax extern size_t strlcpy(char *dst, const char *src, size_t size); Usage strlcpy() takes the full size of the buffer, not only the length, and terminates the result with NUL as long as size is greater than 0. Include a byte for the NUL in your size value. The strlcpy() function returns the total length of the string that would have been copied if there was unlimited space. This might or might not be equal to the length of the string actually copied, depending on whether there was enough space. This means that you can call strlcpy() once to find out how much space is required, then allocate it if you do not have enough, and finally call strlcpy() a second time to do the required copy. This function is a common BSD-derived extension to many C libraries. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-178 4 The C and C++ Library Functions reference 4.38 _sys_close() 4.38 _sys_close() Defined in rt_sys.h, the _sys_close() function closes a file previously opened with _sys_open(). Syntax int _sys_close(FILEHANDLE fh); Usage This function must be defined if any input/output function is to be used. Returns The return value is 0 if successful. A nonzero value indicates an error. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-179 4 The C and C++ Library Functions reference 4.39 _sys_command_string() 4.39 _sys_command_string() Defined in rt_sys.h, the _sys_command_string() function retrieves the command line that invoked the current application from the environment that called the application. Syntax char *_sys_command_string(char *cmd, int len); where: cmd is a pointer to a buffer that can store the command line. It is not required that the command line is stored in cmd. len is the length of the buffer. Usage This function is called by the library startup code to set up argv and argc to pass to main(). Note You must not assume that the C library is fully initialized when this function is called. For example, you must not call malloc() from within this function. This is because the C library startup sequence calls this function before the heap is fully configured. Returns The function must return either: • A pointer to the command line, if successful. This can be either a pointer to the cmd buffer if it is used, or a pointer to wherever else the command line is stored. • NULL, if not successful. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-180 4 The C and C++ Library Functions reference 4.40 _sys_ensure() 4.40 _sys_ensure() This function is deprecated. It is never called by any other library function, and you are not required to re-implement it if you are retargeting standard I/O (stdio). ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-181 4 The C and C++ Library Functions reference 4.41 _sys_exit() 4.41 _sys_exit() Defined in rt_sys.h, this is the library exit function. All exits from the library eventually call _sys_exit(). Syntax void _sys_exit(int return_code); Usage This function must not return. You can intercept application exit at a higher level by either: • Implementing the C library function exit() as part of your application. You lose atexit() processing and library shutdown if you do this. • Implementing the function __rt_exit(int n) as part of your application. You lose library shutdown if you do this, but atexit() processing is still performed when exit() is called or main() returns. Returns The return code is advisory. An implementation might attempt to pass it to the execution environment. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. 4.5 __default_signal_handler() on page 4-144. 4.19 __raise() on page 4-159. 4.51 _ttywrch() on page 4-192. 4.30 __rt_raise() on page 4-170. 4.25 __rt_exit() on page 4-165. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-182 4 The C and C++ Library Functions reference 4.42 _sys_flen() 4.42 _sys_flen() Defined in rt_sys.h, the _sys_flen() function returns the current length of a file. Syntax long _sys_flen(FILEHANDLE fh); Usage This function is used by _sys_seek() to convert an offset relative to the end of a file into an offset relative to the beginning of the file. You do not have to define _sys_flen() if you do not intend to use fseek(). If you retarget at system _sys_*() level, you must supply _sys_flen(), even if the underlying system directly supports seeking relative to the end of a file. Returns This function returns the current length of the file fh, or a negative error indicator. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-183 4 The C and C++ Library Functions reference 4.43 _sys_istty() 4.43 _sys_istty() Defined in rt_sys.h, the _sys_istty() function determines if a file handle identifies a terminal. Syntax int _sys_istty(FILEHANDLE fh); Usage When a file is connected to a terminal device, this function provides unbuffered behavior by default (in the absence of a call to set(v)buf) and prohibits seeking. Returns The return value is one of the following values: 0 There is no interactive device. 1 There is an interactive device. other An error occurred. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. 4.8 _fisatty() on page 4-147. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-184 4 The C and C++ Library Functions reference 4.44 _sys_open() 4.44 _sys_open() Defined in rt_sys.h, the _sys_open() function opens a file. Syntax FILEHANDLE _sys_open(const char *name, int openmode); Usage The _sys_open() function is required by fopen() and freopen(). These functions in turn are required if any file input/output function is to be used. The openmode parameter is a bitmap whose bits mostly correspond directly to the ISO mode specification. Target-dependent extensions are possible, but freopen() must also be extended. Returns The return value is –1 if an error occurs. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-185 4 The C and C++ Library Functions reference 4.45 _sys_read() 4.45 _sys_read() Defined in rt_sys.h, the _sys_read() function reads the contents of a file into a buffer. Syntax int _sys_read(FILEHANDLE fh, unsigned char *buf, unsigned len, int mode); Note The mode parameter is here for historical reasons. It contains nothing useful and must be ignored. Returns The return value is one of the following: • • • The number of bytes not read (that is, len minus the number of bytes that were read). An error indication. An EOF indicator. The EOF indication involves the setting of 0x80000000 in the normal result. Reading up to and including the last byte of data does not turn on the EOF indicator. The EOF indicator is only reached when an attempt is made to read beyond the last byte of data. The target-independent code is capable of handling: • The EOF indicator being returned in the same read as the remaining bytes of data that precede the EOF. • The EOF indicator being returned on its own after the remaining bytes of data have been returned in a previous read. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-186 4 The C and C++ Library Functions reference 4.46 _sys_seek() 4.46 _sys_seek() Defined in rt_sys.h, the _sys_seek() function puts the file pointer at offset pos from the beginning of the file. Syntax int _sys_seek(FILEHANDLE fh, long pos); Usage This function sets the current read or write position to the new location pos relative to the start of the current file fh. Returns The result is: • Negative if an error occurs. • Non-negative if no error occurs. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-187 4 The C and C++ Library Functions reference 4.47 _sys_tmpnam() 4.47 _sys_tmpnam() Defined in rt_sys.h, the _sys_tmpnam() function converts the file number fileno for a temporary file to a unique filename, for example, tmp0001. Syntax void _sys_tmpnam(char *name, int fileno, unsigned maxlength); Usage The function must be defined if tmpnam() or tmpfile() is used. Returns Returns the filename in name. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-188 4 The C and C++ Library Functions reference 4.48 _sys_write() 4.48 _sys_write() Defined in rt_sys.h, the _sys_write() function writes the contents of a buffer to a file previously opened with _sys_open(). Syntax int _sys_write(FILEHANDLE fh, const unsigned char *buf, unsigned len, int mode); Note The mode parameter is here for historical reasons. It contains nothing useful and must be ignored. Returns The return value is either: • A positive number representing the number of characters not written (so any nonzero return value denotes a failure of some sort). • A negative number indicating an error. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-189 4 The C and C++ Library Functions reference 4.49 system() 4.49 system() This is the standard C library system() function from stdlib.h. Syntax int system(const char *string); Usage The default implementation of this function uses semihosting. system() passes the string pointed to by string to the host environment to be executed by a command processor in an implementation-defined manner. A null pointer can be used for string, to inquire whether a command processor exists. Returns If the argument is a NULL pointer, the system function returns nonzero only if a command processor is available. If the argument is not a NULL pointer, the system() function returns an implementation-defined value. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-190 4 The C and C++ Library Functions reference 4.50 time() 4.50 time() This is the standard C library time() function from time.h. The default implementation of this function uses semihosting. Syntax time_t time(time_t *timer); The return value is an approximation of the current calendar time. Returns The value -1 is returned if the calendar time is not available. If timer is not a NULL pointer, the return value is also stored in timer. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-191 4 The C and C++ Library Functions reference 4.51 _ttywrch() 4.51 _ttywrch() Defined in rt_sys.h, the _ttywrch() function writes a character to the console. The console might have been redirected. You can use this function as a last resort error handling routine. Syntax void _ttywrch(int ch); Usage The default implementation of this function uses semihosting. You can redefine this function, or __raise(), even if there is no other input/output. For example, it might write an error message to a log kept in nonvolatile memory. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. 4.5 __default_signal_handler() on page 4-144. 4.19 __raise() on page 4-159. 4.41 _sys_exit() on page 4-182. 4.30 __rt_raise() on page 4-170. Related information What is Semihosting?. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-192 4 The C and C++ Library Functions reference 4.52 __user_heap_extend() 4.52 __user_heap_extend() Defined in rt_misc.h, the __user_heap_extend() function can be defined to return extra blocks of memory, separate from the initial one, to be used by the heap. If defined, this function must return the size and base address of an eight-byte aligned heap extension block. Syntax extern unsigned __user_heap_extend(int var0, void **base, unsigned requested_size); Usage There is no default implementation of this function. If you define this function, it must have the following characteristics: • The returned size must be either: — A multiple of eight bytes of at least the requested size. — 0, denoting that the request cannot be honored. • The returned base address is aligned on an eight-byte boundary. • Size is measured in bytes. • The function is subject only to ARM Architecture Procedure Call Standard (AAPCS) constraints. • The first argument is always zero on entry and can be ignored. The base is returned in the register holding this argument. Returns This function places a pointer to a block of at least the requested size in *base and returns the size of the block. 0 is returned if no such block can be returned, in which case the value stored at *base is never used. Related concepts 1.11.3 Stack pointer initialization and heap bounds on page 1-64. Related references 4.27 __rt_heap_extend() on page 4-167. 4.31 __rt_stackheap_init() on page 4-171. 4.53 __user_heap_extent() on page 4-194. 4.54 __user_setup_stackheap() on page 4-195. Related information Procedure Call Standard for the ARM Architecture. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-193 4 The C and C++ Library Functions reference 4.53 __user_heap_extent() 4.53 __user_heap_extent() If defined, the __user_heap_extent() function returns the bounds of the memory available to the Heap2 allocator. See rt_misc.h. Syntax extern __value_in_regs struct __heap_extent __user_heap_extent(unsigned ignore1, unsigned ignore2); Usage The parameters ignore1 and ignore2 are the default values for the base address and size of the heap. They are for information only and can be ignored. You only need to implement this function if you are using the Heap2 allocator, which is also part of the C library. This function has no default implementation. The Heap2 allocator calls it during heap initialization to determine the maximum address range that the heap can occupy. The function returns the base address of the heap and the total number of bytes available to the heap, rounded up to the next power of two. For example, if you want to specify that all your heap allocations will come from address 0x80000000 and above, and that the heap has a total maximum size of 3MiB, __user_heap_extent() should return base=0x80000000 and range=0x400000, which is 3MiB rounded up to the next power of two. Related concepts 1.11.3 Stack pointer initialization and heap bounds on page 1-64. Related references 1.11.2 Choosing a heap implementation for memory allocation functions on page 1-63. 4.27 __rt_heap_extend() on page 4-167. 4.31 __rt_stackheap_init() on page 4-171. 4.52 __user_heap_extend() on page 4-193. 4.54 __user_setup_stackheap() on page 4-195. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-194 4 The C and C++ Library Functions reference 4.54 __user_setup_stackheap() 4.54 __user_setup_stackheap() __user_setup_stackheap() sets up and returns the locations of the initial stack and heap. If you define this function, it is called by the C library during program start-up. When __user_setup_stackheap() is called, sp has the same value it had on entry to the application. If this was set to a valid value before calling the C library initialization code, it can be left at this value. If sp is not valid, __user_setup_stackheap() must change this value before using any stack and before returning. __user_setup_stackheap() returns the: • • • Heap base in r0 (if the program uses the heap). Stack base in sp. Heap limit in r2 (if the program uses the heap and uses two-region memory). If this function is re-implemented, it must: • Not corrupt registers other than r0 to r3, ip and sp. • Maintain eight-byte alignment of the heap by ensuring that the heap base is a multiple of eight. To create a version of __user_setup_stackheap() that inherits sp from the execution environment and does not have a heap, set r0 and r2 to zero and return. There is no limit to the size of the stack. However, if the heap region grows into the stack, malloc() attempts to detect the overlapping memory and fails the new memory allocation request. Note Any re-implementation of __user_setup_stackheap() must be in assembler. Related concepts 1.11.3 Stack pointer initialization and heap bounds on page 1-64. 1.11.3 Stack pointer initialization and heap bounds on page 1-64. 1.11.4 Legacy support for __user_initial_stackheap() on page 1-66. Related references 1.6.6 Direct semihosting C library function dependencies on page 1-36. 4.61 Legacy function __user_initial_stackheap() on page 4-204. 4.27 __rt_heap_extend() on page 4-167. 4.31 __rt_stackheap_init() on page 4-171. 4.52 __user_heap_extend() on page 4-193. 4.53 __user_heap_extent() on page 4-194. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-195 4 The C and C++ Library Functions reference 4.55 __vectab_stack_and_reset 4.55 __vectab_stack_and_reset __vectab_stack_and_reset is a library section that provides a way for the initial values of sp and pc to be placed in the vector table, starting at address 0 for M-profile processors, such as Cortex-M1 and Cortex-M3 embedded applications. __vectab_stack_and_reset requires the existence of a main() function in your source code. Without a main() function, if you place the __vectab_stack_and_reset section in a scatter file, an error is generated to the following effect: Error: L6236E: No section matches selector - no section to be FIRST/LAST If the normal start-up code is bypassed, that is, if there is intentionally no main() function, you are responsible for setting up the vector table without __vectab_stack_and_reset. The following segment is part of a scatter file. It includes a minimal vector table illustrating the use of __vectab_stack_and_reset to place the initial sp and pc values at addresses 0x0 and 0x4 in the vector table: ;; Maximum of 256 exceptions (256*4 bytes == 0x400) VECTORS 0x0 0x400 { ; First two entries provided by library ; Remaining entries provided by the user in exceptions.c * (:gdef:__vectab_stack_and_reset, +FIRST) * (exceptions_area) } CODE 0x400 FIXED { * (+RO) } Related concepts 1.11.3 Stack pointer initialization and heap bounds on page 1-64. Related information About scatter-loading. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-196 4 The C and C++ Library Functions reference 4.56 wcscasecmp() 4.56 wcscasecmp() Defined in wchar.h, the wcscasecmp() function performs a case-insensitive string comparison test on wide characters. This function is a GNU extension to the libraries. It is not POSIX-standardized. Syntax int wcscasecmp(const wchar_t * __restrict s1, const wchar_t * __restrict s2); ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-197 4 The C and C++ Library Functions reference 4.57 wcsncasecmp() 4.57 wcsncasecmp() Defined in wchar.h, the wcsncasecmp() function performs a case-insensitive string comparison test of not more than a specified number of wide characters. This function is a GNU extension to the libraries. It is not POSIX-standardized. Syntax int wcsncasecmp(const wchar_t * __restrict s1, const wchar_t * __restrict s2, size_t n); ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-198 4 The C and C++ Library Functions reference 4.58 wcstombs() 4.58 wcstombs() Defined in wchar.h, the wcstombs() function works as described in the ISO C standard, with extended functionality as specified by POSIX. That is, if s is a NULL pointer, wcstombs() returns the length required to convert the entire array regardless of the value of n, but no values are stored. Syntax size_t wcstombs(char *s, const wchar_t *pwcs, size_t n); ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-199 4 The C and C++ Library Functions reference 4.59 Thread-safe C library functions 4.59 Thread-safe C library functions The following table shows the C library functions that are thread-safe. Table 4-1 Functions that are thread-safe Functions Description calloc(), free(), malloc(), realloc() The heap functions are thread-safe if the _mutex_* functions are implemented. All threads share a single heap and use mutexes to avoid data corruption when there is concurrent access. Each heap implementation is responsible for doing its own locking. If you supply your own allocator, it must also do its own locking. This enables it to do finegrained locking if required, rather than protecting the entire heap with a single mutex (coarse-grained locking). alloca() alloca() is thread-safe because it allocates memory on the stack. abort(), raise(), signal(), fenv.h The ARM signal handling functions and floating-point exception traps are thread-safe. clearerr(), fclose(), feof(),ferror(), fflush(), fgetc(),fgetpos(), fgets(), fopen(),fputc(), fputs(), fread(),freopen(), fseek(), fsetpos(),ftell(), fwrite(), getc(),getchar(), gets(), perror(),putc(), putchar(), puts(),rewind(), setbuf(), setvbuf(),tmpfile(), tmpnam(), ungetc() The stdio library is thread-safe if the _mutex_* functions are implemented. The settings for signal handlers and floating-point traps are global across the entire process and are protected by locks. Data corruption does not occur if multiple threads call signal() or an fenv.h function at the same time. However, be aware that the effects of the call act on all threads and not only on the calling thread. Each individual stream is protected by a lock, so two threads can each open their own stdio stream and use it, without interfering with one another. If two threads both want to read or write the same stream, locking at the fgetc() and fputc() level prevents data corruption, but it is possible that the individual characters output by each thread might be interleaved in a confusing way. Note tmpnam() also contains a static buffer but this is only used if the argument is NULL. To ensure that your use of tmpnam() is thread-safe, supply your own buffer space. fprintf(), printf(), vfprintf(), vprintf(), fscanf(), scanf() When using these functions: • The standard C printf() and scanf() functions use stdio so they are thread-safe. • The standard C printf() function is susceptible to changes in the locale settings if called in a multithreaded program. clock() clock() contains static data that is written once at program startup and then only ever read. Therefore, clock() is thread-safe provided no extra threads are already running at the time that the library is initialized. errno errno is thread-safe. Each thread has its own errno stored in a __user_perthread_libspace block. This means that each thread can call errno-setting functions independently and then check errno afterwards without interference from other threads. atexit() The list of exit functions maintained by atexit() is process-global and protected by a lock. In the worst case, if more than one thread calls atexit(), the order that exit functions are called cannot be guaranteed. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-200 4 The C and C++ Library Functions reference 4.59 Thread-safe C library functions Table 4-1 Functions that are thread-safe (continued) Functions Description abs(), acos(), asin(),atan(), These functions are inherently thread-safe. atan2(), atof(),atol(), atoi(), bsearch(),ceil(), cos(), cosh(),difftime(), div(), exp(),fabs(), floor(), fmod(),frexp(), labs(), ldexp(),ldiv(), log(), log10(),memchr(), memcmp(), memcpy(),memmove(), memset(), mktime(),modf(), pow(), qsort(),sin(), sinh(), sqrt(),strcat(), strchr(), strcmp(),strcpy(), strcspn(), strlcat(),strlcpy(), strlen(), strncat(),strncmp(), strncpy(), strpbrk(),strrchr(), strspn(), strstr(),strxfrm(), tan(), tanh() longjmp(), setjmp() Although setjmp() and longjmp() keep data in __user_libspace, they call the __alloca_* functions, that are thread-safe. remove(), rename(), time() These functions use interrupts that communicate with the ARM debugging environments. Typically, you have to reimplement these for a real-world application. snprintf(), sprintf(), vsnprintf(),vsprintf(), sscanf(), isalnum(),isalpha(), iscntrl(), isdigit(),isgraph(), islower(), isprint(),ispunct(), isspace(), isupper(),isxdigit(), tolower(), toupper(),strcoll(), strtod(), strtol(),strtoul(), strftime() When using these functions, the string-based functions read the locale settings. Typically, they are thread-safe. However, if you change locale in mid-session, you must ensure that these functions are not affected. stdin, stdout, stderr These functions are thread-safe. The string-based functions, such as sprintf() and sscanf(), do not depend on the stdio library. Related concepts 1.5.11 Thread safety in the ARM C library on page 1-31. Related references 4.2 alloca() on page 4-141. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-201 4 The C and C++ Library Functions reference 4.60 C library functions that are not thread-safe 4.60 C library functions that are not thread-safe The following table shows the C library functions that are not thread-safe. Table 4-2 Functions that are not thread-safe Functions Description asctime(), localtime(), strtok() These functions are all thread-unsafe. Each contains a static buffer that might be overwritten by another thread between a call to the function and the subsequent use of its return value. ARM supplies reentrant versions, _asctime_r(), _localtime_r(), and _strtok_r(). ARM recommends that you use these functions instead to ensure safety. Note These reentrant versions take additional parameters. _asctime_r() takes an additional parameter that is a pointer to a buffer that the output string is written into. _localtime_r() takes an additional parameter that is a pointer to a struct tm, that the result is written into. _strtok_r() takes an additional parameter that is a pointer to a char pointer to the next token. exit() Do not call exit() in a multithreaded program even if you have provided an implementation of the underlying _sys_exit() that actually terminates all threads. In this case, exit() cleans up before calling _sys_exit() so disrupts other threads. gamma(), lgamma(), lgammaf(), lgammal() s These extended mathlib functions use a global variable, _signgam, so are not thread-safe. mbrlen(), mbsrtowcs(), mbrtowc(), wcrtomb(), wcsrtombs() The C90 multibyte conversion functions (defined in stdlib.h) are not thread-safe, for example mblen() and mbtowc(), because they contain internal static state that is shared between all threads without locking. s However, the extended restartable versions (defined in wchar.h) are thread-safe, for example mbrtowc() and wcrtomb(), provided you pass in a pointer to your own mbstate_t object. You must exclusively use these functions with non-NULL mbstate_t * parameters if you want to ensure thread-safety when handling multibyte strings. If migrating from RVCT, be aware that gamma() is deprecated in ARM Compiler 4.1 and later. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-202 4 The C and C++ Library Functions reference 4.60 C library functions that are not thread-safe Table 4-2 Functions that are not thread-safe (continued) Functions Description rand(), srand() These functions keep internal state that is both global and unprotected. This means that calls to rand() are never thread-safe. ARM recommends that you do one of the following: • • • • Use the reentrant versions _rand_r() and _srand_r() supplied by ARM. These use user-provided buffers instead of static data within the C library. Use your own locking to ensure that only one thread ever calls rand() at a time, for example, by defining $Sub$$rand() if you want to avoid changing your code. Arrange that only one thread ever needs to generate random numbers. Supply your own random number generator that can have multiple independent instances. Note _rand_r() and _srand_r() both take an additional parameter that is a pointer to a buffer storing the state of the random number generator. setlocale(), localeconv() setlocale() is used for setting and reading locale settings. The locale settings are global across all threads, and are not protected by a lock. If two threads call setlocale() to simultaneously modify the locale settings, or if one thread reads the settings while another thread is modifying them, data corruption might occur. Also, many other functions, for example strtod() and sprintf(), read the current locale settings. Therefore, if one thread calls setlocale() concurrently with another thread calling such a function, there might be unexpected results. Multiple threads reading the settings simultaneously is thread-safe in simple cases and if no other thread is simultaneously modifying those settings, but where internally an intermediate buffer is required for more complicated returned results, unexpected results can occur unless you use a reentrant version of setlocale(). ARM recommends that you either: • Choose the locale you want and call setlocale() once to initialize it. Do this before creating any additional threads in your program so that any number of threads can read the locale settings concurrently without interfering with one another. • Use the reentrant version _setlocale_r() supplied by ARM. This returns a string that is either a pointer to a constant string, or a pointer to a string stored in a user-supplied buffer that can be used for thread-local storage, rather than using memory within the C library. The buffer must be at least _SETLOCALE_R_BUFSIZE bytes long, including space for a trailing NUL. Be aware that _setlocale_r() is not fully thread-safe when accessed concurrently to change locale settings. This access is not lock-protected. Also, be aware that localeconv() is not thread-safe. Call the ARM function _get_lconv() with a pointer to a user-supplied buffer instead. Related concepts 1.5.11 Thread safety in the ARM C library on page 1-31. Related references 4.20 _rand_r() on page 4-160. 4.33 _srand_r() on page 4-174. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-203 4 The C and C++ Library Functions reference 4.61 Legacy function __user_initial_stackheap() 4.61 Legacy function __user_initial_stackheap() If you have legacy source code you might see __user_initial_stackheap(), from rt_misc.h. This is an old function that is only supported for backwards compatibility with legacy source code. The modern equivalent is __user_setup_stackheap(). Syntax extern __value_in_regs struct __initial_stackheap __user_initial_stackheap(unsigned R0, unsigned SP, unsigned R2, unsigned SL); Usage __user_initial_stackheap() returns the: • • • Heap base in r0. Stack base in r1, that is, the highest address in the stack region. Heap limit in r2. If this function is reimplemented, it must: • • • Use no more than 88 bytes of stack. Not corrupt registers other than r12 (ip). Maintain eight-byte alignment of the heap. The value of sp (r13) at the time __main() is called is passed as an argument in r1. The default implementation of __user_initial_stackheap(), using the semihosting SYS_HEAPINFO, is given by the library in module sys_stackheap.o. To create a version of __user_initial_stackheap() that inherits sp from the execution environment and does not have a heap, set r0 and r2 to the value of r1 and return. There is no limit to the size of the stack. However, if the heap region grows into the stack, malloc() attempts to detect the overlapping memory and fails the new memory allocation request. The definition of __initial_stackheap in rt_misc.h is: struct __initial_stackheap { unsigned heap_base; /* low-address end of initial heap */ unsigned stack_base; /* high-address end of initial stack */ unsigned heap_limit; /* high-address end of initial heap */ unsigned stack_limit; /* unused */ }; Note The value of stack_base is 0x1 greater than the highest address used by the stack because a fulldescending stack is used. Related concepts 1.11.3 Stack pointer initialization and heap bounds on page 1-64. 1.11.4 Legacy support for __user_initial_stackheap() on page 1-66. Related references 4.54 __user_setup_stackheap() on page 4-195. 1.6.6 Direct semihosting C library function dependencies on page 1-36. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 4-204 Chapter 5 Floating-point Support Functions Reference Describes ARM support for floating-point functions. It contains the following sections: • 5.1 _clearfp() on page 5-206. • 5.2 _controlfp() on page 5-207. • 5.3 __fp_status() on page 5-209. • 5.4 gamma(), gamma_r() on page 5-211. • 5.5 __ieee_status() on page 5-212. • 5.6 j0(), j1(), jn(), Bessel functions of the first kind on page 5-215. • 5.7 significand(), fractional part of a number on page 5-216. • 5.8 _statusfp() on page 5-217. • 5.9 y0(), y1(), yn(), Bessel functions of the second kind on page 5-218. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-205 5 Floating-point Support Functions Reference 5.1 _clearfp() 5.1 _clearfp() Defined in float.h, the _clearfp() function is provided for compatibility with Microsoft products. _clearfp() clears all five exception sticky flags and returns their previous values. You can use the _controlfp() argument macros, for example _EM_INVALID and _EM_ZERODIVIDE, to test bits of the returned result. The function prototype for _clearfp() is: unsigned _clearfp(void); Note This function requires you to select a floating-point model that supports exceptions. For example, --fpmode=ieee_full or --fpmode=ieee_fixed. Related concepts 3.3.1 Floating-point functions for compatibility with Microsoft products on page 3-115. Related references 5.2 _controlfp() on page 5-207. 5.8 _statusfp() on page 5-217. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-206 5 Floating-point Support Functions Reference 5.2 _controlfp() 5.2 _controlfp() Defined in float.h, the _controlfp() function is provided for compatibility with Microsoft products. It enables you to control exception traps and rounding modes. The function prototype for _controlfp() is: unsigned int _controlfp(unsigned int new, unsigned int mask); Note This function requires you to select a floating-point model that supports exceptions. For example, -fpmode=ieee_full or --fpmode=ieee_fixed. _controlfp() also modifies a control word using a mask to isolate the bits to modify. For every bit of mask that is zero, the corresponding control word bit is unchanged. For every bit of mask that is nonzero, the corresponding control word bit is set to the value of the corresponding bit of new. The return value is the previous state of the control word. Note This is different behavior to that of __ieee_status() or __fp_status(), where you can toggle a bit by setting a zero in the mask word and a one in the flags word. The following table describes the macros you can use to form the arguments to _controlfp(). Table 5-1 _controlfp argument macros Macro Description _MCW_EM Mask containing all exception bits _EM_INVALID Bit describing the Invalid Operation exception _EM_ZERODIVIDE Bit describing the Divide by Zero exception _EM_OVERFLOW Bit describing the Overflow exception _EM_UNDERFLOW Bit describing the Underflow exception _EM_INEXACT Bit describing the Inexact Result exception _MCW_RC Mask for the rounding mode field _RC_CHOP Rounding mode value describing Round Toward Zero _RC_UP Rounding mode value describing Round Up _RC_DOWN Rounding mode value describing Round Down _RC_NEAR Rounding mode value describing Round To Nearest Note The values of these macros are not guaranteed to remain the same in future versions of ARM products. To ensure that your code continues to work if the value changes in future releases, use the macro rather than its value. For example, to set the rounding mode to round down, call: _controlfp(_RC_DOWN, _MCW_RC); ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-207 5 Floating-point Support Functions Reference 5.2 _controlfp() To trap the Invalid Operation exception and untrap all other exceptions: _controlfp(_EM_INVALID, _MCW_EM); To untrap the Inexact Result exception: _controlfp(0, _EM_INEXACT); Related concepts 3.3.1 Floating-point functions for compatibility with Microsoft products on page 3-115. Related references 5.1 _clearfp() on page 5-206. 5.8 _statusfp() on page 5-217. 5.5 __ieee_status() on page 5-212. 5.3 __fp_status() on page 5-209. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-208 5 Floating-point Support Functions Reference 5.3 __fp_status() 5.3 __fp_status() The ARM Compiler toolchain supports an interface to the status word in the floating-point environment. Some older versions of the ARM libraries implemented a function called __fp_status() to provide this interface. __fp_status() is the same as __ieee_status() but it uses an older style of status word layout. The compiler still supports the __fp_status() function for backwards compatibility. __fp_status() is defined in stdlib.h. The function prototype for __fp_status() is: unsigned int __fp_status(unsigned int mask, unsigned int flags); Note This function requires you to select a floating-point model that supports exceptions. For example, -fpmode=ieee_full or --fpmode=ieee_fixed. The layout of the status word as seen by __fp_status() is as follows: 31 24 System ID 23 21 R 20 16 Masks 15 13 R 12 8 FPA only 7 5 R 4 0 Sticky Figure 5-1 Floating-point status word layout The fields in the status word are as follows: • Bits 0 to 4 (values 0x1 to 0x10, respectively) are the sticky flags, or cumulative flags, for each exception. The sticky flag for an exception is set to 1 whenever that exception happens and is not trapped. Sticky flags are never cleared by the system, only by the user. The mapping of exceptions to bits is: — Bit 0 (0x01) is for the Invalid Operation exception — Bit 1 (0x02) is for the Divide by Zero exception. — Bit 2 (0x04) is for the Overflow exception. — Bit 3 (0x08) is for the Underflow exception. — Bit 4 (0x10) is for the Inexact Result exception. • Bits 8 to 12 (values 0x100 to 0x1000) control various aspects of the Floating-Point Architecture (FPA). The FPA is obsolete and the ARM compilation tools do not support it. Any attempt to write to these bits is ignored. • Bits 16 to 20 (values 0x10000 to 0x100000) are the exception masks. These control whether each exception is trapped or not. If a bit is set to 1, the corresponding exception is trapped. If a bit is set to 0, the corresponding exception sets its sticky flag and returns a plausible result. • Bits 24 to 31 contain the system ID that cannot be changed. It is set to 0x40 for software floatingpoint, to 0x80 or above for hardware floating-point, and to 0 or 1 if a hardware floating-point environment is being faked by an emulator. • Bits marked R are reserved. They cannot be written to by the __fp_status() call, and you must ignore anything you find in them. The rounding mode cannot be changed with the __fp_status() call. In addition to defining the __fp_status() call itself, stdlib.h also defines the following constants to be used for the arguments: #define #define #define #define #define #define #define ARM DUI0378G_02 __fpsr_IXE __fpsr_UFE __fpsr_OFE __fpsr_DZE __fpsr_IOE __fpsr_IXC __fpsr_UFC 0x100000 0x80000 0x40000 0x20000 0x10000 0x10 0x8 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-209 5 Floating-point Support Functions Reference 5.3 __fp_status() #define __fpsr_OFC #define __fpsr_DZC #define __fpsr_IOC 0x4 0x2 0x1 For example, to trap the Invalid Operation exception and untrap all other exceptions, you would call __fp_status() with the following input parameters: __fp_status(_fpsr_IXE | _fpsr_UFE | _fpsr_OFE | _fpsr_DZE | _fpsr_IOE, _fpsr_IOE); To untrap the Inexact Result exception: __fp_status(_fpsr_IXE, 0); To clear the Underflow sticky flag: __fp_status(_fpsr_UFC, 0); Related concepts 3.3 Controlling the ARM floating-point environment on page 3-115. Related references 5.5 __ieee_status() on page 5-212. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-210 5 Floating-point Support Functions Reference 5.4 gamma(), gamma_r() 5.4 gamma(), gamma_r() The gamma() and gamma_r() functions both compute the logarithm of the gamma function. They are synonyms for lgamma and lgamma_r. double gamma(double x); double gamma_r(double x, int *); Note Despite their names, these functions compute the logarithm of the gamma function, not the gamma function itself. To compute the gamma function itself, use tgamma(). Note These functions are deprecated in ARM Compiler 4.1 and later. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-211 5 Floating-point Support Functions Reference 5.5 __ieee_status() 5.5 __ieee_status() The ARM Compiler toolchain supports an interface to the status word in the floating-point environment. This interface is provided as function __ieee_status() and it is generally the most efficient function to use for modifying the status word for VFP. __ieee_status() is defined in fenv.h. The function prototype for __ieee_status() is: unsigned int __ieee_status(unsigned int mask, unsigned int flags); Note This function requires you to select a floating-point model that supports exceptions. For example, -fpmode=ieee_full or --fpmode=ieee_fixed. __ieee_status() modifies the writable parts of the status word according to the parameters, and returns the previous value of the whole word. The writable bits are modified by setting them to: new = (old & ~mask) ^ flags; Four different operations can be performed on each bit of the status word, depending on the corresponding bits in mask and flags. Table 5-2 Status word bit modification Bit of mask Bit of flags Effect 0 0 Leave alone 0 1 Toggle 1 0 Set to 0 1 1 Set to 1 The layout of the status word as seen by __ieee_status() is as follows: 31 28 27 26 25 24 23 22 R QC R FZ RM 21 20 19 18 VFP R 16 15 VFP 13 12 R Masks 0 5 4 8 7 R S ticky Figure 5-2 IEEE status word layout The fields in the status word are as follows: • Bits 0 to 4 (values 0x1 to 0x10, respectively) are the sticky flags, or cumulative flags, for each exception. The sticky flag for an exception is set to 1 whenever that exception happens and is not trapped. Sticky flags are never cleared by the system, only by the user. The mapping of exceptions to bits is: — Bit 0 (0x01) is for the Invalid Operation exception — Bit 1 (0x02) is for the Divide by Zero exception. — Bit 2 (0x04) is for the Overflow exception. — Bit 3 (0x08) is for the Underflow exception. — Bit 4 (0x10) is for the Inexact Result exception. • Bits 8 to 12 (values 0x100 to 0x1000) are the exception masks. These control whether each exception is trapped or not. If a bit is set to 1, the corresponding exception is trapped. If a bit is set to 0, the corresponding exception sets its sticky flag and returns a plausible result. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-212 5 Floating-point Support Functions Reference 5.5 __ieee_status() • • Bits 16 to 18, and bits 20 and 21, are used by VFP hardware to control the VFP vector capability. The __ieee_status() call does not let you modify these bits. Bits 22 and 23 control the rounding mode. See the following table. Table 5-3 Rounding mode control Bits Rounding mode 00 Round to nearest 01 Round up 10 Round down 11 Round toward zero Note The fz*, fj* and f* library variants support only the round-to-nearest rounding mode. If you require support for the other rounding modes, you must use the full IEEE g* libraries. (The relevant compiler options are --fpmode=std, --fpmode=ieee_no_fenv and --fpmode=ieee_fixed.) • Bit 24 enables FZ (Flush to Zero) mode if it is set. In FZ mode, denormals are forced to zero to speed up processing because denormals can be difficult to work with and slow down floating-point systems. Setting this bit reduces accuracy but might increase speed. Note — The FZ bit in the IEEE status word is not supported by any of the fplib variants. This means that switching between flushing to zero and not flushing to zero is not possible with any variant of fplib at runtime. However, flushing to zero or not flushing to zero can be set at compile time as a result of the library you choose to build with. — Some functions are not provided in hardware. They exist only in the software floating-point libraries. So these functions cannot support the FZ mode, even when you are compiling for a hardware VFP architecture. As a result, behavior of the floating-point libraries is not consistent across all functions when you change the FZ mode dynamically. • • Bit 27 indicates that saturation has occurred in an advanced SIMD saturating integer operation. This is accessible through the __ieee_status() call. Bits marked R are reserved. They cannot be written to by the __ieee_status() call, and you must ignore anything you find in them. In addition to defining the __ieee_status() call itself, fenv.h also defines the following constants to be used for the arguments: #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define #define ARM DUI0378G_02 FE_IEEE_FLUSHZERO FE_IEEE_ROUND_TONEAREST FE_IEEE_ROUND_UPWARD FE_IEEE_ROUND_DOWNWARD FE_IEEE_ROUND_TOWARDZERO FE_IEEE_ROUND_MASK FE_IEEE_MASK_INVALID FE_IEEE_MASK_DIVBYZERO FE_IEEE_MASK_OVERFLOW FE_IEEE_MASK_UNDERFLOW FE_IEEE_MASK_INEXACT FE_IEEE_MASK_ALL_EXCEPT FE_IEEE_INVALID FE_IEEE_DIVBYZERO FE_IEEE_OVERFLOW FE_IEEE_UNDERFLOW FE_IEEE_INEXACT FE_IEEE_ALL_EXCEPT (0x01000000) (0x00000000) (0x00400000) (0x00800000) (0x00C00000) (0x00C00000) (0x00000100) (0x00000200) (0x00000400) (0x00000800) (0x00001000) (0x00001F00) (0x00000001) (0x00000002) (0x00000004) (0x00000008) (0x00000010) (0x0000001F) Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-213 5 Floating-point Support Functions Reference 5.5 __ieee_status() For example, to set the rounding mode to round down, you would call: __ieee_status(FE_IEEE_ROUND_MASK, FE_IEEE_ROUND_DOWNWARD); To trap the Invalid Operation exception and untrap all other exceptions: __ieee_status(FE_IEEE_MASK_ALL_EXCEPT, FE_IEEE_MASK_INVALID); To untrap the Inexact Result exception: __ieee_status(FE_IEEE_MASK_INEXACT, 0); To clear the Underflow sticky flag: __ieee_status(FE_IEEE_UNDERFLOW, 0); Related concepts 3.3 Controlling the ARM floating-point environment on page 3-115. 3.6.7 Exceptions arising from IEEE 754 floating-point arithmetic on page 3-133. Related references 3.3.8 ARM floating-point compiler extensions to the C99 interface on page 3-119. 5.3 __fp_status() on page 5-209. 1.26 C and C++ library naming conventions on page 1-91. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-214 5 Floating-point Support Functions Reference 5.6 j0(), j1(), jn(), Bessel functions of the first kind 5.6 j0(), j1(), jn(), Bessel functions of the first kind These functions compute Bessel functions of the first kind. j0 and j1 compute the functions of order 0 and 1 respectively. jn computes the function of order n. double j0(double x); double j1(double x); double jn(int n, double x); If the absolute value of x exceeds pi times 2^52, these functions return an ERANGE error, denoting total loss of significance in the result. Note These functions are deprecated in ARM Compiler 4.1 and later. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-215 5 Floating-point Support Functions Reference 5.7 significand(), fractional part of a number 5.7 significand(), fractional part of a number The significand() function returns the fraction part of x, as a number between 1.0 and 2.0 (not including 2.0). double significand(double x); Note This functions is deprecated in ARM Compiler 4.1 and later. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-216 5 Floating-point Support Functions Reference 5.8 _statusfp() 5.8 _statusfp() Defined in float.h, the _statusfp() function is provided for compatibility with Microsoft products. It returns the current value of the exception sticky flags. You can use the _controlfp() argument macros, for example _EM_INVALID and _EM_ZERODIVIDE, to test bits of the returned result. The function prototype for _statusfp() is: unsigned _statusfp(void); Note This function requires you to select a floating-point model that supports exceptions. For example, --fpmode=ieee_full or --fpmode=ieee_fixed. Related concepts 3.3.1 Floating-point functions for compatibility with Microsoft products on page 3-115. Related references 5.1 _clearfp() on page 5-206. 5.2 _controlfp() on page 5-207. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-217 5 Floating-point Support Functions Reference 5.9 y0(), y1(), yn(), Bessel functions of the second kind 5.9 y0(), y1(), yn(), Bessel functions of the second kind These functions compute Bessel functions of the second kind. y0 and y1 compute the functions of order 0 and 1 respectively. yn computes the function of order n. double y0(double x); double y1(double x); double yn(int, double); If x is positive and exceeds pi times 2^52, these functions return an ERANGE error, denoting total loss of significance in the result. Note These functions are deprecated in ARM Compiler 4.1 and later. ARM DUI0378G_02 Copyright © 2007, 2008, 2011, 2012, 2014, 2015 ARM. All rights reserved. Confidential - Draft - Beta 5-218
Source Exif Data:
File Type : PDF File Type Extension : pdf MIME Type : application/pdf PDF Version : 1.7 Linearized : No Encryption : Standard V4.4 (128-bit) User Access : Print, Copy, Annotate, Fill forms, Extract, Print high-res Author : ARM Create Date : 2015:08:05 15:36:50+00:00 Modify Date : 2015:08:05 15:36:50+00:00 Creator : AH XSL Formatter V6.2 MR4 for Linux64 : 6.2.6.18551 (2014/09/24 15:00JST) Producer : Antenna House PDF Output Library 6.2.609 (Linux64) Keywords : ARM Compiler, Related to Keil Boards Title : ARM® Compiler v5.06 for µVision® ARM C and C++ Libraries and Floating-Point Support User Guide Subject : ARM® Compiler for µVision® ARM C and C++ Libraries and Floating-Point Support User Guide. This manual provides user information for the ARM libraries and floating-point support. It is also available as a PDF. Trapped : False Page Count : 218 Page Mode : UseOutlines Language : ENEXIF Metadata provided by EXIF.tools