@database gcc.guide @Master /ade-src/fsf/gcc/./gcc.texi @Width 72 This is the AmigaGuideŽ file gcc.guide, produced by Makeinfo-1.67 from the input file /ade-src/fsf/gcc/./gcc.texi. This file documents the use and the internals of the GNU compiler. Published by the Free Software Foundation 59 Temple Place - Suite 330 Boston, MA 02111-1307 USA Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995 Free Software Foundation, Inc. Permission is granted to make and distribute verbatim copies of this manual provided the copyright notice and this permission notice are preserved on all copies. Permission is granted to copy and distribute modified versions of this manual under the conditions for verbatim copying, provided also that the sections entitled "GNU General Public License," "Funding for Free Software," and "Protect Your Freedom--Fight `Look And Feel'" are included exactly as in the original, and provided that the entire resulting derived work is distributed under the terms of a permission notice identical to this one. Permission is granted to copy and distribute translations of this manual into another language, under the above conditions for modified versions, except that the sections entitled "GNU General Public License," "Funding for Free Software," and "Protect Your Freedom--Fight `Look And Feel'", and this permission notice, may be included in translations approved by the Free Software Foundation instead of in the original English. @Node Main "gcc.guide" @Next "Copying" Introduction ************ This manual documents how to run, install and port the GNU compiler, as well as its new features and incompatibilities, and how to report bugs. It corresponds to GNU CC version 2.7.2. @{" Copying " Link "Copying"} GNU General Public License says how you can copy and share GNU CC. @{" Contributors " Link "Contributors"} People who have contributed to GNU CC. @{" Funding " Link "Funding"} How to help assure funding for free software. @{" Look and Feel " Link "Look and Feel"} Protect your freedom--fight "look and feel". @{" G++ and GCC " Link "G++ and GCC"} You can compile C or C++ programs. @{" Invoking GCC " Link "Invoking GCC"} Command options supported by @{b}gcc@{ub}. @{" Installation " Link "Installation"} How to configure, compile and install GNU CC. @{" C Extensions " Link "C Extensions"} GNU extensions to the C language family. @{" C++ Extensions " Link "C++ Extensions"} GNU extensions to the C++ language. @{" Trouble " Link "Trouble"} If you have trouble installing GNU CC. @{" Bugs " Link "Bugs"} How, why and where to report bugs. @{" Service " Link "Service"} How to find suppliers of support for GNU CC. @{" VMS " Link "VMS"} Using GNU CC on VMS. @{" Portability " Link "Portability"} Goals of GNU CC's portability features. @{" Interface " Link "Interface"} Function-call interface of GNU CC output. @{" Passes " Link "Passes"} Order of passes, what they do, and what each file is for. @{" RTL " Link "RTL"} The intermediate representation that most passes work on. @{" Machine Desc " Link "Machine Desc"} How to write machine description instruction patterns. @{" Target Macros " Link "Target Macros"} How to write the machine description C macros. @{" Config " Link "Config"} Writing the @{b}xm-@{I}MACHINE@{ui}.h@{ub} file. @{" Fragments " Link "Fragments"} Writing the @{b}t-@{I}TARGET@{ui}@{ub} and @{b}x-@{I}HOST@{ui}@{ub} files. @{" Index " Link "Index"} Index of concepts and symbol names. @EndNode @Node "Copying" "gcc.guide/Copying" @Next "Contributors" @Prev "Main" @Toc "Main" GNU GENERAL PUBLIC LICENSE ************************** Version 2, June 1991 Copyright (C) 1989, 1991 Free Software Foundation, Inc. 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed. Preamble ======== The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundation's software and to any other program whose authors commit to using it. (Some other Free Software Foundation software is covered by the GNU Library General Public License instead.) You can apply it to your programs, too. When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things. To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it. For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights. We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software. Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations. Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all. The precise terms and conditions for copying, distribution and modification follow. TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION 0. This License applies to any program or other work which contains a notice placed by the copyright holder saying it may be distributed under the terms of this General Public License. The "Program", below, refers to any such program or work, and a "work based on the Program" means either the Program or any derivative work under copyright law: that is to say, a work containing the Program or a portion of it, either verbatim or with modifications and/or translated into another language. (Hereinafter, translation is included without limitation in the term "modification".) Each licensee is addressed as "you". Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does. 1. You may copy and distribute verbatim copies of the Program's source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Program a copy of this License along with the Program. You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee. 2. You may modify your copy or copies of the Program or any portion of it, thus forming a work based on the Program, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions: a. You must cause the modified files to carry prominent notices stating that you changed the files and the date of any change. b. You must cause any work that you distribute or publish, that in whole or in part contains or is derived from the Program or any part thereof, to be licensed as a whole at no charge to all third parties under the terms of this License. c. If the modified program normally reads commands interactively when run, you must cause it, when started running for such interactive use in the most ordinary way, to print or display an announcement including an appropriate copyright notice and a notice that there is no warranty (or else, saying that you provide a warranty) and that users may redistribute the program under these conditions, and telling the user how to view a copy of this License. (Exception: if the Program itself is interactive but does not normally print such an announcement, your work based on the Program is not required to print an announcement.) These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Program. In addition, mere aggregation of another work not based on the Program with the Program (or with a work based on the Program) on a volume of a storage or distribution medium does not bring the other work under the scope of this License. 3. You may copy and distribute the Program (or a work based on it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you also do one of the following: a. Accompany it with the complete corresponding machine-readable source code, which must be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, b. Accompany it with a written offer, valid for at least three years, to give any third party, for a charge no more than your cost of physically performing source distribution, a complete machine-readable copy of the corresponding source code, to be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, c. Accompany it with the information you received as to the offer to distribute corresponding source code. (This alternative is allowed only for noncommercial distribution and only if you received the program in object code or executable form with such an offer, in accord with Subsection b above.) The source code for a work means the preferred form of the work for making modifications to it. For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable. If distribution of executable or object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place counts as distribution of the source code, even though third parties are not compelled to copy the source along with the object code. 4. You may not copy, modify, sublicense, or distribute the Program except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense or distribute the Program is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance. 5. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Program or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Program (or any work based on the Program), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Program or works based on it. 6. Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties to this License. 7. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program. If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply and the section as a whole is intended to apply in other circumstances. It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system, which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice. This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License. 8. If the distribution and/or use of the Program is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Program under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License. 9. The Free Software Foundation may publish revised and/or new versions of the General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. Each version is given a distinguishing version number. If the Program specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation. 10. If you wish to incorporate parts of the Program into other free programs whose distribution conditions are different, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally. NO WARRANTY 11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. END OF TERMS AND CONDITIONS How to Apply These Terms to Your New Programs ============================================= If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms. To do so, attach the following notices to the program. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the "copyright" line and a pointer to where the full notice is found. @{I}ONE LINE TO GIVE THE PROGRAM'S NAME AND A BRIEF IDEA OF WHAT IT DOES.@{ui} Copyright (C) 19@{I}YY@{ui} @{I}NAME OF AUTHOR@{ui} This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. Also add information on how to contact you by electronic and paper mail. If the program is interactive, make it output a short notice like this when it starts in an interactive mode: Gnomovision version 69, Copyright (C) 19@{I}YY@{ui} @{I}NAME OF AUTHOR@{ui} Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. This is free software, and you are welcome to redistribute it under certain conditions; type `show c' for details. The hypothetical commands @{b}show w@{ub} and @{b}show c@{ub} should show the appropriate parts of the General Public License. Of course, the commands you use may be called something other than @{b}show w@{ub} and @{b}show c@{ub}; they could even be mouse-clicks or menu items--whatever suits your program. You should also get your employer (if you work as a programmer) or your school, if any, to sign a "copyright disclaimer" for the program, if necessary. Here is a sample; alter the names: Yoyodyne, Inc., hereby disclaims all copyright interest in the program `Gnomovision' (which makes passes at compilers) written by James Hacker. @{I}SIGNATURE OF TY COON@{ui}, 1 April 1989 Ty Coon, President of Vice This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Library General Public License instead of this License. @EndNode @Node "Contributors" "gcc.guide/Contributors" @Next "Funding" @Prev "Copying" @Toc "Main" Contributors to GNU CC ********************** In addition to Richard Stallman, several people have written parts of GNU CC. @{b}*@{ub} The idea of using RTL and some of the optimization ideas came from the program PO written at the University of Arizona by Jack Davidson and Christopher Fraser. See "Register Allocation and Exhaustive Peephole Optimization", Software Practice and Experience 14 (9), Sept. 1984, 857-866. @{b}*@{ub} Paul Rubin wrote most of the preprocessor. @{b}*@{ub} Leonard Tower wrote parts of the parser, RTL generator, and RTL definitions, and of the Vax machine description. @{b}*@{ub} Ted Lemon wrote parts of the RTL reader and printer. @{b}*@{ub} Jim Wilson implemented loop strength reduction and some other loop optimizations. @{b}*@{ub} Nobuyuki Hikichi of Software Research Associates, Tokyo, contributed the support for the Sony NEWS machine. @{b}*@{ub} Charles LaBrec contributed the support for the Integrated Solutions 68020 system. @{b}*@{ub} Michael Tiemann of Cygnus Support wrote the front end for C++, as well as the support for inline functions and instruction scheduling. Also the descriptions of the National Semiconductor 32000 series cpu, the SPARC cpu and part of the Motorola 88000 cpu. @{b}*@{ub} Gerald Baumgartner added the signature extension to the C++ front-end. @{b}*@{ub} Jan Stein of the Chalmers Computer Society provided support for Genix, as well as part of the 32000 machine description. @{b}*@{ub} Randy Smith finished the Sun FPA support. @{b}*@{ub} Robert Brown implemented the support for Encore 32000 systems. @{b}*@{ub} David Kashtan of SRI adapted GNU CC to VMS. @{b}*@{ub} Alex Crain provided changes for the 3b1. @{b}*@{ub} Greg Satz and Chris Hanson assisted in making GNU CC work on HP-UX for the 9000 series 300. @{b}*@{ub} William Schelter did most of the work on the Intel 80386 support. @{b}*@{ub} Christopher Smith did the port for Convex machines. @{b}*@{ub} Paul Petersen wrote the machine description for the Alliant FX/8. @{b}*@{ub} Dario Dariol contributed the four varieties of sample programs that print a copy of their source. @{b}*@{ub} Alain Lichnewsky ported GNU CC to the Mips cpu. @{b}*@{ub} Devon Bowen, Dale Wiles and Kevin Zachmann ported GNU CC to the Tahoe. @{b}*@{ub} Jonathan Stone wrote the machine description for the Pyramid computer. @{b}*@{ub} Gary Miller ported GNU CC to Charles River Data Systems machines. @{b}*@{ub} Richard Kenner of the New York University Ultracomputer Research Laboratory wrote the machine descriptions for the AMD 29000, the DEC Alpha, the IBM RT PC, and the IBM RS/6000 as well as the support for instruction attributes. He also made changes to better support RISC processors including changes to common subexpression elimination, strength reduction, function calling sequence handling, and condition code support, in addition to generalizing the code for frame pointer elimination. @{b}*@{ub} Richard Kenner and Michael Tiemann jointly developed reorg.c, the delay slot scheduler. @{b}*@{ub} Mike Meissner and Tom Wood of Data General finished the port to the Motorola 88000. @{b}*@{ub} Masanobu Yuhara of Fujitsu Laboratories implemented the machine description for the Tron architecture (specifically, the Gmicro). @{b}*@{ub} NeXT, Inc. donated the front end that supports the Objective C language. @{b}*@{ub} James van Artsdalen wrote the code that makes efficient use of the Intel 80387 register stack. @{b}*@{ub} Mike Meissner at the Open Software Foundation finished the port to the MIPS cpu, including adding ECOFF debug support, and worked on the Intel port for the Intel 80386 cpu. @{b}*@{ub} Ron Guilmette implemented the @{b}protoize@{ub} and @{b}unprotoize@{ub} tools, the support for Dwarf symbolic debugging information, and much of the support for System V Release 4. He has also worked heavily on the Intel 386 and 860 support. @{b}*@{ub} Torbjorn Granlund implemented multiply- and divide-by-constant optimization, improved long long support, and improved leaf function register allocation. @{b}*@{ub} Mike Stump implemented the support for Elxsi 64 bit CPU. @{b}*@{ub} John Wehle added the machine description for the Western Electric 32000 processor used in several 3b series machines (no relation to the National Semiconductor 32000 processor). @{b}*@{ub} Holger Teutsch provided the support for the Clipper cpu. @{b}*@{ub} Kresten Krab Thorup wrote the run time support for the Objective C language. @{b}*@{ub} Stephen Moshier contributed the floating point emulator that assists in cross-compilation and permits support for floating point numbers wider than 64 bits. @{b}*@{ub} David Edelsohn contributed the changes to RS/6000 port to make it support the PowerPC and POWER2 architectures. @{b}*@{ub} Steve Chamberlain wrote the support for the Hitachi SH processor. @{b}*@{ub} Peter Schauer wrote the code to allow debugging to work on the Alpha. @{b}*@{ub} Oliver M. Kellogg of Deutsche Aerospace contributed the port to the MIL-STD-1750A. @{b}*@{ub} Michael K. Gschwind contributed the port to the PDP-11. @EndNode @Node "Funding" "gcc.guide/Funding" @Next "Look and Feel" @Prev "Contributors" @Toc "Main" Funding Free Software ********************* If you want to have more free software a few years from now, it makes sense for you to help encourage people to contribute funds for its development. The most effective approach known is to encourage commercial redistributors to donate. Users of free software systems can boost the pace of development by encouraging for-a-fee distributors to donate part of their selling price to free software developers--the Free Software Foundation, and others. The way to convince distributors to do this is to demand it and expect it from them. So when you compare distributors, judge them partly by how much they give to free software development. Show distributors they must compete to be the one who gives the most. To make this approach work, you must insist on numbers that you can compare, such as, "We will donate ten dollars to the Frobnitz project for each disk sold." Don't be satisfied with a vague promise, such as "A portion of the profits are donated," since it doesn't give a basis for comparison. Even a precise fraction "of the profits from this disk" is not very meaningful, since creative accounting and unrelated business decisions can greatly alter what fraction of the sales price counts as profit. If the price you pay is $50, ten percent of the profit is probably less than a dollar; it might be a few cents, or nothing at all. Some redistributors do development work themselves. This is useful too; but to keep everyone honest, you need to inquire how much they do, and what kind. Some kinds of development make much more long-term difference than others. For example, maintaining a separate version of a program contributes very little; maintaining the standard version of a program for the whole community contributes much. Easy new ports contribute little, since someone else would surely do them; difficult ports such as adding a new CPU to the GNU C compiler contribute more; major new features or packages contribute the most. By establishing the idea that supporting further development is "the proper thing to do" when distributing free software for a fee, we can assure a steady flow of resources into making more free software. Copyright (C) 1994 Free Software Foundation, Inc. Verbatim copying and redistribution of this section is permitted without royalty; alteration is not permitted. @EndNode @Node "Look and Feel" "gcc.guide/Look and Feel" @Next "G++ and GCC" @Prev "Funding" @Toc "Main" Protect Your Freedom--Fight "Look And Feel" ******************************************* @{i}This section is a political message from the League for Programming Freedom to the users of GNU CC. We have included it here because the issue of interface copyright is important to the GNU project.@{ui} Apple, Lotus, and now CDC have tried to create a new form of legal monopoly: a copyright on a user interface. An interface is a kind of language--a set of conventions for communication between two entities, human or machine. Until a few years ago, the law seemed clear: interfaces were outside the domain of copyright, so programmers could program freely and implement whatever interface the users demanded. Imitating de-facto standard interfaces, sometimes with improvements, was standard practice in the computer field. These improvements, if accepted by the users, caught on and became the norm; in this way, much progress took place. Computer users, and most software developers, were happy with this state of affairs. However, large companies such as Apple and Lotus would prefer a different system--one in which they can own interfaces and thereby rid themselves of all serious competitors. They hope that interface copyright will give them, in effect, monopolies on major classes of software. Other large companies such as IBM and Digital also favor interface monopolies, for the same reason: if languages become property, they expect to own many de-facto standard languages. But Apple and Lotus are the ones who have actually sued. Apple's lawsuit was defeated, for reasons only partly related to the general issue of interface copyright. Lotus won lawsuits against two small companies, which were thus put out of business. Then Lotus sued Borland; Lotus won in the trial court (no surprise, since it was the same court that had ruled for Lotus twice before), but the court of appeals ruled in favor of Borland, which was assisted by a friend-of-the-court brief from the League for Programming Freedom. Lotus appealed the case to the Supreme Court, which heard the case but was unable to reach a decision. This failure means that the appeals court decision stands, in one portion of the United States, and may influence the other appeals courts, but it does not set a nationwide precedent. The battle is not over, and it is not limited to the United States. The battle is extending into other areas of software as well. In 1995 a company that produced a simulator for a CDC computer was shut down by a copyright lawsuit, in which CDC charged that the simulator infringed the copyright on the manuals for the computer. If the monopolists get their way, they will hobble the software field: @{b}*@{ub} Gratuitous incompatibilities will burden users. Imagine if each car manufacturer had to design a different way to start, stop, and steer a car. @{b}*@{ub} Users will be "locked in" to whichever interface they learn; then they will be prisoners of one supplier, who will charge a monopolistic price. @{b}*@{ub} Large companies have an unfair advantage wherever lawsuits become commonplace. Since they can afford to sue, they can intimidate smaller developers with threats even when they don't really have a case. @{b}*@{ub} Interface improvements will come slower, since incremental evolution through creative partial imitation will no longer occur. If interface monopolies are accepted, other large companies are waiting to grab theirs: @{b}*@{ub} Adobe is expected to claim a monopoly on the interfaces of various popular application programs, if Lotus ultimately wins the case against Borland. @{b}*@{ub} Open Computing magazine reported a Microsoft vice president as threatening to sue people who imitate the interface of Windows. Users invest a great deal of time and money in learning to use computer interfaces. Far more, in fact, than software developers invest in developing @{i}and even implementing@{ui} the interfaces. Whoever can own an interface, has made its users into captives, and misappropriated their investment. To protect our freedom from monopolies like these, a group of programmers and users have formed a grass-roots political organization, the League for Programming Freedom. The purpose of the League is to oppose monopolistic practices such as interface copyright and software patents. The League calls for a return to the legal policies of the recent past, in which programmers could program freely. The League is not concerned with free software as an issue, and is not affiliated with the Free Software Foundation. The League's activities include publicizing the issues, as is being done here, and filing friend-of-the-court briefs on behalf of defendants sued by monopolists. The League's membership rolls include Donald Knuth, the foremost authority on algorithms, John McCarthy, inventor of Lisp, Marvin Minsky, founder of the MIT Artificial Intelligence lab, Guy L. Steele, Jr., author of well-known books on Lisp and C, as well as Richard Stallman, the developer of GNU CC. Please join and add your name to the list. Membership dues in the League are $42 per year for programmers, managers and professionals; $10.50 for students; $21 for others. Activist members are especially important, but members who have no time to give are also important. Surveys at major ACM conferences have indicated a vast majority of attendees agree with the League on both issues (interface copyrights and software patents). If just ten percent of the programmers who agree with the League join the League, we will probably triumph. To join, or for more information, phone (617) 243-4091 or write to: League for Programming Freedom 1 Kendall Square #143 P.O. Box 9171 Cambridge, MA 02139 You can also send electronic mail to @{b}lpf@uunet.uu.net@{ub}. In addition to joining the League, here are some suggestions from the League for other things you can do to protect your freedom to write programs: @{b}*@{ub} Tell your friends and colleagues about this issue and how it threatens to ruin the computer industry. @{b}*@{ub} Mention that you are a League member in your @{b}.signature@{ub}, and mention the League's email address for inquiries. @{b}*@{ub} Ask the companies you consider working for or working with to make statements against software monopolies, and give preference to those that do. @{b}*@{ub} When employers ask you to sign contracts giving them copyright on your work, insist on a clause saying they will not claim the copyright covers imitating the interface. @{b}*@{ub} When employers ask you to sign contracts giving them patent rights, insist on clauses saying they can use these rights only defensively. Don't rely on "company policy," since that can change at any time; don't rely on an individual executive's private word, since that person may be replaced. Get a commitment just as binding as the commitment they get from you. @{b}*@{ub} Write to Congress to explain the importance of these issues. House Subcommittee on Intellectual Property 2137 Rayburn Bldg Washington, DC 20515 Senate Subcommittee on Patents, Trademarks and Copyrights United States Senate Washington, DC 20510 (These committees have received lots of mail already; let's give them even more.) Democracy means nothing if you don't use it. Stand up and be counted! @EndNode @Node "G++ and GCC" "gcc.guide/G++ and GCC" @Next "Invoking GCC" @Prev "Look and Feel" @Toc "Main" Compile C, C++, or Objective C ****************************** The C, C++, and Objective C versions of the compiler are integrated; the GNU C compiler can compile programs written in C, C++, or Objective C. "GCC" is a common shorthand term for the GNU C compiler. This is both the most general name for the compiler, and the name used when the emphasis is on compiling C programs. When referring to C++ compilation, it is usual to call the compiler "G++". Since there is only one compiler, it is also accurate to call it "GCC" no matter what the language context; however, the term "G++" is more useful when the emphasis is on compiling C++ programs. We use the name "GNU CC" to refer to the compilation system as a whole, and more specifically to the language-independent part of the compiler. For example, we refer to the optimization options as affecting the behavior of "GNU CC" or sometimes just "the compiler". Front ends for other languages, such as Ada 9X, Fortran, Modula-3, and Pascal, are under development. These front-ends, like that for C++, are built in subdirectories of GNU CC and link to it. The result is an integrated compiler that can compile programs written in C, C++, Objective C, or any of the languages for which you have installed front ends. In this manual, we only discuss the options for the C, Objective-C, and C++ compilers and those of the GNU CC core. Consult the documentation of the other front ends for the options to use when compiling programs written in other languages. G++ is a @{i}compiler@{ui}, not merely a preprocessor. G++ builds object code directly from your C++ program source. There is no intermediate C version of the program. (By contrast, for example, some other implementations use a program that generates a C program from your C++ source.) Avoiding an intermediate C representation of the program means that you get better object code, and better debugging information. The GNU debugger, GDB, works with this information in the object code to give you comprehensive C++ source-level editing capabilities (see @{"C and C++" Link "gdb.guide/C"}). @EndNode @Node "Invoking GCC" "gcc.guide/Invoking GCC" @Next "Installation" @Prev "G++ and GCC" @Toc "Main" GNU CC Command Options ********************** When you invoke GNU CC, it normally does preprocessing, compilation, assembly and linking. The "overall options" allow you to stop this process at an intermediate stage. For example, the @{b}-c@{ub} option says not to run the linker. Then the output consists of object files output by the assembler. Other options are passed on to one stage of processing. Some options control the preprocessor and others the compiler itself. Yet other options control the assembler and linker; most of these are not documented here, since you rarely need to use any of them. Most of the command line options that you can use with GNU CC are useful for C programs; when an option is only useful with another language (usually C++), the explanation says so explicitly. If the description for a particular option does not mention a source language, you can use that option with all supported languages. See @{"Compiling C++ Programs" Link "Invoking G++"}, for a summary of special options for compiling C++ programs. The @{b}gcc@{ub} program accepts options and file names as operands. Many options have multiletter names; therefore multiple single-letter options may @{i}not@{ui} be grouped: @{b}-dr@{ub} is very different from @{b}-d -r@{ub}. You can mix options and other arguments. For the most part, the order you use doesn't matter. Order does matter when you use several options of the same kind; for example, if you specify @{b}-L@{ub} more than once, the directories are searched in the order specified. Many options have long names starting with @{b}-f@{ub} or with @{b}-W@{ub}--for example, @{b}-fforce-mem@{ub}, @{b}-fstrength-reduce@{ub}, @{b}-Wformat@{ub} and so on. Most of these have both positive and negative forms; the negative form of @{b}-ffoo@{ub} would be @{b}-fno-foo@{ub}. This manual documents only one of these two forms, whichever one is not the default. @{" Option Summary " Link "Option Summary"} Brief list of all options, without explanations. @{" Overall Options " Link "Overall Options"} Controlling the kind of output: an executable, object files, assembler files, or preprocessed source. @{" Invoking G++ " Link "Invoking G++"} Compiling C++ programs. @{" C Dialect Options " Link "C Dialect Options"} Controlling the variant of C language compiled. @{" C++ Dialect Options " Link "C++ Dialect Options"} Variations on C++. @{" Warning Options " Link "Warning Options"} How picky should the compiler be? @{" Debugging Options " Link "Debugging Options"} Symbol tables, measurements, and debugging dumps. @{" Optimize Options " Link "Optimize Options"} How much optimization? @{" Preprocessor Options " Link "Preprocessor Options"} Controlling header files and macro definitions. Also, getting dependency information for Make. @{" Assembler Options " Link "Assembler Options"} Passing options to the assembler. @{" Link Options " Link "Link Options"} Specifying libraries and so on. @{" Directory Options " Link "Directory Options"} Where to find header files and libraries. Where to find the compiler executable files. @{" Target Options " Link "Target Options"} Running a cross-compiler, or an old version of GNU CC. @{" Submodel Options " Link "Submodel Options"} Specifying minor hardware or convention variations, such as 68010 vs 68020. @{" Code Gen Options " Link "Code Gen Options"} Specifying conventions for function calls, data layout and register usage. @{" Environment Variables " Link "Environment Variables"} Env vars that affect GNU CC. @{" Running Protoize " Link "Running Protoize"} Automatically adding or removing function prototypes. @EndNode @Node "Option Summary" "gcc.guide/Option Summary" @Next "Overall Options" @Toc "Invoking GCC" Option Summary ============== Here is a summary of all the options, grouped by type. Explanations are in the following sections. @{i}Overall Options@{ui} See @{"Options Controlling the Kind of Output" Link "Overall Options"}. -c -S -E -o @{I}FILE@{ui} -pipe -v -x @{I}LANGUAGE@{ui} @{i}C Language Options@{ui} See @{"Options Controlling C Dialect" Link "C Dialect Options"}. -ansi -fallow-single-precision -fcond-mismatch -fno-asm -fno-builtin -fsigned-bitfields -fsigned-char -funsigned-bitfields -funsigned-char -fwritable-strings -traditional -traditional-cpp -trigraphs @{i}C++ Language Options@{ui} See @{"Options Controlling C++ Dialect" Link "C++ Dialect Options"}. -fall-virtual -fdollars-in-identifiers -felide-constructors -fenum-int-equiv -fexternal-templates -ffor-scope -fno-for-scope -fhandle-signatures -fmemoize-lookups -fno-default-inline -fno-gnu-keywords -fnonnull-objects -foperator-names -fstrict-prototype -fthis-is-variable -nostdinc++ -traditional +e@{I}N@{ui} @{i}Warning Options@{ui} See @{"Options to Request or Suppress Warnings" Link "Warning Options"}. -fsyntax-only -pedantic -pedantic-errors -w -W -Wall -Waggregate-return -Wbad-function-cast -Wcast-align -Wcast-qual -Wchar-subscript -Wcomment -Wconversion -Wenum-clash -Werror -Wformat -Wid-clash-@{I}LEN@{ui} -Wimplicit -Wimport -Winline -Wlarger-than-@{I}LEN@{ui} -Wmissing-declarations -Wmissing-prototypes -Wnested-externs -Wno-import -Woverloaded-virtual -Wparentheses -Wpointer-arith -Wredundant-decls -Wreorder -Wreturn-type -Wshadow -Wstrict-prototypes -Wswitch -Wsynth -Wtemplate-debugging -Wtraditional -Wtrigraphs -Wuninitialized -Wunused -Wwrite-strings @{i}Debugging Options@{ui} See @{"Options for Debugging Your Program or GCC" Link "Debugging Options"}. -a -d@{I}LETTERS@{ui} -fpretend-float -g -g@{I}LEVEL@{ui} -gcoff -gdwarf -gdwarf+ -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+ -p -pg -print-file-name=@{I}LIBRARY@{ui} -print-libgcc-file-name -print-prog-name=@{I}PROGRAM@{ui} -print-search-dirs -save-temps @{i}Optimization Options@{ui} See @{"Options that Control Optimization" Link "Optimize Options"}. -fcaller-saves -fcse-follow-jumps -fcse-skip-blocks -fdelayed-branch -fexpensive-optimizations -ffast-math -ffloat-store -fforce-addr -fforce-mem -finline-functions -fkeep-inline-functions -fno-default-inline -fno-defer-pop -fno-function-cse -fno-inline -fno-peephole -fomit-frame-pointer -frerun-cse-after-loop -fschedule-insns -fschedule-insns2 -fstrength-reduce -fthread-jumps -funroll-all-loops -funroll-loops -O -O0 -O1 -O2 -O3 @{i}Preprocessor Options@{ui} See @{"Options Controlling the Preprocessor" Link "Preprocessor Options"}. -A@{I}QUESTION@{ui}(@{I}ANSWER@{ui}) -C -dD -dM -dN -D@{I}MACRO@{ui} @{i}[@{ui}=@{I}DEFN@{ui} @{i}]@{ui} -E -H -idirafter @{I}DIR@{ui} -include @{I}FILE@{ui} -imacros @{I}FILE@{ui} -iprefix @{I}FILE@{ui} -iwithprefix @{I}DIR@{ui} -iwithprefixbefore @{I}DIR@{ui} -isystem @{I}DIR@{ui} -M -MD -MM -MMD -MG -nostdinc -P -trigraphs -undef -U@{I}MACRO@{ui} -Wp,@{I}OPTION@{ui} @{i}Assembler Option@{ui} See @{"Passing Options to the Assembler" Link "Assembler Options"}. -Wa,@{I}OPTION@{ui} @{i}Linker Options@{ui} See @{"Options for Linking" Link "Link Options"}. @{I}OBJECT-FILE-NAME@{ui} -l@{I}LIBRARY@{ui} -nostartfiles -nodefaultlibs -nostdlib -s -static -shared -symbolic -Wl,@{I}OPTION@{ui} -Xlinker @{I}OPTION@{ui} -u @{I}SYMBOL@{ui} @{i}Directory Options@{ui} See @{"Options for Directory Search" Link "Directory Options"}. -B@{I}PREFIX@{ui} -I@{I}DIR@{ui} -I- -L@{I}DIR@{ui} @{i}Target Options@{ui} See @{"Target Options" Link "Target Options"}. -b @{I}MACHINE@{ui} -V @{I}VERSION@{ui} @{i}Machine Dependent Options@{ui} See @{"Hardware Models and Configurations" Link "Submodel Options"}. @{i}M680x0 Options@{ui} -m68000 -m68020 -m68020-40 -m68030 -m68040 -m68881 -mbitfield -mc68000 -mc68020 -mfpa -mnobitfield -mrtd -mshort -msoft-float @{i}VAX Options@{ui} -mg -mgnu -munix @{i}SPARC Options@{ui} -mapp-regs -mcypress -mepilogue -mflat -mfpu -mhard-float -mhard-quad-float -mno-app-regs -mno-flat -mno-fpu -mno-epilogue -mno-unaligned-doubles -msoft-float -msoft-quad-float -msparclite -msupersparc -munaligned-doubles -mv8 SPARC V9 compilers support the following options in addition to the above: -mmedlow -mmedany -mint32 -mint64 -mlong32 -mlong64 -mno-stack-bias -mstack-bias @{i}Convex Options@{ui} -mc1 -mc2 -mc32 -mc34 -mc38 -margcount -mnoargcount -mlong32 -mlong64 -mvolatile-cache -mvolatile-nocache @{i}AMD29K Options@{ui} -m29000 -m29050 -mbw -mnbw -mdw -mndw -mlarge -mnormal -msmall -mkernel-registers -mno-reuse-arg-regs -mno-stack-check -mno-storem-bug -mreuse-arg-regs -msoft-float -mstack-check -mstorem-bug -muser-registers @{i}ARM Options@{ui} -mapcs -m2 -m3 -m6 -mbsd -mxopen -mno-symrename @{i}M88K Options@{ui} -m88000 -m88100 -m88110 -mbig-pic -mcheck-zero-division -mhandle-large-shift -midentify-revision -mno-check-zero-division -mno-ocs-debug-info -mno-ocs-frame-position -mno-optimize-arg-area -mno-serialize-volatile -mno-underscores -mocs-debug-info -mocs-frame-position -moptimize-arg-area -mserialize-volatile -mshort-data-@{I}NUM@{ui} -msvr3 -msvr4 -mtrap-large-shift -muse-div-instruction -mversion-03.00 -mwarn-passed-structs @{i}RS/6000 and PowerPC Options@{ui} -mcpu=@{I}CPU TYPE@{ui} -mpower -mno-power -mpower2 -mno-power2 -mpowerpc -mno-powerpc -mpowerpc-gpopt -mno-powerpc-gpopt -mpowerpc-gfxopt -mno-powerpc-gfxopt -mnew-mnemonics -mno-new-mnemonics -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc -msoft-float -mhard-float -mmultiple -mno-multiple -mstring -mno-string -mbit-align -mno-bit-align -mstrict-align -mno-strict-align -mrelocatable -mno-relocatable -mtoc -mno-toc -mtraceback -mno-traceback -mlittle -mlittle-endian -mbig -mbig-endian -mcall-aix -mcall-sysv -mprototype @{i}RT Options@{ui} -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs -mfull-fp-blocks -mhc-struct-return -min-line-mul -mminimum-fp-blocks -mnohc-struct-return @{i}MIPS Options@{ui} -mabicalls -mcpu=@{I}CPU TYPE@{ui} -membedded-data -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 -mips2 -mips3 -mlong64 -mlong-calls -mmemcpy -mmips-as -mmips-tfile -mno-abicalls -mno-embedded-data -mno-embedded-pic -mno-gpopt -mno-long-calls -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats -mrnames -msoft-float -m4650 -msingle-float -mmad -mstats -EL -EB -G @{I}NUM@{ui} -nocpp @{i}i386 Options@{ui} -m486 -m386 -mieee-fp -mno-fancy-math-387 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib -mno-wide-multiply -mrtd -malign-double -mreg-alloc=@{I}LIST@{ui} -mregparm=@{I}NUM@{ui} -malign-jumps=@{I}NUM@{ui} -malign-loops=@{I}NUM@{ui} -malign-functions=@{I}NUM@{ui} @{i}HPPA Options@{ui} -mdisable-fpregs -mdisable-indexing -mfast-indirect-calls -mgas -mjump-in-delay -mlong-millicode-calls -mno-disable-fpregs -mno-disable-indexing -mno-fast-indirect-calls -mno-gas -mno-jump-in-delay -mno-millicode-long-calls -mno-portable-runtime -mno-soft-float -msoft-float -mpa-risc-1-0 -mpa-risc-1-1 -mportable-runtime -mschedule=@{I}LIST@{ui} @{i}Intel 960 Options@{ui} -m@{I}CPU TYPE@{ui} -masm-compat -mclean-linkage -mcode-align -mcomplex-addr -mleaf-procedures -mic-compat -mic2.0-compat -mic3.0-compat -mintel-asm -mno-clean-linkage -mno-code-align -mno-complex-addr -mno-leaf-procedures -mno-old-align -mno-strict-align -mno-tail-call -mnumerics -mold-align -msoft-float -mstrict-align -mtail-call @{i}DEC Alpha Options@{ui} -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float @{i}Clipper Options@{ui} -mc300 -mc400 @{i}H8/300 Options@{ui} -mrelax -mh @{i}System V Options@{ui} -Qy -Qn -YP,@{I}PATHS@{ui} -Ym,@{I}DIR@{ui} @{i}Code Generation Options@{ui} See @{"Options for Code Generation Conventions" Link "Code Gen Options"}. -fcall-saved-@{I}REG@{ui} -fcall-used-@{I}REG@{ui} -ffixed-@{I}REG@{ui} -finhibit-size-directive -fno-common -fno-ident -fno-gnu-linker -fpcc-struct-return -fpic -fPIC -freg-struct-return -fshared-data -fshort-enums -fshort-double -fvolatile -fvolatile-global -fverbose-asm -fpack-struct +e0 +e1 @{" Overall Options " Link "Overall Options"} Controlling the kind of output: an executable, object files, assembler files, or preprocessed source. @{" C Dialect Options " Link "C Dialect Options"} Controlling the variant of C language compiled. @{" C++ Dialect Options " Link "C++ Dialect Options"} Variations on C++. @{" Warning Options " Link "Warning Options"} How picky should the compiler be? @{" Debugging Options " Link "Debugging Options"} Symbol tables, measurements, and debugging dumps. @{" Optimize Options " Link "Optimize Options"} How much optimization? @{" Preprocessor Options " Link "Preprocessor Options"} Controlling header files and macro definitions. Also, getting dependency information for Make. @{" Assembler Options " Link "Assembler Options"} Passing options to the assembler. @{" Link Options " Link "Link Options"} Specifying libraries and so on. @{" Directory Options " Link "Directory Options"} Where to find header files and libraries. Where to find the compiler executable files. @{" Target Options " Link "Target Options"} Running a cross-compiler, or an old version of GNU CC. @EndNode @Node "Overall Options" "gcc.guide/Overall Options" @Next "Invoking G++" @Prev "Option Summary" @Toc "Invoking GCC" Options Controlling the Kind of Output ====================================== Compilation can involve up to four stages: preprocessing, compilation proper, assembly and linking, always in that order. The first three stages apply to an individual source file, and end by producing an object file; linking combines all the object files (those newly compiled, and those specified as input) into an executable file. For any given input file, the file name suffix determines what kind of compilation is done: @{b}@{I}FILE@{ui}.c@{ub} C source code which must be preprocessed. @{b}@{I}FILE@{ui}.i@{ub} C source code which should not be preprocessed. @{b}@{I}FILE@{ui}.ii@{ub} C++ source code which should not be preprocessed. @{b}@{I}FILE@{ui}.m@{ub} Objective-C source code. Note that you must link with the library @{b}libobjc.a@{ub} to make an Objective-C program work. @{b}@{I}FILE@{ui}.h@{ub} C header file (not to be compiled or linked). @{b}@{I}FILE@{ui}.cc@{ub} @{b}@{I}FILE@{ui}.cxx@{ub} @{b}@{I}FILE@{ui}.cpp@{ub} @{b}@{I}FILE@{ui}.C@{ub} C++ source code which must be preprocessed. Note that in @{b}.cxx@{ub}, the last two letters must both be literally @{b}x@{ub}. Likewise, @{b}.C@{ub} refers to a literal capital C. @{b}@{I}FILE@{ui}.s@{ub} Assembler code. @{b}@{I}FILE@{ui}.S@{ub} Assembler code which must be preprocessed. @{b}@{I}OTHER@{ui}@{ub} An object file to be fed straight into linking. Any file name with no recognized suffix is treated this way. You can specify the input language explicitly with the @{b}-x@{ub} option: @{b}-x @{I}LANGUAGE@{ui}@{ub} Specify explicitly the @{I}LANGUAGE@{ui} for the following input files (rather than letting the compiler choose a default based on the file name suffix). This option applies to all following input files until the next @{b}-x@{ub} option. Possible values for @{I}LANGUAGE@{ui} are: c objective-c c++ c-header cpp-output c++-cpp-output assembler assembler-with-cpp @{b}-x none@{ub} Turn off any specification of a language, so that subsequent files are handled according to their file name suffixes (as they are if @{b}-x@{ub} has not been used at all). If you only want some of the stages of compilation, you can use @{b}-x@{ub} (or filename suffixes) to tell @{b}gcc@{ub} where to start, and one of the options @{b}-c@{ub}, @{b}-S@{ub}, or @{b}-E@{ub} to say where @{b}gcc@{ub} is to stop. Note that some combinations (for example, @{b}-x cpp-output -E@{ub} instruct @{b}gcc@{ub} to do nothing at all. @{b}-c@{ub} Compile or assemble the source files, but do not link. The linking stage simply is not done. The ultimate output is in the form of an object file for each source file. By default, the object file name for a source file is made by replacing the suffix @{b}.c@{ub}, @{b}.i@{ub}, @{b}.s@{ub}, etc., with @{b}.o@{ub}. Unrecognized input files, not requiring compilation or assembly, are ignored. @{b}-S@{ub} Stop after the stage of compilation proper; do not assemble. The output is in the form of an assembler code file for each non-assembler input file specified. By default, the assembler file name for a source file is made by replacing the suffix @{b}.c@{ub}, @{b}.i@{ub}, etc., with @{b}.s@{ub}. Input files that don't require compilation are ignored. @{b}-E@{ub} Stop after the preprocessing stage; do not run the compiler proper. The output is in the form of preprocessed source code, which is sent to the standard output. Input files which don't require preprocessing are ignored. @{b}-o @{I}FILE@{ui}@{ub} Place output in file @{I}FILE@{ui}. This applies regardless to whatever sort of output is being produced, whether it be an executable file, an object file, an assembler file or preprocessed C code. Since only one output file can be specified, it does not make sense to use @{b}-o@{ub} when compiling more than one input file, unless you are producing an executable file as output. If @{b}-o@{ub} is not specified, the default is to put an executable file in @{b}a.out@{ub}, the object file for @{b}@{I}SOURCE@{ui}.@{I}SUFFIX@{ui}@{ub} in @{b}@{I}SOURCE@{ui}.o@{ub}, its assembler file in @{b}@{I}SOURCE@{ui}.s@{ub}, and all preprocessed C source on standard output. @{b}-v@{ub} Print (on standard error output) the commands executed to run the stages of compilation. Also print the version number of the compiler driver program and of the preprocessor and the compiler proper. @{b}-pipe@{ub} Use pipes rather than temporary files for communication between the various stages of compilation. This fails to work on some systems where the assembler is unable to read from a pipe; but the GNU assembler has no trouble. @EndNode @Node "Invoking G++" "gcc.guide/Invoking G++" @Next "C Dialect Options" @Prev "Overall Options" @Toc "Invoking GCC" Compiling C++ Programs ====================== C++ source files conventionally use one of the suffixes @{b}.C@{ub}, @{b}.cc@{ub}, @{b}cpp@{ub}, or @{b}.cxx@{ub}; preprocessed C++ files use the suffix @{b}.ii@{ub}. GNU CC recognizes files with these names and compiles them as C++ programs even if you call the compiler the same way as for compiling C programs (usually with the name @{b}gcc@{ub}). However, C++ programs often require class libraries as well as a compiler that understands the C++ language--and under some circumstances, you might want to compile programs from standard input, or otherwise without a suffix that flags them as C++ programs. @{b}g++@{ub} is a program that calls GNU CC with the default language set to C++, and automatically specifies linking against the GNU class library libg++. (1) On many systems, the script @{b}g++@{ub} is also installed with the name @{b}c++@{ub}. When you compile C++ programs, you may specify many of the same command-line options that you use for compiling programs in any language; or command-line options meaningful for C and related languages; or options that are meaningful only for C++ programs. See @{"Options Controlling C Dialect" Link "C Dialect Options"}, for explanations of options for languages related to C. See @{"Options Controlling C++ Dialect" Link "C++ Dialect Options"}, for explanations of options that are meaningful only for C++ programs. ---------- Footnotes ---------- (1) Prior to release 2 of the compiler, there was a separate @{b}g++@{ub} compiler. That version was based on GNU CC, but not integrated with it. Versions of @{b}g++@{ub} with a @{b}1.@{I}XX@{ui}@{ub} version number--for example, @{b}g++@{ub} version 1.37 or 1.42--are much less reliable than the versions integrated with GCC 2. Moreover, combining G++ @{b}1.@{I}XX@{ui}@{ub} with a version 2 GCC will simply not work. @EndNode @Node "C Dialect Options" "gcc.guide/C Dialect Options" @Next "C++ Dialect Options" @Prev "Invoking G++" @Toc "Invoking GCC" Options Controlling C Dialect ============================= The following options control the dialect of C (or languages derived from C, such as C++ and Objective C) that the compiler accepts: @{b}-ansi@{ub} Support all ANSI standard C programs. This turns off certain features of GNU C that are incompatible with ANSI C, such as the @{b}asm@{ub}, @{b}inline@{ub} and @{b}typeof@{ub} keywords, and predefined macros such as @{b}unix@{ub} and @{b}vax@{ub} that identify the type of system you are using. It also enables the undesirable and rarely used ANSI trigraph feature, disallows @{b}$@{ub} as part of identifiers, and disables recognition of C++ style @{b}//@{ub} comments. The alternate keywords @{b}__asm__@{ub}, @{b}__extension__@{ub}, @{b}__inline__@{ub} and @{b}__typeof__@{ub} continue to work despite @{b}-ansi@{ub}. You would not want to use them in an ANSI C program, of course, but it is useful to put them in header files that might be included in compilations done with @{b}-ansi@{ub}. Alternate predefined macros such as @{b}__unix__@{ub} and @{b}__vax__@{ub} are also available, with or without @{b}-ansi@{ub}. The @{b}-ansi@{ub} option does not cause non-ANSI programs to be rejected gratuitously. For that, @{b}-pedantic@{ub} is required in addition to @{b}-ansi@{ub}. See @{"Warning Options" Link "Warning Options"}. The macro @{b}__STRICT_ANSI__@{ub} is predefined when the @{b}-ansi@{ub} option is used. Some header files may notice this macro and refrain from declaring certain functions or defining certain macros that the ANSI standard doesn't call for; this is to avoid interfering with any programs that might use these names for other things. The functions @{b}alloca@{ub}, @{b}abort@{ub}, @{b}exit@{ub}, and @{b}_exit@{ub} are not builtin functions when @{b}-ansi@{ub} is used. @{b}-fno-asm@{ub} Do not recognize @{b}asm@{ub}, @{b}inline@{ub} or @{b}typeof@{ub} as a keyword, so that code can use these words as identifiers. You can use the keywords @{b}__asm__@{ub}, @{b}__inline__@{ub} and @{b}__typeof__@{ub} instead. @{b}-ansi@{ub} implies @{b}-fno-asm@{ub}. In C++, this switch only affects the @{b}typeof@{ub} keyword, since @{b}asm@{ub} and @{b}inline@{ub} are standard keywords. You may want to use the @{b}-fno-gnu-keywords@{ub} flag instead, as it also disables the other, C++-specific, extension keywords such as @{b}headof@{ub}. @{b}-fno-builtin@{ub} Don't recognize builtin functions that do not begin with two leading underscores. Currently, the functions affected include @{b}abort@{ub}, @{b}abs@{ub}, @{b}alloca@{ub}, @{b}cos@{ub}, @{b}exit@{ub}, @{b}fabs@{ub}, @{b}ffs@{ub}, @{b}labs@{ub}, @{b}memcmp@{ub}, @{b}memcpy@{ub}, @{b}sin@{ub}, @{b}sqrt@{ub}, @{b}strcmp@{ub}, @{b}strcpy@{ub}, and @{b}strlen@{ub}. GCC normally generates special code to handle certain builtin functions more efficiently; for instance, calls to @{b}alloca@{ub} may become single instructions that adjust the stack directly, and calls to @{b}memcpy@{ub} may become inline copy loops. The resulting code is often both smaller and faster, but since the function calls no longer appear as such, you cannot set a breakpoint on those calls, nor can you change the behavior of the functions by linking with a different library. The @{b}-ansi@{ub} option prevents @{b}alloca@{ub} and @{b}ffs@{ub} from being builtin functions, since these functions do not have an ANSI standard meaning. @{b}-trigraphs@{ub} Support ANSI C trigraphs. You don't want to know about this brain-damage. The @{b}-ansi@{ub} option implies @{b}-trigraphs@{ub}. @{b}-traditional@{ub} Attempt to support some aspects of traditional C compilers. Specifically: @{b}*@{ub} All @{b}extern@{ub} declarations take effect globally even if they are written inside of a function definition. This includes implicit declarations of functions. @{b}*@{ub} The newer keywords @{b}typeof@{ub}, @{b}inline@{ub}, @{b}signed@{ub}, @{b}const@{ub} and @{b}volatile@{ub} are not recognized. (You can still use the alternative keywords such as @{b}__typeof__@{ub}, @{b}__inline__@{ub}, and so on.) @{b}*@{ub} Comparisons between pointers and integers are always allowed. @{b}*@{ub} Integer types @{b}unsigned short@{ub} and @{b}unsigned char@{ub} promote to @{b}unsigned int@{ub}. @{b}*@{ub} Out-of-range floating point literals are not an error. @{b}*@{ub} Certain constructs which ANSI regards as a single invalid preprocessing number, such as @{b}0xe-0xd@{ub}, are treated as expressions instead. @{b}*@{ub} String "constants" are not necessarily constant; they are stored in writable space, and identical looking constants are allocated separately. (This is the same as the effect of @{b}-fwritable-strings@{ub}.) @{b}*@{ub} All automatic variables not declared @{b}register@{ub} are preserved by @{b}longjmp@{ub}. Ordinarily, GNU C follows ANSI C: automatic variables not declared @{b}volatile@{ub} may be clobbered. @{b}*@{ub} The character escape sequences @{b}\\x@{ub} and @{b}\\a@{ub} evaluate as the literal characters @{b}x@{ub} and @{b}a@{ub} respectively. Without @{b}-traditional@{ub}, @{b}\\x@{ub} is a prefix for the hexadecimal representation of a character, and @{b}\\a@{ub} produces a bell. @{b}*@{ub} In C++ programs, assignment to @{b}this@{ub} is permitted with @{b}-traditional@{ub}. (The option @{b}-fthis-is-variable@{ub} also has this effect.) You may wish to use @{b}-fno-builtin@{ub} as well as @{b}-traditional@{ub} if your program uses names that are normally GNU C builtin functions for other purposes of its own. You cannot use @{b}-traditional@{ub} if you include any header files that rely on ANSI C features. Some vendors are starting to ship systems with ANSI C header files and you cannot use @{b}-traditional@{ub} on such systems to compile files that include any system headers. @{b}@{ub} In the preprocessor, comments convert to nothing at all, rather than to a space. This allows traditional token concatenation. @{b}@{ub} In preprocessing directive, the @{b}#@{ub} symbol must appear as the first character of a line. @{b}@{ub} In the preprocessor, macro arguments are recognized within string constants in a macro definition (and their values are stringified, though without additional quote marks, when they appear in such a context). The preprocessor always considers a string constant to end at a newline. @{b}@{ub} The predefined macro @{b}__STDC__@{ub} is not defined when you use @{b}-traditional@{ub}, but @{b}__GNUC__@{ub} is (since the GNU extensions which @{b}__GNUC__@{ub} indicates are not affected by @{b}-traditional@{ub}). If you need to write header files that work differently depending on whether @{b}-traditional@{ub} is in use, by testing both of these predefined macros you can distinguish four situations: GNU C, traditional GNU C, other ANSI C compilers, and other old C compilers. The predefined macro @{b}__STDC_VERSION__@{ub} is also not defined when you use @{b}-traditional@{ub}. See @{"Standard Predefined Macros" Link "cpp.guide/Standard Predefined"}, for more discussion of these and other predefined macros. @{b}@{ub} The preprocessor considers a string constant to end at a newline (unless the newline is escaped with @{b}\\ @{ub}). (Without @{b}-traditional@{ub}, string constants can contain the newline character as typed.) @{b}-traditional-cpp@{ub} Attempt to support some aspects of traditional C preprocessors. This includes the last five items in the table immediately above, but none of the other effects of @{b}-traditional@{ub}. @{b}-fcond-mismatch@{ub} Allow conditional expressions with mismatched types in the second and third arguments. The value of such an expression is void. @{b}-funsigned-char@{ub} Let the type @{b}char@{ub} be unsigned, like @{b}unsigned char@{ub}. Each kind of machine has a default for what @{b}char@{ub} should be. It is either like @{b}unsigned char@{ub} by default or like @{b}signed char@{ub} by default. Ideally, a portable program should always use @{b}signed char@{ub} or @{b}unsigned char@{ub} when it depends on the signedness of an object. But many programs have been written to use plain @{b}char@{ub} and expect it to be signed, or expect it to be unsigned, depending on the machines they were written for. This option, and its inverse, let you make such a program work with the opposite default. The type @{b}char@{ub} is always a distinct type from each of @{b}signed char@{ub} or @{b}unsigned char@{ub}, even though its behavior is always just like one of those two. @{b}-fsigned-char@{ub} Let the type @{b}char@{ub} be signed, like @{b}signed char@{ub}. Note that this is equivalent to @{b}-fno-unsigned-char@{ub}, which is the negative form of @{b}-funsigned-char@{ub}. Likewise, the option @{b}-fno-signed-char@{ub} is equivalent to @{b}-funsigned-char@{ub}. @{b}-fsigned-bitfields@{ub} @{b}-funsigned-bitfields@{ub} @{b}-fno-signed-bitfields@{ub} @{b}-fno-unsigned-bitfields@{ub} These options control whether a bitfield is signed or unsigned, when the declaration does not use either @{b}signed@{ub} or @{b}unsigned@{ub}. By default, such a bitfield is signed, because this is consistent: the basic integer types such as @{b}int@{ub} are signed types. However, when @{b}-traditional@{ub} is used, bitfields are all unsigned no matter what. @{b}-fwritable-strings@{ub} Store string constants in the writable data segment and don't uniquize them. This is for compatibility with old programs which assume they can write into string constants. The option @{b}-traditional@{ub} also has this effect. Writing into string constants is a very bad idea; "constants" should be constant. @{b}-fallow-single-precision@{ub} Do not promote single precision math operations to double precision, even when compiling with @{b}-traditional@{ub}. Traditional K&R C promotes all floating point operations to double precision, regardless of the sizes of the operands. On the architecture for which you are compiling, single precision may be faster than double precision. If you must use @{b}-traditional@{ub}, but want to use single precision operations when the operands are single precision, use this option. This option has no effect when compiling with ANSI or GNU C conventions (the default). @EndNode @Node "C++ Dialect Options" "gcc.guide/C++ Dialect Options" @Next "Warning Options" @Prev "C Dialect Options" @Toc "Invoking GCC" Options Controlling C++ Dialect =============================== This section describes the command-line options that are only meaningful for C++ programs; but you can also use most of the GNU compiler options regardless of what language your program is in. For example, you might compile a file @{b}firstClass.C@{ub} like this: g++ -g -felide-constructors -O -c firstClass.C In this example, only @{b}-felide-constructors@{ub} is an option meant only for C++ programs; you can use the other options with any language supported by GNU CC. Here is a list of options that are @{i}only@{ui} for compiling C++ programs: @{b}-fno-access-control@{ub} Turn off all access checking. This switch is mainly useful for working around bugs in the access control code. @{b}-fall-virtual@{ub} Treat all possible member functions as virtual, implicitly. All member functions (except for constructor functions and @{b}new@{ub} or @{b}delete@{ub} member operators) are treated as virtual functions of the class where they appear. This does not mean that all calls to these member functions will be made through the internal table of virtual functions. Under some circumstances, the compiler can determine that a call to a given virtual function can be made directly; in these cases the calls are direct in any case. @{b}-fcheck-new@{ub} Check that the pointer returned by @{b}operator new@{ub} is non-null before attempting to modify the storage allocated. The current Working Paper requires that @{b}operator new@{ub} never return a null pointer, so this check is normally unnecessary. @{b}-fconserve-space@{ub} Put uninitialized or runtime-initialized global variables into the common segment, as C does. This saves space in the executable at the cost of not diagnosing duplicate definitions. If you compile with this flag and your program mysteriously crashes after @{b}main()@{ub} has completed, you may have an object that is being destroyed twice because two definitions were merged. @{b}-fdollars-in-identifiers@{ub} Accept @{b}$@{ub} in identifiers. You can also explicitly prohibit use of @{b}$@{ub} with the option @{b}-fno-dollars-in-identifiers@{ub}. (GNU C++ allows @{b}$@{ub} by default on some target systems but not others.) Traditional C allowed the character @{b}$@{ub} to form part of identifiers. However, ANSI C and C++ forbid @{b}$@{ub} in identifiers. @{b}-fenum-int-equiv@{ub} Anachronistically permit implicit conversion of @{b}int@{ub} to enumeration types. Current C++ allows conversion of @{b}enum@{ub} to @{b}int@{ub}, but not the other way around. @{b}-fexternal-templates@{ub} Cause template instantiations to obey @{b}#pragma interface@{ub} and @{b}implementation@{ub}; template instances are emitted or not according to the location of the template definition. See @{"Template Instantiation" Link "Template Instantiation"}, for more information. @{b}-falt-external-templates@{ub} Similar to -fexternal-templates, but template instances are emitted or not according to the place where they are first instantiated. See @{"Template Instantiation" Link "Template Instantiation"}, for more information. @{b}-ffor-scope@{ub} @{b}-fno-for-scope@{ub} If -ffor-scope is specified, the scope of variables declared in a @{i}for-init-statement@{ui} is limited to the @{b}for@{ub} loop itself, as specified by the draft C++ standard. If -fno-for-scope is specified, the scope of variables declared in a @{i}for-init-statement@{ui} extends to the end of the enclosing scope, as was the case in old versions of gcc, and other (traditional) implementations of C++. The default if neither flag is given to follow the standard, but to allow and give a warning for old-style code that would otherwise be invalid, or have different behavior. @{b}-fno-gnu-keywords@{ub} Do not recognize @{b}classof@{ub}, @{b}headof@{ub}, @{b}signature@{ub}, @{b}sigof@{ub} or @{b}typeof@{ub} as a keyword, so that code can use these words as identifiers. You can use the keywords @{b}__classof__@{ub}, @{b}__headof__@{ub}, @{b}__signature__@{ub}, @{b}__sigof__@{ub}, and @{b}__typeof__@{ub} instead. @{b}-ansi@{ub} implies @{b}-fno-gnu-keywords@{ub}. @{b}-fno-implicit-templates@{ub} Never emit code for templates which are instantiated implicitly (i.e. by use); only emit code for explicit instantiations. See @{"Template Instantiation" Link "Template Instantiation"}, for more information. @{b}-fhandle-signatures@{ub} Recognize the @{b}signature@{ub} and @{b}sigof@{ub} keywords for specifying abstract types. The default (@{b}-fno-handle-signatures@{ub}) is not to recognize them. See @{"Type Abstraction using Signatures" Link "C++ Signatures"}. @{b}-fhuge-objects@{ub} Support virtual function calls for objects that exceed the size representable by a @{b}short int@{ub}. Users should not use this flag by default; if you need to use it, the compiler will tell you so. If you compile any of your code with this flag, you must compile @{i}all@{ui} of your code with this flag (including libg++, if you use it). This flag is not useful when compiling with -fvtable-thunks. @{b}-fno-implement-inlines@{ub} To save space, do not emit out-of-line copies of inline functions controlled by @{b}#pragma implementation@{ub}. This will cause linker errors if these functions are not inlined everywhere they are called. @{b}-fmemoize-lookups@{ub} @{b}-fsave-memoized@{ub} Use heuristics to compile faster. These heuristics are not enabled by default, since they are only effective for certain input files. Other input files compile more slowly. The first time the compiler must build a call to a member function (or reference to a data member), it must (1) determine whether the class implements member functions of that name; (2) resolve which member function to call (which involves figuring out what sorts of type conversions need to be made); and (3) check the visibility of the member function to the caller. All of this adds up to slower compilation. Normally, the second time a call is made to that member function (or reference to that data member), it must go through the same lengthy process again. This means that code like this: cout << "This " << p << " has " << n << " legs.\\n"; makes six passes through all three steps. By using a software cache, a "hit" significantly reduces this cost. Unfortunately, using the cache introduces another layer of mechanisms which must be implemented, and so incurs its own overhead. @{b}-fmemoize-lookups@{ub} enables the software cache. Because access privileges (visibility) to members and member functions may differ from one function context to the next, G++ may need to flush the cache. With the @{b}-fmemoize-lookups@{ub} flag, the cache is flushed after every function that is compiled. The @{b}-fsave-memoized@{ub} flag enables the same software cache, but when the compiler determines that the context of the last function compiled would yield the same access privileges of the next function to compile, it preserves the cache. This is most helpful when defining many member functions for the same class: with the exception of member functions which are friends of other classes, each member function has exactly the same access privileges as every other, and the cache need not be flushed. The code that implements these flags has rotted; you should probably avoid using them. @{b}-fstrict-prototype@{ub} Within an @{b}extern "C"@{ub} linkage specification, treat a function declaration with no arguments, such as @{b}int foo ();@{ub}, as declaring the function to take no arguments. Normally, such a declaration means that the function @{b}foo@{ub} can take any combination of arguments, as in C. @{b}-pedantic@{ub} implies @{b}-fstrict-prototype@{ub} unless overridden with @{b}-fno-strict-prototype@{ub}. This flag no longer affects declarations with C++ linkage. @{b}-fno-nonnull-objects@{ub} Don't assume that a reference is initialized to refer to a valid object. Although the current C++ Working Paper prohibits null references, some old code may rely on them, and you can use @{b}-fno-nonnull-objects@{ub} to turn on checking. At the moment, the compiler only does this checking for conversions to virtual base classes. @{b}-foperator-names@{ub} Recognize the operator name keywords @{b}and@{ub}, @{b}bitand@{ub}, @{b}bitor@{ub}, @{b}compl@{ub}, @{b}not@{ub}, @{b}or@{ub} and @{b}xor@{ub} as synonyms for the symbols they refer to. @{b}-ansi@{ub} implies @{b}-foperator-names@{ub}. @{b}-fthis-is-variable@{ub} Permit assignment to @{b}this@{ub}. The incorporation of user-defined free store management into C++ has made assignment to @{b}this@{ub} an anachronism. Therefore, by default it is invalid to assign to @{b}this@{ub} within a class member function; that is, GNU C++ treats @{b}this@{ub} in a member function of class @{b}X@{ub} as a non-lvalue of type @{b}X *@{ub}. However, for backwards compatibility, you can make it valid with @{b}-fthis-is-variable@{ub}. @{b}-fvtable-thunks@{ub} Use @{b}thunks@{ub} to implement the virtual function dispatch table (@{b}vtable@{ub}). The traditional (cfront-style) approach to implementing vtables was to store a pointer to the function and two offsets for adjusting the @{b}this@{ub} pointer at the call site. Newer implementations store a single pointer to a @{b}thunk@{ub} function which does any necessary adjustment and then calls the target function. This option also enables a heuristic for controlling emission of vtables; if a class has any non-inline virtual functions, the vtable will be emitted in the translation unit containing the first one of those. @{b}-nostdinc++@{ub} Do not search for header files in the standard directories specific to C++, but do still search the other standard directories. (This option is used when building libg++.) @{b}-traditional@{ub} For C++ programs (in addition to the effects that apply to both C and C++), this has the same effect as @{b}-fthis-is-variable@{ub}. See @{"Options Controlling C Dialect" Link "C Dialect Options"}. In addition, these optimization, warning, and code generation options have meanings only for C++ programs: @{b}-fno-default-inline@{ub} Do not assume @{b}inline@{ub} for functions defined inside a class scope. See @{"Options That Control Optimization" Link "Optimize Options"}. @{b}-Wenum-clash@{ub} @{b}-Woverloaded-virtual@{ub} @{b}-Wtemplate-debugging@{ub} Warnings that apply only to C++ programs. See @{"Options to Request or Suppress Warnings" Link "Warning Options"}. @{b}+e@{I}N@{ui}@{ub} Control how virtual function definitions are used, in a fashion compatible with @{b}cfront@{ub} 1.x. See @{"Options for Code Generation Conventions" Link "Code Gen Options"}. @EndNode @Node "Warning Options" "gcc.guide/Warning Options" @Next "Debugging Options" @Prev "C++ Dialect Options" @Toc "Invoking GCC" Options to Request or Suppress Warnings ======================================= Warnings are diagnostic messages that report constructions which are not inherently erroneous but which are risky or suggest there may have been an error. You can request many specific warnings with options beginning @{b}-W@{ub}, for example @{b}-Wimplicit@{ub} to request warnings on implicit declarations. Each of these specific warning options also has a negative form beginning @{b}-Wno-@{ub} to turn off warnings; for example, @{b}-Wno-implicit@{ub}. This manual lists only one of the two forms, whichever is not the default. These options control the amount and kinds of warnings produced by GNU CC: @{b}-fsyntax-only@{ub} Check the code for syntax errors, but don't do anything beyond that. @{b}-pedantic@{ub} Issue all the warnings demanded by strict ANSI standard C; reject all programs that use forbidden extensions. Valid ANSI standard C programs should compile properly with or without this option (though a rare few will require @{b}-ansi@{ub}). However, without this option, certain GNU extensions and traditional C features are supported as well. With this option, they are rejected. @{b}-pedantic@{ub} does not cause warning messages for use of the alternate keywords whose names begin and end with @{b}__@{ub}. Pedantic warnings are also disabled in the expression that follows @{b}__extension__@{ub}. However, only system header files should use these escape routes; application programs should avoid them. See @{"Alternate Keywords" Link "Alternate Keywords"}. This option is not intended to be @{i}useful@{ui}; it exists only to satisfy pedants who would otherwise claim that GNU CC fails to support the ANSI standard. Some users try to use @{b}-pedantic@{ub} to check programs for strict ANSI C conformance. They soon find that it does not do quite what they want: it finds some non-ANSI practices, but not all--only those for which ANSI C @{i}requires@{ui} a diagnostic. A feature to report any failure to conform to ANSI C might be useful in some instances, but would require considerable additional work and would be quite different from @{b}-pedantic@{ub}. We recommend, rather, that users take advantage of the extensions of GNU C and disregard the limitations of other compilers. Aside from certain supercomputers and obsolete small machines, there is less and less reason ever to use any other C compiler other than for bootstrapping GNU CC. @{b}-pedantic-errors@{ub} Like @{b}-pedantic@{ub}, except that errors are produced rather than warnings. @{b}-w@{ub} Inhibit all warning messages. @{b}-Wno-import@{ub} Inhibit warning messages about the use of @{b}#import@{ub}. @{b}-Wchar-subscripts@{ub} Warn if an array subscript has type @{b}char@{ub}. This is a common cause of error, as programmers often forget that this type is signed on some machines. @{b}-Wcomment@{ub} Warn whenever a comment-start sequence @{b}/*@{ub} appears in a comment. @{b}-Wformat@{ub} Check calls to @{b}printf@{ub} and @{b}scanf@{ub}, etc., to make sure that the arguments supplied have types appropriate to the format string specified. @{b}-Wimplicit@{ub} Warn whenever a function or parameter is implicitly declared. @{b}-Wparentheses@{ub} Warn if parentheses are omitted in certain contexts, such as when there is an assignment in a context where a truth value is expected, or when operators are nested whose precedence people often get confused about. @{b}-Wreturn-type@{ub} Warn whenever a function is defined with a return-type that defaults to @{b}int@{ub}. Also warn about any @{b}return@{ub} statement with no return-value in a function whose return-type is not @{b}void@{ub}. @{b}-Wswitch@{ub} Warn whenever a @{b}switch@{ub} statement has an index of enumeral type and lacks a @{b}case@{ub} for one or more of the named codes of that enumeration. (The presence of a @{b}default@{ub} label prevents this warning.) @{b}case@{ub} labels outside the enumeration range also provoke warnings when this option is used. @{b}-Wtrigraphs@{ub} Warn if any trigraphs are encountered (assuming they are enabled). @{b}-Wunused@{ub} Warn whenever a variable is unused aside from its declaration, whenever a function is declared static but never defined, whenever a label is declared but not used, and whenever a statement computes a result that is explicitly not used. To suppress this warning for an expression, simply cast it to void. For unused variables and parameters, use the @{b}unused@{ub} attribute (see @{"Variable Attributes" Link "Variable Attributes"}). @{b}-Wuninitialized@{ub} An automatic variable is used without first being initialized. These warnings are possible only in optimizing compilation, because they require data flow information that is computed only when optimizing. If you don't specify @{b}-O@{ub}, you simply won't get these warnings. These warnings occur only for variables that are candidates for register allocation. Therefore, they do not occur for a variable that is declared @{b}volatile@{ub}, or whose address is taken, or whose size is other than 1, 2, 4 or 8 bytes. Also, they do not occur for structures, unions or arrays, even when they are in registers. Note that there may be no warning about a variable that is used only to compute a value that itself is never used, because such computations may be deleted by data flow analysis before the warnings are printed. These warnings are made optional because GNU CC is not smart enough to see all the reasons why the code might be correct despite appearing to have an error. Here is one example of how this can happen: { int x; switch (y) { case 1: x = 1; break; case 2: x = 4; break; case 3: x = 5; } foo (x); } If the value of @{b}y@{ub} is always 1, 2 or 3, then @{b}x@{ub} is always initialized, but GNU CC doesn't know this. Here is another common case: { int save_y; if (change_y) save_y = y, y = new_y; ... if (change_y) y = save_y; } This has no bug because @{b}save_y@{ub} is used only if it is set. Some spurious warnings can be avoided if you declare all the functions you use that never return as @{b}noreturn@{ub}. See @{"Function Attributes" Link "Function Attributes"}. @{b}-Wenum-clash@{ub} Warn about conversion between different enumeration types. (C++ only). @{b}-Wreorder (C++ only)@{ub} Warn when the order of member initializers given in the code does not match the order in which they must be executed. For instance: struct A { int i; int j; A(): j (0), i (1) { } }; Here the compiler will warn that the member initializers for @{b}i@{ub} and @{b}j@{ub} will be rearranged to match the declaration order of the members. @{b}-Wtemplate-debugging@{ub} When using templates in a C++ program, warn if debugging is not yet fully available (C++ only). @{b}-Wall@{ub} All of the above @{b}-W@{ub} options combined. These are all the options which pertain to usage that we recommend avoiding and that we believe is easy to avoid, even in conjunction with macros. The remaining @{b}-W...@{ub} options are not implied by @{b}-Wall@{ub} because they warn about constructions that we consider reasonable to use, on occasion, in clean programs. @{b}-W@{ub} Print extra warning messages for these events: @{b}*@{ub} A nonvolatile automatic variable might be changed by a call to @{b}longjmp@{ub}. These warnings as well are possible only in optimizing compilation. The compiler sees only the calls to @{b}setjmp@{ub}. It cannot know where @{b}longjmp@{ub} will be called; in fact, a signal handler could call it at any point in the code. As a result, you may get a warning even when there is in fact no problem because @{b}longjmp@{ub} cannot in fact be called at the place which would cause a problem. @{b}*@{ub} A function can return either with or without a value. (Falling off the end of the function body is considered returning without a value.) For example, this function would evoke such a warning: foo (a) { if (a > 0) return a; } @{b}*@{ub} An expression-statement or the left-hand side of a comma expression contains no side effects. To suppress the warning, cast the unused expression to void. For example, an expression such as @{b}x[i,j]@{ub} will cause a warning, but @{b}x[(void)i,j]@{ub} will not. @{b}*@{ub} An unsigned value is compared against zero with @{b}<@{ub} or @{b}<=@{ub}. @{b}*@{ub} A comparison like @{b}x<=y<=z@{ub} appears; this is equivalent to @{b}(x<=y ? 1 : 0) <= z@{ub}, which is a different interpretation from that of ordinary mathematical notation. @{b}*@{ub} Storage-class specifiers like @{b}static@{ub} are not the first things in a declaration. According to the C Standard, this usage is obsolescent. @{b}*@{ub} If @{b}-Wall@{ub} or @{b}-Wunused@{ub} is also specified, warn about unused arguments. @{b}*@{ub} An aggregate has a partly bracketed initializer. For example, the following code would evoke such a warning, because braces are missing around the initializer for @{b}x.h@{ub}: struct s { int f, g; }; struct t { struct s h; int i; }; struct t x = { 1, 2, 3 }; @{b}-Wtraditional@{ub} Warn about certain constructs that behave differently in traditional and ANSI C. @{b}*@{ub} Macro arguments occurring within string constants in the macro body. These would substitute the argument in traditional C, but are part of the constant in ANSI C. @{b}*@{ub} A function declared external in one block and then used after the end of the block. @{b}*@{ub} A @{b}switch@{ub} statement has an operand of type @{b}long@{ub}. @{b}-Wshadow@{ub} Warn whenever a local variable shadows another local variable. @{b}-Wid-clash-@{I}LEN@{ui}@{ub} Warn whenever two distinct identifiers match in the first @{I}LEN@{ui} characters. This may help you prepare a program that will compile with certain obsolete, brain-damaged compilers. @{b}-Wlarger-than-@{I}LEN@{ui}@{ub} Warn whenever an object of larger than @{I}LEN@{ui} bytes is defined. @{b}-Wpointer-arith@{ub} Warn about anything that depends on the "size of" a function type or of @{b}void@{ub}. GNU C assigns these types a size of 1, for convenience in calculations with @{b}void *@{ub} pointers and pointers to functions. @{b}-Wbad-function-cast@{ub} Warn whenever a function call is cast to a non-matching type. For example, warn if @{b}int malloc()@{ub} is cast to @{b}anything *@{ub}. @{b}-Wcast-qual@{ub} Warn whenever a pointer is cast so as to remove a type qualifier from the target type. For example, warn if a @{b}const char *@{ub} is cast to an ordinary @{b}char *@{ub}. @{b}-Wcast-align@{ub} Warn whenever a pointer is cast such that the required alignment of the target is increased. For example, warn if a @{b}char *@{ub} is cast to an @{b}int *@{ub} on machines where integers can only be accessed at two- or four-byte boundaries. @{b}-Wwrite-strings@{ub} Give string constants the type @{b}const char[@{I}LENGTH@{ui}]@{ub} so that copying the address of one into a non-@{b}const@{ub} @{b}char *@{ub} pointer will get a warning. These warnings will help you find at compile time code that can try to write into a string constant, but only if you have been very careful about using @{b}const@{ub} in declarations and prototypes. Otherwise, it will just be a nuisance; this is why we did not make @{b}-Wall@{ub} request these warnings. @{b}-Wconversion@{ub} Warn if a prototype causes a type conversion that is different from what would happen to the same argument in the absence of a prototype. This includes conversions of fixed point to floating and vice versa, and conversions changing the width or signedness of a fixed point argument except when the same as the default promotion. Also, warn if a negative integer constant expression is implicitly converted to an unsigned type. For example, warn about the assignment @{b}x = -1@{ub} if @{b}x@{ub} is unsigned. But do not warn about explicit casts like @{b}(unsigned) -1@{ub}. @{b}-Waggregate-return@{ub} Warn if any functions that return structures or unions are defined or called. (In languages where you can return an array, this also elicits a warning.) @{b}-Wstrict-prototypes@{ub} Warn if a function is declared or defined without specifying the argument types. (An old-style function definition is permitted without a warning if preceded by a declaration which specifies the argument types.) @{b}-Wmissing-prototypes@{ub} Warn if a global function is defined without a previous prototype declaration. This warning is issued even if the definition itself provides a prototype. The aim is to detect global functions that fail to be declared in header files. @{b}-Wmissing-declarations@{ub} Warn if a global function is defined without a previous declaration. Do so even if the definition itself provides a prototype. Use this option to detect global functions that are not declared in header files. @{b}-Wredundant-decls@{ub} Warn if anything is declared more than once in the same scope, even in cases where multiple declaration is valid and changes nothing. @{b}-Wnested-externs@{ub} Warn if an @{b}extern@{ub} declaration is encountered within an function. @{b}-Winline@{ub} Warn if a function can not be inlined, and either it was declared as inline, or else the @{b}-finline-functions@{ub} option was given. @{b}-Woverloaded-virtual@{ub} Warn when a derived class function declaration may be an error in defining a virtual function (C++ only). In a derived class, the definitions of virtual functions must match the type signature of a virtual function declared in the base class. With this option, the compiler warns when you define a function with the same name as a virtual function, but with a type signature that does not match any declarations from the base class. @{b}-Wsynth (C++ only)@{ub} Warn when g++'s synthesis behavior does not match that of cfront. For instance: struct A { operator int (); A& operator = (int); }; main () { A a,b; a = b; } In this example, g++ will synthesize a default @{b}A& operator = (const A&);@{ub}, while cfront will use the user-defined @{b}operator =@{ub}. @{b}-Werror@{ub} Make all warnings into errors. @EndNode @Node "Debugging Options" "gcc.guide/Debugging Options" @Next "Optimize Options" @Prev "Warning Options" @Toc "Invoking GCC" Options for Debugging Your Program or GNU CC ============================================ GNU CC has various special options that are used for debugging either your program or GCC: @{b}-g@{ub} Produce debugging information in the operating system's native format (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging information. On most systems that use stabs format, @{b}-g@{ub} enables use of extra debugging information that only GDB can use; this extra information makes debugging work better in GDB but will probably make other debuggers crash or refuse to read the program. If you want to control for certain whether to generate the extra information, use @{b}-gstabs+@{ub}, @{b}-gstabs@{ub}, @{b}-gxcoff+@{ub}, @{b}-gxcoff@{ub}, @{b}-gdwarf+@{ub}, or @{b}-gdwarf@{ub} (see below). Unlike most other C compilers, GNU CC allows you to use @{b}-g@{ub} with @{b}-O@{ub}. The shortcuts taken by optimized code may occasionally produce surprising results: some variables you declared may not exist at all; flow of control may briefly move where you did not expect it; some statements may not be executed because they compute constant results or their values were already at hand; some statements may execute in different places because they were moved out of loops. Nevertheless it proves possible to debug optimized output. This makes it reasonable to use the optimizer for programs that might have bugs. The following options are useful when GNU CC is generated with the capability for more than one debugging format. @{b}-ggdb@{ub} Produce debugging information in the native format (if that is supported), including GDB extensions if at all possible. @{b}-gstabs@{ub} Produce debugging information in stabs format (if that is supported), without GDB extensions. This is the format used by DBX on most BSD systems. On MIPS, Alpha and System V Release 4 systems this option produces stabs debugging output which is not understood by DBX or SDB. On System V Release 4 systems this option requires the GNU assembler. @{b}-gstabs+@{ub} Produce debugging information in stabs format (if that is supported), using GNU extensions understood only by the GNU debugger (GDB). The use of these extensions is likely to make other debuggers crash or refuse to read the program. @{b}-gcoff@{ub} Produce debugging information in COFF format (if that is supported). This is the format used by SDB on most System V systems prior to System V Release 4. @{b}-gxcoff@{ub} Produce debugging information in XCOFF format (if that is supported). This is the format used by the DBX debugger on IBM RS/6000 systems. @{b}-gxcoff+@{ub} Produce debugging information in XCOFF format (if that is supported), using GNU extensions understood only by the GNU debugger (GDB). The use of these extensions is likely to make other debuggers crash or refuse to read the program, and may cause assemblers other than the GNU assembler (GAS) to fail with an error. @{b}-gdwarf@{ub} Produce debugging information in DWARF format (if that is supported). This is the format used by SDB on most System V Release 4 systems. @{b}-gdwarf+@{ub} Produce debugging information in DWARF format (if that is supported), using GNU extensions understood only by the GNU debugger (GDB). The use of these extensions is likely to make other debuggers crash or refuse to read the program. @{b}-g@{I}LEVEL@{ui}@{ub} @{b}-ggdb@{I}LEVEL@{ui}@{ub} @{b}-gstabs@{I}LEVEL@{ui}@{ub} @{b}-gcoff@{I}LEVEL@{ui}@{ub} @{b}-gxcoff@{I}LEVEL@{ui}@{ub} @{b}-gdwarf@{I}LEVEL@{ui}@{ub} Request debugging information and also use @{I}LEVEL@{ui} to specify how much information. The default level is 2. Level 1 produces minimal information, enough for making backtraces in parts of the program that you don't plan to debug. This includes descriptions of functions and external variables, but no information about local variables and no line numbers. Level 3 includes extra information, such as all the macro definitions present in the program. Some debuggers support macro expansion when you use @{b}-g3@{ub}. @{b}-p@{ub} Generate extra code to write profile information suitable for the analysis program @{b}prof@{ub}. You must use this option when compiling the source files you want data about, and you must also use it when linking. @{b}-pg@{ub} Generate extra code to write profile information suitable for the analysis program @{b}gprof@{ub}. You must use this option when compiling the source files you want data about, and you must also use it when linking. @{b}-a@{ub} Generate extra code to write profile information for basic blocks, which will record the number of times each basic block is executed, the basic block start address, and the function name containing the basic block. If @{b}-g@{ub} is used, the line number and filename of the start of the basic block will also be recorded. If not overridden by the machine description, the default action is to append to the text file @{b}bb.out@{ub}. This data could be analyzed by a program like @{b}tcov@{ub}. Note, however, that the format of the data is not what @{b}tcov@{ub} expects. Eventually GNU @{b}gprof@{ub} should be extended to process this data. @{b}-d@{I}LETTERS@{ui}@{ub} Says to make debugging dumps during compilation at times specified by @{I}LETTERS@{ui}. This is used for debugging the compiler. The file names for most of the dumps are made by appending a word to the source file name (e.g. @{b}foo.c.rtl@{ub} or @{b}foo.c.jump@{ub}). Here are the possible letters for use in @{I}LETTERS@{ui}, and their meanings: @{b}M@{ub} Dump all macro definitions, at the end of preprocessing, and write no output. @{b}N@{ub} Dump all macro names, at the end of preprocessing. @{b}D@{ub} Dump all macro definitions, at the end of preprocessing, in addition to normal output. @{b}y@{ub} Dump debugging information during parsing, to standard error. @{b}r@{ub} Dump after RTL generation, to @{b}@{I}FILE@{ui}.rtl@{ub}. @{b}x@{ub} Just generate RTL for a function instead of compiling it. Usually used with @{b}r@{ub}. @{b}j@{ub} Dump after first jump optimization, to @{b}@{I}FILE@{ui}.jump@{ub}. @{b}s@{ub} Dump after CSE (including the jump optimization that sometimes follows CSE), to @{b}@{I}FILE@{ui}.cse@{ub}. @{b}L@{ub} Dump after loop optimization, to @{b}@{I}FILE@{ui}.loop@{ub}. @{b}t@{ub} Dump after the second CSE pass (including the jump optimization that sometimes follows CSE), to @{b}@{I}FILE@{ui}.cse2@{ub}. @{b}f@{ub} Dump after flow analysis, to @{b}@{I}FILE@{ui}.flow@{ub}. @{b}c@{ub} Dump after instruction combination, to the file @{b}@{I}FILE@{ui}.combine@{ub}. @{b}S@{ub} Dump after the first instruction scheduling pass, to @{b}@{I}FILE@{ui}.sched@{ub}. @{b}l@{ub} Dump after local register allocation, to @{b}@{I}FILE@{ui}.lreg@{ub}. @{b}g@{ub} Dump after global register allocation, to @{b}@{I}FILE@{ui}.greg@{ub}. @{b}R@{ub} Dump after the second instruction scheduling pass, to @{b}@{I}FILE@{ui}.sched2@{ub}. @{b}J@{ub} Dump after last jump optimization, to @{b}@{I}FILE@{ui}.jump2@{ub}. @{b}d@{ub} Dump after delayed branch scheduling, to @{b}@{I}FILE@{ui}.dbr@{ub}. @{b}k@{ub} Dump after conversion from registers to stack, to @{b}@{I}FILE@{ui}.stack@{ub}. @{b}a@{ub} Produce all the dumps listed above. @{b}m@{ub} Print statistics on memory usage, at the end of the run, to standard error. @{b}p@{ub} Annotate the assembler output with a comment indicating which pattern and alternative was used. @{b}-fpretend-float@{ub} When running a cross-compiler, pretend that the target machine uses the same floating point format as the host machine. This causes incorrect output of the actual floating constants, but the actual instruction sequence will probably be the same as GNU CC would make when running on the target machine. @{b}-save-temps@{ub} Store the usual "temporary" intermediate files permanently; place them in the current directory and name them based on the source file. Thus, compiling @{b}foo.c@{ub} with @{b}-c -save-temps@{ub} would produce files @{b}foo.i@{ub} and @{b}foo.s@{ub}, as well as @{b}foo.o@{ub}. @{b}-print-file-name=@{I}LIBRARY@{ui}@{ub} Print the full absolute name of the library file @{I}LIBRARY@{ui} that would be used when linking--and don't do anything else. With this option, GNU CC does not compile or link anything; it just prints the file name. @{b}-print-prog-name=@{I}PROGRAM@{ui}@{ub} Like @{b}-print-file-name@{ub}, but searches for a program such as @{b}cpp@{ub}. @{b}-print-libgcc-file-name@{ub} Same as @{b}-print-file-name=libgcc.a@{ub}. This is useful when you use @{b}-nostdlib@{ub} or @{b}-nodefaultlibs@{ub} but you do want to link with @{b}libgcc.a@{ub}. You can do gcc -nostdlib @{I}FILES@{ui} ... `gcc -print-libgcc-file-name` @{b}-print-search-dirs@{ub} Print the name of the configured installation directory and a list of program and library directories gcc will search--and don't do anything else. This is useful when gcc prints the error message @{b}installation problem, cannot exec cpp: No such file or directory@{ub}. To resolve this you either need to put @{b}cpp@{ub} and the other compiler components where gcc expects to find them, or you can set the environment variable @{b}GCC_EXEC_PREFIX@{ub} to the directory where you installed them. Don't forget the trailing '/'. See @{"Environment Variables" Link "Environment Variables"}. @EndNode @Node "Optimize Options" "gcc.guide/Optimize Options" @Next "Preprocessor Options" @Prev "Debugging Options" @Toc "Invoking GCC" Options That Control Optimization ================================= These options control various sorts of optimizations: @{b}-O@{ub} @{b}-O1@{ub} Optimize. Optimizing compilation takes somewhat more time, and a lot more memory for a large function. Without @{b}-O@{ub}, the compiler's goal is to reduce the cost of compilation and to make debugging produce the expected results. Statements are independent: if you stop the program with a breakpoint between statements, you can then assign a new value to any variable or change the program counter to any other statement in the function and get exactly the results you would expect from the source code. Without @{b}-O@{ub}, the compiler only allocates variables declared @{b}register@{ub} in registers. The resulting compiled code is a little worse than produced by PCC without @{b}-O@{ub}. With @{b}-O@{ub}, the compiler tries to reduce code size and execution time. When you specify @{b}-O@{ub}, the compiler turns on @{b}-fthread-jumps@{ub} and @{b}-fdefer-pop@{ub} on all machines. The compiler turns on @{b}-fdelayed-branch@{ub} on machines that have delay slots, and @{b}-fomit-frame-pointer@{ub} on machines that can support debugging even without a frame pointer. On some machines the compiler also turns on other flags. @{b}-O2@{ub} Optimize even more. GNU CC performs nearly all supported optimizations that do not involve a space-speed tradeoff. The compiler does not perform loop unrolling or function inlining when you specify @{b}-O2@{ub}. As compared to @{b}-O@{ub}, this option increases both compilation time and the performance of the generated code. @{b}-O2@{ub} turns on all optional optimizations except for loop unrolling and function inlining. It also turns on the @{b}-fforce-mem@{ub} option on all machines and frame pointer elimination on machines where doing so does not interfere with debugging. @{b}-O3@{ub} Optimize yet more. @{b}-O3@{ub} turns on all optimizations specified by @{b}-O2@{ub} and also turns on the @{b}inline-functions@{ub} option. @{b}-O0@{ub} Do not optimize. If you use multiple @{b}-O@{ub} options, with or without level numbers, the last such option is the one that is effective. Options of the form @{b}-f@{I}FLAG@{ui}@{ub} specify machine-independent flags. Most flags have both positive and negative forms; the negative form of @{b}-ffoo@{ub} would be @{b}-fno-foo@{ub}. In the table below, only one of the forms is listed--the one which is not the default. You can figure out the other form by either removing @{b}no-@{ub} or adding it. @{b}-ffloat-store@{ub} Do not store floating point variables in registers, and inhibit other options that might change whether a floating point value is taken from a register or memory. This option prevents undesirable excess precision on machines such as the 68000 where the floating registers (of the 68881) keep more precision than a @{b}double@{ub} is supposed to have. For most programs, the excess precision does only good, but a few programs rely on the precise definition of IEEE floating point. Use @{b}-ffloat-store@{ub} for such programs. @{b}-fno-default-inline@{ub} Do not make member functions inline by default merely because they are defined inside the class scope (C++ only). Otherwise, when you specify @{b}-O@{ub}, member functions defined inside class scope are compiled inline by default; i.e., you don't need to add @{b}inline@{ub} in front of the member function name. @{b}-fno-defer-pop@{ub} Always pop the arguments to each function call as soon as that function returns. For machines which must pop arguments after a function call, the compiler normally lets arguments accumulate on the stack for several function calls and pops them all at once. @{b}-fforce-mem@{ub} Force memory operands to be copied into registers before doing arithmetic on them. This produces better code by making all memory references potential common subexpressions. When they are not common subexpressions, instruction combination should eliminate the separate register-load. The @{b}-O2@{ub} option turns on this option. @{b}-fforce-addr@{ub} Force memory address constants to be copied into registers before doing arithmetic on them. This may produce better code just as @{b}-fforce-mem@{ub} may. @{b}-fomit-frame-pointer@{ub} Don't keep the frame pointer in a register for functions that don't need one. This avoids the instructions to save, set up and restore frame pointers; it also makes an extra register available in many functions. @{i}It also makes debugging impossible on some machines.@{ui} On some machines, such as the Vax, this flag has no effect, because the standard calling sequence automatically handles the frame pointer and nothing is saved by pretending it doesn't exist. The machine-description macro @{b}FRAME_POINTER_REQUIRED@{ub} controls whether a target machine supports this flag. See @{"Registers" Link "Registers"}. @{b}-fno-inline@{ub} Don't pay attention to the @{b}inline@{ub} keyword. Normally this option is used to keep the compiler from expanding any functions inline. Note that if you are not optimizing, no functions can be expanded inline. @{b}-finline-functions@{ub} Integrate all simple functions into their callers. The compiler heuristically decides which functions are simple enough to be worth integrating in this way. If all calls to a given function are integrated, and the function is declared @{b}static@{ub}, then the function is normally not output as assembler code in its own right. @{b}-fkeep-inline-functions@{ub} Even if all calls to a given function are integrated, and the function is declared @{b}static@{ub}, nevertheless output a separate run-time callable version of the function. @{b}-fno-function-cse@{ub} Do not put function addresses in registers; make each instruction that calls a constant function contain the function's address explicitly. This option results in less efficient code, but some strange hacks that alter the assembler output may be confused by the optimizations performed when this option is not used. @{b}-ffast-math@{ub} This option allows GCC to violate some ANSI or IEEE rules and/or specifications in the interest of optimizing code for speed. For example, it allows the compiler to assume arguments to the @{b}sqrt@{ub} function are non-negative numbers and that no floating-point values are NaNs. This option should never be turned on by any @{b}-O@{ub} option since it can result in incorrect output for programs which depend on an exact implementation of IEEE or ANSI rules/specifications for math functions. The following options control specific optimizations. The @{b}-O2@{ub} option turns on all of these optimizations except @{b}-funroll-loops@{ub} and @{b}-funroll-all-loops@{ub}. On most machines, the @{b}-O@{ub} option turns on the @{b}-fthread-jumps@{ub} and @{b}-fdelayed-branch@{ub} options, but specific machines may handle it differently. You can use the following flags in the rare cases when "fine-tuning" of optimizations to be performed is desired. @{b}-fstrength-reduce@{ub} Perform the optimizations of loop strength reduction and elimination of iteration variables. @{b}-fthread-jumps@{ub} Perform optimizations where we check to see if a jump branches to a location where another comparison subsumed by the first is found. If so, the first branch is redirected to either the destination of the second branch or a point immediately following it, depending on whether the condition is known to be true or false. @{b}-fcse-follow-jumps@{ub} In common subexpression elimination, scan through jump instructions when the target of the jump is not reached by any other path. For example, when CSE encounters an @{b}if@{ub} statement with an @{b}else@{ub} clause, CSE will follow the jump when the condition tested is false. @{b}-fcse-skip-blocks@{ub} This is similar to @{b}-fcse-follow-jumps@{ub}, but causes CSE to follow jumps which conditionally skip over blocks. When CSE encounters a simple @{b}if@{ub} statement with no else clause, @{b}-fcse-skip-blocks@{ub} causes CSE to follow the jump around the body of the @{b}if@{ub}. @{b}-frerun-cse-after-loop@{ub} Re-run common subexpression elimination after loop optimizations has been performed. @{b}-fexpensive-optimizations@{ub} Perform a number of minor optimizations that are relatively expensive. @{b}-fdelayed-branch@{ub} If supported for the target machine, attempt to reorder instructions to exploit instruction slots available after delayed branch instructions. @{b}-fschedule-insns@{ub} If supported for the target machine, attempt to reorder instructions to eliminate execution stalls due to required data being unavailable. This helps machines that have slow floating point or memory load instructions by allowing other instructions to be issued until the result of the load or floating point instruction is required. @{b}-fschedule-insns2@{ub} Similar to @{b}-fschedule-insns@{ub}, but requests an additional pass of instruction scheduling after register allocation has been done. This is especially useful on machines with a relatively small number of registers and where memory load instructions take more than one cycle. @{b}-fcaller-saves@{ub} Enable values to be allocated in registers that will be clobbered by function calls, by emitting extra instructions to save and restore the registers around such calls. Such allocation is done only when it seems to result in better code than would otherwise be produced. This option is enabled by default on certain machines, usually those which have no call-preserved registers to use instead. @{b}-funroll-loops@{ub} Perform the optimization of loop unrolling. This is only done for loops whose number of iterations can be determined at compile time or run time. @{b}-funroll-loop@{ub} implies both @{b}-fstrength-reduce@{ub} and @{b}-frerun-cse-after-loop@{ub}. @{b}-funroll-all-loops@{ub} Perform the optimization of loop unrolling. This is done for all loops and usually makes programs run more slowly. @{b}-funroll-all-loops@{ub} implies @{b}-fstrength-reduce@{ub} as well as @{b}-frerun-cse-after-loop@{ub}. @{b}-fno-peephole@{ub} Disable any machine-specific peephole optimizations. @EndNode @Node "Preprocessor Options" "gcc.guide/Preprocessor Options" @Next "Assembler Options" @Prev "Optimize Options" @Toc "Invoking GCC" Options Controlling the Preprocessor ==================================== These options control the C preprocessor, which is run on each C source file before actual compilation. If you use the @{b}-E@{ub} option, nothing is done except preprocessing. Some of these options make sense only together with @{b}-E@{ub} because they cause the preprocessor output to be unsuitable for actual compilation. @{b}-include @{I}FILE@{ui}@{ub} Process @{I}FILE@{ui} as input before processing the regular input file. In effect, the contents of @{I}FILE@{ui} are compiled first. Any @{b}-D@{ub} and @{b}-U@{ub} options on the command line are always processed before @{b}-include @{I}FILE@{ui}@{ub}, regardless of the order in which they are written. All the @{b}-include@{ub} and @{b}-imacros@{ub} options are processed in the order in which they are written. @{b}-imacros @{I}FILE@{ui}@{ub} Process @{I}FILE@{ui} as input, discarding the resulting output, before processing the regular input file. Because the output generated from @{I}FILE@{ui} is discarded, the only effect of @{b}-imacros @{I}FILE@{ui}@{ub} is to make the macros defined in @{I}FILE@{ui} available for use in the main input. Any @{b}-D@{ub} and @{b}-U@{ub} options on the command line are always processed before @{b}-imacros @{I}FILE@{ui}@{ub}, regardless of the order in which they are written. All the @{b}-include@{ub} and @{b}-imacros@{ub} options are processed in the order in which they are written. @{b}-idirafter @{I}DIR@{ui}@{ub} Add the directory @{I}DIR@{ui} to the second include path. The directories on the second include path are searched when a header file is not found in any of the directories in the main include path (the one that @{b}-I@{ub} adds to). @{b}-iprefix @{I}PREFIX@{ui}@{ub} Specify @{I}PREFIX@{ui} as the prefix for subsequent @{b}-iwithprefix@{ub} options. @{b}-iwithprefix @{I}DIR@{ui}@{ub} Add a directory to the second include path. The directory's name is made by concatenating @{I}PREFIX@{ui} and @{I}DIR@{ui}, where @{I}PREFIX@{ui} was specified previously with @{b}-iprefix@{ub}. If you have not specified a prefix yet, the directory containing the installed passes of the compiler is used as the default. @{b}-iwithprefixbefore @{I}DIR@{ui}@{ub} Add a directory to the main include path. The directory's name is made by concatenating @{I}PREFIX@{ui} and @{I}DIR@{ui}, as in the case of @{b}-iwithprefix@{ub}. @{b}-isystem @{I}DIR@{ui}@{ub} Add a directory to the beginning of the second include path, marking it as a system directory, so that it gets the same special treatment as is applied to the standard system directories. @{b}-nostdinc@{ub} Do not search the standard system directories for header files. Only the directories you have specified with @{b}-I@{ub} options (and the current directory, if appropriate) are searched. See @{"Directory Options" Link "Directory Options"}, for information on @{b}-I@{ub}. By using both @{b}-nostdinc@{ub} and @{b}-I-@{ub}, you can limit the include-file search path to only those directories you specify explicitly. @{b}-undef@{ub} Do not predefine any nonstandard macros. (Including architecture flags). @{b}-E@{ub} Run only the C preprocessor. Preprocess all the C source files specified and output the results to standard output or to the specified output file. @{b}-C@{ub} Tell the preprocessor not to discard comments. Used with the @{b}-E@{ub} option. @{b}-P@{ub} Tell the preprocessor not to generate @{b}#line@{ub} directives. Used with the @{b}-E@{ub} option. @{b}-M@{ub} Tell the preprocessor to output a rule suitable for @{b}make@{ub} describing the dependencies of each object file. For each source file, the preprocessor outputs one @{b}make@{ub}-rule whose target is the object file name for that source file and whose dependencies are all the @{b}#include@{ub} header files it uses. This rule may be a single line or may be continued with @{b}\\ @{ub}-newline if it is long. The list of rules is printed on standard output instead of the preprocessed C program. @{b}-M@{ub} implies @{b}-E@{ub}. Another way to specify output of a @{b}make@{ub} rule is by setting the environment variable @{b}DEPENDENCIES_OUTPUT@{ub} (see @{"Environment Variables" Link "Environment Variables"}). @{b}-MM@{ub} Like @{b}-M@{ub} but the output mentions only the user header files included with @{b}#include "@{I}FILE@{ui}"@{ub}. System header files included with @{b}#include <@{I}FILE@{ui}>@{ub} are omitted. @{b}-MD@{ub} Like @{b}-M@{ub} but the dependency information is written to a file made by replacing ".c" with ".d" at the end of the input file names. This is in addition to compiling the file as specified--@{b}-MD@{ub} does not inhibit ordinary compilation the way @{b}-M@{ub} does. In Mach, you can use the utility @{b}md@{ub} to merge multiple dependency files into a single dependency file suitable for using with the @{b}make@{ub} command. @{b}-MMD@{ub} Like @{b}-MD@{ub} except mention only user header files, not system header files. @{b}-MG@{ub} Treat missing header files as generated files and assume they live in the same directory as the source file. If you specify @{b}-MG@{ub}, you must also specify either @{b}-M@{ub} or @{b}-MM@{ub}. @{b}-MG@{ub} is not supported with @{b}-MD@{ub} or @{b}-MMD@{ub}. @{b}-H@{ub} Print the name of each header file used, in addition to other normal activities. @{b}-A@{I}QUESTION@{ui}(@{I}ANSWER@{ui})@{ub} Assert the answer @{I}ANSWER@{ui} for @{I}QUESTION@{ui}, in case it is tested with a preprocessing conditional such as @{b}#if #@{I}QUESTION@{ui}(@{I}ANSWER@{ui})@{ub}. @{b}-A-@{ub} disables the standard assertions that normally describe the target machine. @{b}-D@{I}MACRO@{ui}@{ub} Define macro @{I}MACRO@{ui} with the string @{b}1@{ub} as its definition. @{b}-D@{I}MACRO@{ui}=@{I}DEFN@{ui}@{ub} Define macro @{I}MACRO@{ui} as @{I}DEFN@{ui}. All instances of @{b}-D@{ub} on the command line are processed before any @{b}-U@{ub} options. @{b}-U@{I}MACRO@{ui}@{ub} Undefine macro @{I}MACRO@{ui}. @{b}-U@{ub} options are evaluated after all @{b}-D@{ub} options, but before any @{b}-include@{ub} and @{b}-imacros@{ub} options. @{b}-dM@{ub} Tell the preprocessor to output only a list of the macro definitions that are in effect at the end of preprocessing. Used with the @{b}-E@{ub} option. @{b}-dD@{ub} Tell the preprocessing to pass all macro definitions into the output, in their proper sequence in the rest of the output. @{b}-dN@{ub} Like @{b}-dD@{ub} except that the macro arguments and contents are omitted. Only @{b}#define @{I}NAME@{ui}@{ub} is included in the output. @{b}-trigraphs@{ub} Support ANSI C trigraphs. The @{b}-ansi@{ub} option also has this effect. @{b}-Wp,@{I}OPTION@{ui}@{ub} Pass @{I}OPTION@{ui} as an option to the preprocessor. If @{I}OPTION@{ui} contains commas, it is split into multiple options at the commas. @EndNode @Node "Assembler Options" "gcc.guide/Assembler Options" @Next "Link Options" @Prev "Preprocessor Options" @Toc "Invoking GCC" Passing Options to the Assembler ================================ You can pass options to the assembler. @{b}-Wa,@{I}OPTION@{ui}@{ub} Pass @{I}OPTION@{ui} as an option to the assembler. If @{I}OPTION@{ui} contains commas, it is split into multiple options at the commas. @EndNode @Node "Link Options" "gcc.guide/Link Options" @Next "Directory Options" @Prev "Assembler Options" @Toc "Invoking GCC" Options for Linking =================== These options come into play when the compiler links object files into an executable output file. They are meaningless if the compiler is not doing a link step. @{b}@{I}OBJECT-FILE-NAME@{ui}@{ub} A file name that does not end in a special recognized suffix is considered to name an object file or library. (Object files are distinguished from libraries by the linker according to the file contents.) If linking is done, these object files are used as input to the linker. @{b}-c@{ub} @{b}-S@{ub} @{b}-E@{ub} If any of these options is used, then the linker is not run, and object file names should not be used as arguments. See @{"Overall Options" Link "Overall Options"}. @{b}-l@{I}LIBRARY@{ui}@{ub} Search the library named @{I}LIBRARY@{ui} when linking. It makes a difference where in the command you write this option; the linker searches processes libraries and object files in the order they are specified. Thus, @{b}foo.o -lz bar.o@{ub} searches library @{b}z@{ub} after file @{b}foo.o@{ub} but before @{b}bar.o@{ub}. If @{b}bar.o@{ub} refers to functions in @{b}z@{ub}, those functions may not be loaded. The linker searches a standard list of directories for the library, which is actually a file named @{b}lib@{I}LIBRARY@{ui}.a@{ub}. The linker then uses this file as if it had been specified precisely by name. The directories searched include several standard system directories plus any that you specify with @{b}-L@{ub}. Normally the files found this way are library files--archive files whose members are object files. The linker handles an archive file by scanning through it for members which define symbols that have so far been referenced but not defined. But if the file that is found is an ordinary object file, it is linked in the usual fashion. The only difference between using an @{b}-l@{ub} option and specifying a file name is that @{b}-l@{ub} surrounds @{I}LIBRARY@{ui} with @{b}lib@{ub} and @{b}.a@{ub} and searches several directories. @{b}-lobjc@{ub} You need this special case of the @{b}-l@{ub} option in order to link an Objective C program. @{b}-nostartfiles@{ub} Do not use the standard system startup files when linking. The standard system libraries are used normally, unless @{b}-nostdlib@{ub} or @{b}-nodefaultlibs@{ub} is used. @{b}-nodefaultlibs@{ub} Do not use the standard system libraries when linking. Only the libraries you specify will be passed to the linker. The standard startup files are used normally, unless @{b}-nostartfiles@{ub} is used. @{b}-nostdlib@{ub} Do not use the standard system startup files or libraries when linking. No startup files and only the libraries you specify will be passed to the linker. One of the standard libraries bypassed by @{b}-nostdlib@{ub} and @{b}-nodefaultlibs@{ub} is @{b}libgcc.a@{ub}, a library of internal subroutines that GNU CC uses to overcome shortcomings of particular machines, or special needs for some languages. (See @{"Interfacing to GNU CC Output" Link "Interface"}, for more discussion of @{b}libgcc.a@{ub}.) In most cases, you need @{b}libgcc.a@{ub} even when you want to avoid other standard libraries. In other words, when you specify @{b}-nostdlib@{ub} or @{b}-nodefaultlibs@{ub} you should usually specify @{b}-lgcc@{ub} as well. This ensures that you have no unresolved references to internal GNU CC library subroutines. (For example, @{b}__main@{ub}, used to ensure C++ constructors will be called; see @{"collect2" Link "Collect2"}.) @{b}-s@{ub} Remove all symbol table and relocation information from the executable. @{b}-static@{ub} On systems that support dynamic linking, this prevents linking with the shared libraries. On other systems, this option has no effect. @{b}-shared@{ub} Produce a shared object which can then be linked with other objects to form an executable. Only a few systems support this option. @{b}-symbolic@{ub} Bind references to global symbols when building a shared object. Warn about any unresolved references (unless overridden by the link editor option @{b}-Xlinker -z -Xlinker defs@{ub}). Only a few systems support this option. @{b}-Xlinker @{I}OPTION@{ui}@{ub} Pass @{I}OPTION@{ui} as an option to the linker. You can use this to supply system-specific linker options which GNU CC does not know how to recognize. If you want to pass an option that takes an argument, you must use @{b}-Xlinker@{ub} twice, once for the option and once for the argument. For example, to pass @{b}-assert definitions@{ub}, you must write @{b}-Xlinker -assert -Xlinker definitions@{ub}. It does not work to write @{b}-Xlinker "-assert definitions"@{ub}, because this passes the entire string as a single argument, which is not what the linker expects. @{b}-Wl,@{I}OPTION@{ui}@{ub} Pass @{I}OPTION@{ui} as an option to the linker. If @{I}OPTION@{ui} contains commas, it is split into multiple options at the commas. @{b}-u @{I}SYMBOL@{ui}@{ub} Pretend the symbol @{I}SYMBOL@{ui} is undefined, to force linking of library modules to define it. You can use @{b}-u@{ub} multiple times with different symbols to force loading of additional library modules. @EndNode @Node "Directory Options" "gcc.guide/Directory Options" @Next "Target Options" @Prev "Link Options" @Toc "Invoking GCC" Options for Directory Search ============================ These options specify directories to search for header files, for libraries and for parts of the compiler: @{b}-I@{I}DIR@{ui}@{ub} Add the directory @{I}DIRECTORY@{ui} to the head of the list of directories to be searched for header files. This can be used to override a system header file, substituting your own version, since these directories are searched before the system header file directories. If you use more than one @{b}-I@{ub} option, the directories are scanned in left-to-right order; the standard system directories come after. @{b}-I-@{ub} Any directories you specify with @{b}-I@{ub} options before the @{b}-I-@{ub} option are searched only for the case of @{b}#include "@{I}FILE@{ui}"@{ub}; they are not searched for @{b}#include <@{I}FILE@{ui}>@{ub}. If additional directories are specified with @{b}-I@{ub} options after the @{b}-I-@{ub}, these directories are searched for all @{b}#include@{ub} directives. (Ordinarily @{i}all@{ui} @{b}-I@{ub} directories are used this way.) In addition, the @{b}-I-@{ub} option inhibits the use of the current directory (where the current input file came from) as the first search directory for @{b}#include "@{I}FILE@{ui}"@{ub}. There is no way to override this effect of @{b}-I-@{ub}. With @{b}-I.@{ub} you can specify searching the directory which was current when the compiler was invoked. That is not exactly the same as what the preprocessor does by default, but it is often satisfactory. @{b}-I-@{ub} does not inhibit the use of the standard system directories for header files. Thus, @{b}-I-@{ub} and @{b}-nostdinc@{ub} are independent. @{b}-L@{I}DIR@{ui}@{ub} Add directory @{I}DIR@{ui} to the list of directories to be searched for @{b}-l@{ub}. @{b}-B@{I}PREFIX@{ui}@{ub} This option specifies where to find the executables, libraries, include files, and data files of the compiler itself. The compiler driver program runs one or more of the subprograms @{b}cpp@{ub}, @{b}cc1@{ub}, @{b}as@{ub} and @{b}ld@{ub}. It tries @{I}PREFIX@{ui} as a prefix for each program it tries to run, both with and without @{b}@{I}MACHINE@{ui}/@{I}VERSION@{ui}/@{ub} (see @{"Target Options" Link "Target Options"}). For each subprogram to be run, the compiler driver first tries the @{b}-B@{ub} prefix, if any. If that name is not found, or if @{b}-B@{ub} was not specified, the driver tries two standard prefixes, which are @{b}/usr/lib/gcc/@{ub} and @{b}/usr/local/lib/gcc-lib/@{ub}. If neither of those results in a file name that is found, the unmodified program name is searched for using the directories specified in your @{b}PATH@{ub} environment variable. @{b}-B@{ub} prefixes that effectively specify directory names also apply to libraries in the linker, because the compiler translates these options into @{b}-L@{ub} options for the linker. They also apply to includes files in the preprocessor, because the compiler translates these options into @{b}-isystem@{ub} options for the preprocessor. In this case, the compiler appends @{b}include@{ub} to the prefix. The run-time support file @{b}libgcc.a@{ub} can also be searched for using the @{b}-B@{ub} prefix, if needed. If it is not found there, the two standard prefixes above are tried, and that is all. The file is left out of the link if it is not found by those means. Another way to specify a prefix much like the @{b}-B@{ub} prefix is to use the environment variable @{b}GCC_EXEC_PREFIX@{ub}. See @{"Environment Variables" Link "Environment Variables"}. @EndNode @Node "Target Options" "gcc.guide/Target Options" @Next "Submodel Options" @Prev "Directory Options" @Toc "Invoking GCC" Specifying Target Machine and Compiler Version ============================================== By default, GNU CC compiles code for the same type of machine that you are using. However, it can also be installed as a cross-compiler, to compile for some other type of machine. In fact, several different configurations of GNU CC, for different target machines, can be installed side by side. Then you specify which one to use with the @{b}-b@{ub} option. In addition, older and newer versions of GNU CC can be installed side by side. One of them (probably the newest) will be the default, but you may sometimes wish to use another. @{b}-b @{I}MACHINE@{ui}@{ub} The argument @{I}MACHINE@{ui} specifies the target machine for compilation. This is useful when you have installed GNU CC as a cross-compiler. The value to use for @{I}MACHINE@{ui} is the same as was specified as the machine type when configuring GNU CC as a cross-compiler. For example, if a cross-compiler was configured with @{b}configure i386v@{ub}, meaning to compile for an 80386 running System V, then you would specify @{b}-b i386v@{ub} to run that cross compiler. When you do not specify @{b}-b@{ub}, it normally means to compile for the same type of machine that you are using. @{b}-V @{I}VERSION@{ui}@{ub} The argument @{I}VERSION@{ui} specifies which version of GNU CC to run. This is useful when multiple versions are installed. For example, @{I}VERSION@{ui} might be @{b}2.0@{ub}, meaning to run GNU CC version 2.0. The default version, when you do not specify @{b}-V@{ub}, is the last version of GNU CC that you installed. The @{b}-b@{ub} and @{b}-V@{ub} options actually work by controlling part of the file name used for the executable files and libraries used for compilation. A given version of GNU CC, for a given target machine, is normally kept in the directory @{b}/usr/local/lib/gcc-lib/@{I}MACHINE@{ui}/@{I}VERSION@{ui}@{ub}. Thus, sites can customize the effect of @{b}-b@{ub} or @{b}-V@{ub} either by changing the names of these directories or adding alternate names (or symbolic links). If in directory @{b}/usr/local/lib/gcc-lib/@{ub} the file @{b}80386@{ub} is a link to the file @{b}i386v@{ub}, then @{b}-b 80386@{ub} becomes an alias for @{b}-b i386v@{ub}. In one respect, the @{b}-b@{ub} or @{b}-V@{ub} do not completely change to a different compiler: the top-level driver program @{b}gcc@{ub} that you originally invoked continues to run and invoke the other executables (preprocessor, compiler per se, assembler and linker) that do the real work. However, since no real work is done in the driver program, it usually does not matter that the driver program in use is not the one for the specified target and version. The only way that the driver program depends on the target machine is in the parsing and handling of special machine-specific options. However, this is controlled by a file which is found, along with the other executables, in the directory for the specified version and target machine. As a result, a single installed driver program adapts to any specified target machine and compiler version. The driver program executable does control one significant thing, however: the default version and target machine. Therefore, you can install different instances of the driver program, compiled for different targets or versions, under different names. For example, if the driver for version 2.0 is installed as @{b}ogcc@{ub} and that for version 2.1 is installed as @{b}gcc@{ub}, then the command @{b}gcc@{ub} will use version 2.1 by default, while @{b}ogcc@{ub} will use 2.0 by default. However, you can choose either version with either command with the @{b}-V@{ub} option. @EndNode @Node "Submodel Options" "gcc.guide/Submodel Options" @Next "Code Gen Options" @Prev "Target Options" @Toc "Invoking GCC" Hardware Models and Configurations ================================== Earlier we discussed the standard option @{b}-b@{ub} which chooses among different installed compilers for completely different target machines, such as Vax vs. 68000 vs. 80386. In addition, each of these target machine types can have its own special options, starting with @{b}-m@{ub}, to choose among various hardware models or configurations--for example, 68010 vs 68020, floating coprocessor or none. A single installed version of the compiler can compile for any model or configuration, according to the options specified. Some configurations of the compiler also support additional special options, usually for compatibility with other compilers on the same platform. These options are defined by the macro @{b}TARGET_SWITCHES@{ub} in the machine description. The default for the options is also defined by that macro, which enables you to change the defaults. @{" M680x0 Options " Link "M680x0 Options"} @{" VAX Options " Link "VAX Options"} @{" SPARC Options " Link "SPARC Options"} @{" Convex Options " Link "Convex Options"} @{" AMD29K Options " Link "AMD29K Options"} @{" ARM Options " Link "ARM Options"} @{" M88K Options " Link "M88K Options"} @{" RS-6000 and PowerPC Options " Link "RS-6000 and PowerPC Options"} @{" RT Options " Link "RT Options"} @{" MIPS Options " Link "MIPS Options"} @{" i386 Options " Link "i386 Options"} @{" HPPA Options " Link "HPPA Options"} @{" Intel 960 Options " Link "Intel 960 Options"} @{" DEC Alpha Options " Link "DEC Alpha Options"} @{" Clipper Options " Link "Clipper Options"} @{" H8-300 Options " Link "H8-300 Options"} @{" System V Options " Link "System V Options"} @EndNode @Node "M680x0 Options" "gcc.guide/M680x0 Options" @Next "VAX Options" @Toc "Submodel Options" M680x0 Options -------------- These are the @{b}-m@{ub} options defined for the 68000 series. The default values for these options depends on which style of 68000 was selected when the compiler was configured; the defaults for the most common choices are given below. @{b}-m68000@{ub} @{b}-mc68000@{ub} Generate output for a 68000. This is the default when the compiler is configured for 68000-based systems. @{b}-m68020@{ub} @{b}-mc68020@{ub} Generate output for a 68020. This is the default when the compiler is configured for 68020-based systems. @{b}-m68881@{ub} Generate output containing 68881 instructions for floating point. This is the default for most 68020 systems unless @{b}-nfp@{ub} was specified when the compiler was configured. @{b}-m68030@{ub} Generate output for a 68030. This is the default when the compiler is configured for 68030-based systems. @{b}-m68040@{ub} Generate output for a 68040. This is the default when the compiler is configured for 68040-based systems. This option inhibits the use of 68881/68882 instructions that have to be emulated by software on the 68040. If your 68040 does not have code to emulate those instructions, use @{b}-m68040@{ub}. @{b}-m68020-40@{ub} Generate output for a 68040, without using any of the new instructions. This results in code which can run relatively efficiently on either a 68020/68881 or a 68030 or a 68040. The generated code does use the 68881 instructions that are emulated on the 68040. @{b}-mfpa@{ub} Generate output containing Sun FPA instructions for floating point. @{b}-msoft-float@{ub} Generate output containing library calls for floating point. @{i}Warning:@{ui} the requisite libraries are not available for all m68k targets. Normally the facilities of the machine's usual C compiler are used, but this can't be done directly in cross-compilation. You must make your own arrangements to provide suitable library functions for cross-compilation. The embedded targets @{b}m68k-*-aout@{ub} and @{b}m68k-*-coff@{ub} do provide software floating point support. @{b}-mshort@{ub} Consider type @{b}int@{ub} to be 16 bits wide, like @{b}short int@{ub}. @{b}-mnobitfield@{ub} Do not use the bit-field instructions. The @{b}-m68000@{ub} option implies @{b}-mnobitfield@{ub}. @{b}-mbitfield@{ub} Do use the bit-field instructions. The @{b}-m68020@{ub} option implies @{b}-mbitfield@{ub}. This is the default if you use a configuration designed for a 68020. @{b}-mrtd@{ub} Use a different function-calling convention, in which functions that take a fixed number of arguments return with the @{b}rtd@{ub} instruction, which pops their arguments while returning. This saves one instruction in the caller since there is no need to pop the arguments there. This calling convention is incompatible with the one normally used on Unix, so you cannot use it if you need to call libraries compiled with the Unix compiler. Also, you must provide function prototypes for all functions that take variable numbers of arguments (including @{b}printf@{ub}); otherwise incorrect code will be generated for calls to those functions. In addition, seriously incorrect code will result if you call a function with too many arguments. (Normally, extra arguments are harmlessly ignored.) The @{b}rtd@{ub} instruction is supported by the 68010 and 68020 processors, but not by the 68000. @EndNode @Node "VAX Options" "gcc.guide/VAX Options" @Next "SPARC Options" @Prev "M680x0 Options" @Toc "Submodel Options" VAX Options ----------- These @{b}-m@{ub} options are defined for the Vax: @{b}-munix@{ub} Do not output certain jump instructions (@{b}aobleq@{ub} and so on) that the Unix assembler for the Vax cannot handle across long ranges. @{b}-mgnu@{ub} Do output those jump instructions, on the assumption that you will assemble with the GNU assembler. @{b}-mg@{ub} Output code for g-format floating point numbers instead of d-format. @EndNode @Node "SPARC Options" "gcc.guide/SPARC Options" @Next "Convex Options" @Prev "VAX Options" @Toc "Submodel Options" SPARC Options ------------- These @{b}-m@{ub} switches are supported on the SPARC: @{b}-mno-app-regs@{ub} @{b}-mapp-regs@{ub} Specify @{b}-mapp-regs@{ub} to generate output using the global registers 2 through 4, which the SPARC SVR4 ABI reserves for applications. This is the default. To be fully SVR4 ABI compliant at the cost of some performance loss, specify @{b}-mno-app-regs@{ub}. You should compile libraries and system software with this option. @{b}-mfpu@{ub} @{b}-mhard-float@{ub} Generate output containing floating point instructions. This is the default. @{b}-mno-fpu@{ub} @{b}-msoft-float@{ub} Generate output containing library calls for floating point. @{i}Warning:@{ui} the requisite libraries are not available for all SPARC targets. Normally the facilities of the machine's usual C compiler are used, but this cannot be done directly in cross-compilation. You must make your own arrangements to provide suitable library functions for cross-compilation. The embedded targets @{b}sparc-*-aout@{ub} and @{b}sparclite-*-*@{ub} do provide software floating point support. @{b}-msoft-float@{ub} changes the calling convention in the output file; therefore, it is only useful if you compile @{i}all@{ui} of a program with this option. In particular, you need to compile @{b}libgcc.a@{ub}, the library that comes with GNU CC, with @{b}-msoft-float@{ub} in order for this to work. @{b}-mhard-quad-float@{ub} Generate output containing quad-word (long double) floating point instructions. @{b}-msoft-quad-float@{ub} Generate output containing library calls for quad-word (long double) floating point instructions. The functions called are those specified in the SPARC ABI. This is the default. As of this writing, there are no sparc implementations that have hardware support for the quad-word floating point instructions. They all invoke a trap handler for one of these instructions, and then the trap handler emulates the effect of the instruction. Because of the trap handler overhead, this is much slower than calling the ABI library routines. Thus the @{b}-msoft-quad-float@{ub} option is the default. @{b}-mno-epilogue@{ub} @{b}-mepilogue@{ub} With @{b}-mepilogue@{ub} (the default), the compiler always emits code for function exit at the end of each function. Any function exit in the middle of the function (such as a return statement in C) will generate a jump to the exit code at the end of the function. With @{b}-mno-epilogue@{ub}, the compiler tries to emit exit code inline at every function exit. @{b}-mno-flat@{ub} @{b}-mflat@{ub} With @{b}-mflat@{ub}, the compiler does not generate save/restore instructions and will use a "flat" or single register window calling convention. This model uses %i7 as the frame pointer and is compatible with the normal register window model. Code from either may be intermixed although debugger support is still incomplete. The local registers and the input registers (0-5) are still treated as "call saved" registers and will be saved on the stack as necessary. With @{b}-mno-flat@{ub} (the default), the compiler emits save/restore instructions (except for leaf functions) and is the normal mode of operation. @{b}-mno-unaligned-doubles@{ub} @{b}-munaligned-doubles@{ub} Assume that doubles have 8 byte alignment. This is the default. With @{b}-munaligned-doubles@{ub}, GNU CC assumes that doubles have 8 byte alignment only if they are contained in another type, or if they have an absolute address. Otherwise, it assumes they have 4 byte alignment. Specifying this option avoids some rare compatibility problems with code generated by other compilers. It is not the default because it results in a performance loss, especially for floating point code. @{b}-mv8@{ub} @{b}-msparclite@{ub} These two options select variations on the SPARC architecture. By default (unless specifically configured for the Fujitsu SPARClite), GCC generates code for the v7 variant of the SPARC architecture. @{b}-mv8@{ub} will give you SPARC v8 code. The only difference from v7 code is that the compiler emits the integer multiply and integer divide instructions which exist in SPARC v8 but not in SPARC v7. @{b}-msparclite@{ub} will give you SPARClite code. This adds the integer multiply, integer divide step and scan (@{b}ffs@{ub}) instructions which exist in SPARClite but not in SPARC v7. @{b}-mcypress@{ub} @{b}-msupersparc@{ub} These two options select the processor for which the code is optimised. With @{b}-mcypress@{ub} (the default), the compiler optimizes code for the Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series. This is also appropriate for the older SparcStation 1, 2, IPX etc. With @{b}-msupersparc@{ub} the compiler optimizes code for the SuperSparc cpu, as used in the SparcStation 10, 1000 and 2000 series. This flag also enables use of the full SPARC v8 instruction set. In a future version of GCC, these options will very likely be renamed to @{b}-mcpu=cypress@{ub} and @{b}-mcpu=supersparc@{ub}. These @{b}-m@{ub} switches are supported in addition to the above on SPARC V9 processors: @{b}-mmedlow@{ub} Generate code for the Medium/Low code model: assume a 32 bit address space. Programs are statically linked, PIC is not supported. Pointers are still 64 bits. It is very likely that a future version of GCC will rename this option. @{b}-mmedany@{ub} Generate code for the Medium/Anywhere code model: assume a 32 bit text segment starting at offset 0, and a 32 bit data segment starting anywhere (determined at link time). Programs are statically linked, PIC is not supported. Pointers are still 64 bits. It is very likely that a future version of GCC will rename this option. @{b}-mint64@{ub} Types long and int are 64 bits. @{b}-mlong32@{ub} Types long and int are 32 bits. @{b}-mlong64@{ub} @{b}-mint32@{ub} Type long is 64 bits, and type int is 32 bits. @{b}-mstack-bias@{ub} @{b}-mno-stack-bias@{ub} With @{b}-mstack-bias@{ub}, GNU CC assumes that the stack pointer, and frame pointer if present, are offset by -2047 which must be added back when making stack frame references. Otherwise, assume no such offset is present. @EndNode @Node "Convex Options" "gcc.guide/Convex Options" @Next "AMD29K Options" @Prev "SPARC Options" @Toc "Submodel Options" Convex Options -------------- These @{b}-m@{ub} options are defined for Convex: @{b}-mc1@{ub} Generate output for C1. The code will run on any Convex machine. The preprocessor symbol @{b}__convex__c1__@{ub} is defined. @{b}-mc2@{ub} Generate output for C2. Uses instructions not available on C1. Scheduling and other optimizations are chosen for max performance on C2. The preprocessor symbol @{b}__convex_c2__@{ub} is defined. @{b}-mc32@{ub} Generate output for C32xx. Uses instructions not available on C1. Scheduling and other optimizations are chosen for max performance on C32. The preprocessor symbol @{b}__convex_c32__@{ub} is defined. @{b}-mc34@{ub} Generate output for C34xx. Uses instructions not available on C1. Scheduling and other optimizations are chosen for max performance on C34. The preprocessor symbol @{b}__convex_c34__@{ub} is defined. @{b}-mc38@{ub} Generate output for C38xx. Uses instructions not available on C1. Scheduling and other optimizations are chosen for max performance on C38. The preprocessor symbol @{b}__convex_c38__@{ub} is defined. @{b}-margcount@{ub} Generate code which puts an argument count in the word preceding each argument list. This is compatible with regular CC, and a few programs may need the argument count word. GDB and other source-level debuggers do not need it; this info is in the symbol table. @{b}-mnoargcount@{ub} Omit the argument count word. This is the default. @{b}-mvolatile-cache@{ub} Allow volatile references to be cached. This is the default. @{b}-mvolatile-nocache@{ub} Volatile references bypass the data cache, going all the way to memory. This is only needed for multi-processor code that does not use standard synchronization instructions. Making non-volatile references to volatile locations will not necessarily work. @{b}-mlong32@{ub} Type long is 32 bits, the same as type int. This is the default. @{b}-mlong64@{ub} Type long is 64 bits, the same as type long long. This option is useless, because no library support exists for it. @EndNode @Node "AMD29K Options" "gcc.guide/AMD29K Options" @Next "ARM Options" @Prev "Convex Options" @Toc "Submodel Options" AMD29K Options -------------- These @{b}-m@{ub} options are defined for the AMD Am29000: @{b}-mdw@{ub} Generate code that assumes the @{b}DW@{ub} bit is set, i.e., that byte and halfword operations are directly supported by the hardware. This is the default. @{b}-mndw@{ub} Generate code that assumes the @{b}DW@{ub} bit is not set. @{b}-mbw@{ub} Generate code that assumes the system supports byte and halfword write operations. This is the default. @{b}-mnbw@{ub} Generate code that assumes the systems does not support byte and halfword write operations. @{b}-mnbw@{ub} implies @{b}-mndw@{ub}. @{b}-msmall@{ub} Use a small memory model that assumes that all function addresses are either within a single 256 KB segment or at an absolute address of less than 256k. This allows the @{b}call@{ub} instruction to be used instead of a @{b}const@{ub}, @{b}consth@{ub}, @{b}calli@{ub} sequence. @{b}-mnormal@{ub} Use the normal memory model: Generate @{b}call@{ub} instructions only when calling functions in the same file and @{b}calli@{ub} instructions otherwise. This works if each file occupies less than 256 KB but allows the entire executable to be larger than 256 KB. This is the default. @{b}-mlarge@{ub} Always use @{b}calli@{ub} instructions. Specify this option if you expect a single file to compile into more than 256 KB of code. @{b}-m29050@{ub} Generate code for the Am29050. @{b}-m29000@{ub} Generate code for the Am29000. This is the default. @{b}-mkernel-registers@{ub} Generate references to registers @{b}gr64-gr95@{ub} instead of to registers @{b}gr96-gr127@{ub}. This option can be used when compiling kernel code that wants a set of global registers disjoint from that used by user-mode code. Note that when this option is used, register names in @{b}-f@{ub} flags must use the normal, user-mode, names. @{b}-muser-registers@{ub} Use the normal set of global registers, @{b}gr96-gr127@{ub}. This is the default. @{b}-mstack-check@{ub} @{b}-mno-stack-check@{ub} Insert (or do not insert) a call to @{b}__msp_check@{ub} after each stack adjustment. This is often used for kernel code. @{b}-mstorem-bug@{ub} @{b}-mno-storem-bug@{ub} @{b}-mstorem-bug@{ub} handles 29k processors which cannot handle the separation of a mtsrim insn and a storem instruction (most 29000 chips to date, but not the 29050). @{b}-mno-reuse-arg-regs@{ub} @{b}-mreuse-arg-regs@{ub} @{b}-mno-reuse-arg-regs@{ub} tells the compiler to only use incoming argument registers for copying out arguments. This helps detect calling a function with fewer arguments than it was declared with. @{b}-msoft-float@{ub} Generate output containing library calls for floating point. @{i}Warning:@{ui} the requisite libraries are not part of GNU CC. Normally the facilities of the machine's usual C compiler are used, but this can't be done directly in cross-compilation. You must make your own arrangements to provide suitable library functions for cross-compilation. @EndNode @Node "ARM Options" "gcc.guide/ARM Options" @Next "M88K Options" @Prev "AMD29K Options" @Toc "Submodel Options" ARM Options ----------- These @{b}-m@{ub} options are defined for Advanced RISC Machines (ARM) architectures: @{b}-m2@{ub} @{b}-m3@{ub} These options are identical. Generate code for the ARM2 and ARM3 processors. This option is the default. You should also use this option to generate code for ARM6 processors that are running with a 26-bit program counter. @{b}-m6@{ub} Generate code for the ARM6 processor when running with a 32-bit program counter. @{b}-mapcs@{ub} Generate a stack frame that is compliant with the ARM Procedure Call Standard for all functions, even if this is not strictly necessary for correct execution of the code. @{b}-mbsd@{ub} This option only applies to RISC iX. Emulate the native BSD-mode compiler. This is the default if @{b}-ansi@{ub} is not specified. @{b}-mxopen@{ub} This option only applies to RISC iX. Emulate the native X/Open-mode compiler. @{b}-mno-symrename@{ub} This option only applies to RISC iX. Do not run the assembler post-processor, @{b}symrename@{ub}, after code has been assembled. Normally it is necessary to modify some of the standard symbols in preparation for linking with the RISC iX C library; this option suppresses this pass. The post-processor is never run when the compiler is built for cross-compilation. @EndNode @Node "M88K Options" "gcc.guide/M88K Options" @Next "RS-6000 and PowerPC Options" @Prev "ARM Options" @Toc "Submodel Options" M88K Options ------------ These @{b}-m@{ub} options are defined for Motorola 88k architectures: @{b}-m88000@{ub} Generate code that works well on both the m88100 and the m88110. @{b}-m88100@{ub} Generate code that works best for the m88100, but that also runs on the m88110. @{b}-m88110@{ub} Generate code that works best for the m88110, and may not run on the m88100. @{b}-mbig-pic@{ub} Obsolete option to be removed from the next revision. Use @{b}-fPIC@{ub}. @{b}-midentify-revision@{ub} Include an @{b}ident@{ub} directive in the assembler output recording the source file name, compiler name and version, timestamp, and compilation flags used. @{b}-mno-underscores@{ub} In assembler output, emit symbol names without adding an underscore character at the beginning of each name. The default is to use an underscore as prefix on each name. @{b}-mocs-debug-info@{ub} @{b}-mno-ocs-debug-info@{ub} Include (or omit) additional debugging information (about registers used in each stack frame) as specified in the 88open Object Compatibility Standard, "OCS". This extra information allows debugging of code that has had the frame pointer eliminated. The default for DG/UX, SVr4, and Delta 88 SVr3.2 is to include this information; other 88k configurations omit this information by default. @{b}-mocs-frame-position@{ub} When emitting COFF debugging information for automatic variables and parameters stored on the stack, use the offset from the canonical frame address, which is the stack pointer (register 31) on entry to the function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use @{b}-mocs-frame-position@{ub}; other 88k configurations have the default @{b}-mno-ocs-frame-position@{ub}. @{b}-mno-ocs-frame-position@{ub} When emitting COFF debugging information for automatic variables and parameters stored on the stack, use the offset from the frame pointer register (register 30). When this option is in effect, the frame pointer is not eliminated when debugging information is selected by the -g switch. @{b}-moptimize-arg-area@{ub} @{b}-mno-optimize-arg-area@{ub} Control how function arguments are stored in stack frames. @{b}-moptimize-arg-area@{ub} saves space by optimizing them, but this conflicts with the 88open specifications. The opposite alternative, @{b}-mno-optimize-arg-area@{ub}, agrees with 88open standards. By default GNU CC does not optimize the argument area. @{b}-mshort-data-@{I}NUM@{ui}@{ub} Generate smaller data references by making them relative to @{b}r0@{ub}, which allows loading a value using a single instruction (rather than the usual two). You control which data references are affected by specifying @{I}NUM@{ui} with this option. For example, if you specify @{b}-mshort-data-512@{ub}, then the data references affected are those involving displacements of less than 512 bytes. @{b}-mshort-data-@{I}NUM@{ui}@{ub} is not effective for @{I}NUM@{ui} greater than 64k. @{b}-mserialize-volatile@{ub} @{b}-mno-serialize-volatile@{ub} Do, or don't, generate code to guarantee sequential consistency of volatile memory references. By default, consistency is guaranteed. The order of memory references made by the MC88110 processor does not always match the order of the instructions requesting those references. In particular, a load instruction may execute before a preceding store instruction. Such reordering violates sequential consistency of volatile memory references, when there are multiple processors. When consistency must be guaranteed, GNU C generates special instructions, as needed, to force execution in the proper order. The MC88100 processor does not reorder memory references and so always provides sequential consistency. However, by default, GNU C generates the special instructions to guarantee consistency even when you use @{b}-m88100@{ub}, so that the code may be run on an MC88110 processor. If you intend to run your code only on the MC88100 processor, you may use @{b}-mno-serialize-volatile@{ub}. The extra code generated to guarantee consistency may affect the performance of your application. If you know that you can safely forgo this guarantee, you may use @{b}-mno-serialize-volatile@{ub}. @{b}-msvr4@{ub} @{b}-msvr3@{ub} Turn on (@{b}-msvr4@{ub}) or off (@{b}-msvr3@{ub}) compiler extensions related to System V release 4 (SVr4). This controls the following: 1. Which variant of the assembler syntax to emit. 2. @{b}-msvr4@{ub} makes the C preprocessor recognize @{b}#pragma weak@{ub} that is used on System V release 4. 3. @{b}-msvr4@{ub} makes GNU CC issue additional declaration directives used in SVr4. @{b}-msvr4@{ub} is the default for the m88k-motorola-sysv4 and m88k-dg-dgux m88k configurations. @{b}-msvr3@{ub} is the default for all other m88k configurations. @{b}-mversion-03.00@{ub} This option is obsolete, and is ignored. @{b}-mno-check-zero-division@{ub} @{b}-mcheck-zero-division@{ub} Do, or don't, generate code to guarantee that integer division by zero will be detected. By default, detection is guaranteed. Some models of the MC88100 processor fail to trap upon integer division by zero under certain conditions. By default, when compiling code that might be run on such a processor, GNU C generates code that explicitly checks for zero-valued divisors and traps with exception number 503 when one is detected. Use of mno-check-zero-division suppresses such checking for code generated to run on an MC88100 processor. GNU C assumes that the MC88110 processor correctly detects all instances of integer division by zero. When @{b}-m88110@{ub} is specified, both @{b}-mcheck-zero-division@{ub} and @{b}-mno-check-zero-division@{ub} are ignored, and no explicit checks for zero-valued divisors are generated. @{b}-muse-div-instruction@{ub} Use the div instruction for signed integer division on the MC88100 processor. By default, the div instruction is not used. On the MC88100 processor the signed integer division instruction div) traps to the operating system on a negative operand. The operating system transparently completes the operation, but at a large cost in execution time. By default, when compiling code that might be run on an MC88100 processor, GNU C emulates signed integer division using the unsigned integer division instruction divu), thereby avoiding the large penalty of a trap to the operating system. Such emulation has its own, smaller, execution cost in both time and space. To the extent that your code's important signed integer division operations are performed on two nonnegative operands, it may be desirable to use the div instruction directly. On the MC88110 processor the div instruction (also known as the divs instruction) processes negative operands without trapping to the operating system. When @{b}-m88110@{ub} is specified, @{b}-muse-div-instruction@{ub} is ignored, and the div instruction is used for signed integer division. Note that the result of dividing INT_MIN by -1 is undefined. In particular, the behavior of such a division with and without @{b}-muse-div-instruction@{ub} may differ. @{b}-mtrap-large-shift@{ub} @{b}-mhandle-large-shift@{ub} Include code to detect bit-shifts of more than 31 bits; respectively, trap such shifts or emit code to handle them properly. By default GNU CC makes no special provision for large bit shifts. @{b}-mwarn-passed-structs@{ub} Warn when a function passes a struct as an argument or result. Structure-passing conventions have changed during the evolution of the C language, and are often the source of portability problems. By default, GNU CC issues no such warning. @EndNode @Node "RS-6000 and PowerPC Options" "gcc.guide/RS-6000 and PowerPC Options" @Next "RT Options" @Prev "M88K Options" @Toc "Submodel Options" IBM RS/6000 and PowerPC Options ------------------------------- These @{b}-m@{ub} options are defined for the IBM RS/6000 and PowerPC: @{b}-mpower@{ub} @{b}-mno-power@{ub} @{b}-mpower2@{ub} @{b}-mno-power2@{ub} @{b}-mpowerpc@{ub} @{b}-mno-powerpc@{ub} @{b}-mpowerpc-gpopt@{ub} @{b}-mno-powerpc-gpopt@{ub} @{b}-mpowerpc-gfxopt@{ub} @{b}-mno-powerpc-gfxopt@{ub} GNU CC supports two related instruction set architectures for the RS/6000 and PowerPC. The @{i}POWER@{ui} instruction set are those instructions supported by the @{b}rios@{ub} chip set used in the original RS/6000 systems and the @{i}PowerPC@{ui} instruction set is the architecture of the Motorola MPC6xx microprocessors. The PowerPC architecture defines 64-bit instructions, but they are not supported by any current processors. Neither architecture is a subset of the other. However there is a large common subset of instructions supported by both. An MQ register is included in processors supporting the POWER architecture. You use these options to specify which instructions are available on the processor you are using. The default value of these options is determined when configuring GNU CC. Specifying the @{b}-mcpu=@{I}CPU_TYPE@{ui}@{ub} overrides the specification of these options. We recommend you use that option rather than these. The @{b}-mpower@{ub} option allows GNU CC to generate instructions that are found only in the POWER architecture and to use the MQ register. Specifying @{b}-mpower2@{ub} implies @{b}-power@{ub} and also allows GNU CC to generate instructions that are present in the POWER2 architecture but not the original POWER architecture. The @{b}-mpowerpc@{ub} option allows GNU CC to generate instructions that are found only in the 32-bit subset of the PowerPC architecture. Specifying @{b}-mpowerpc-gpopt@{ub} implies @{b}-mpowerpc@{ub} and also allows GNU CC to use the optional PowerPC architecture instructions in the General Purpose group, including floating-point square root. Specifying @{b}-mpowerpc-gfxopt@{ub} implies @{b}-mpowerpc@{ub} and also allows GNU CC to use the optional PowerPC architecture instructions in the Graphics group, including floating-point select. If you specify both @{b}-mno-power@{ub} and @{b}-mno-powerpc@{ub}, GNU CC will use only the instructions in the common subset of both architectures plus some special AIX common-mode calls, and will not use the MQ register. Specifying both @{b}-mpower@{ub} and @{b}-mpowerpc@{ub} permits GNU CC to use any instruction from either architecture and to allow use of the MQ register; specify this for the Motorola MPC601. @{b}-mnew-mnemonics@{ub} @{b}-mold-mnemonics@{ub} Select which mnemonics to use in the generated assembler code. @{b}-mnew-mnemonics@{ub} requests output that uses the assembler mnemonics defined for the PowerPC architecture, while @{b}-mold-mnemonics@{ub} requests the assembler mnemonics defined for the POWER architecture. Instructions defined in only one architecture have only one mnemonic; GNU CC uses that mnemonic irrespective of which of these options is specified. PowerPC assemblers support both the old and new mnemonics, as will later POWER assemblers. Current POWER assemblers only support the old mnemonics. Specify @{b}-mnew-mnemonics@{ub} if you have an assembler that supports them, otherwise specify @{b}-mold-mnemonics@{ub}. The default value of these options depends on how GNU CC was configured. Specifying @{b}-mcpu=@{I}CPU_TYPE@{ui}@{ub} sometimes overrides the value of these option. Unless you are building a cross-compiler, you should normally not specify either @{b}-mnew-mnemonics@{ub} or @{b}-mold-mnemonics@{ub}, but should instead accept the default. @{b}-mcpu=@{I}CPU_TYPE@{ui}@{ub} Set architecture type, register usage, choice of mnemonics, and instruction scheduling parameters for machine type @{I}CPU_TYPE@{ui}. By default, @{I}CPU_TYPE@{ui} is the target system defined when GNU CC was configured. Supported values for @{I}CPU_TYPE@{ui} are @{b}rios1@{ub}, @{b}rios2@{ub}, @{b}rsc@{ub}, @{b}601@{ub}, @{b}603@{ub}, @{b}604@{ub}, @{b}power@{ub}, @{b}powerpc@{ub}, @{b}403@{ub}, and @{b}common@{ub}. @{b}-mcpu=power@{ub} and @{b}-mcpu=powerpc@{ub} specify generic POWER and pure PowerPC (i.e., not MPC601) architecture machine types, with an appropriate, generic processor model assumed for scheduling purposes. Specifying @{b}-mcpu=rios1@{ub}, @{b}-mcpu=rios2@{ub}, @{b}-mcpu=rsc@{ub}, or @{b}-mcpu=power@{ub} enables the @{b}-mpower@{ub} option and disables the @{b}-mpowerpc@{ub} option; @{b}-mcpu=601@{ub} enables both the @{b}-mpower@{ub} and @{b}-mpowerpc@{ub} options; @{b}-mcpu=603@{ub}, @{b}-mcpu=604@{ub}, @{b}-mcpu=403@{ub}, and @{b}-mcpu=powerpc@{ub} enable the @{b}-mpowerpc@{ub} option and disable the @{b}-mpower@{ub} option; @{b}-mcpu=common@{ub} disables both the @{b}-mpower@{ub} and @{b}-mpowerpc@{ub} options. To generate code that will operate on all members of the RS/6000 and PowerPC families, specify @{b}-mcpu=common@{ub}. In that case, GNU CC will use only the instructions in the common subset of both architectures plus some special AIX common-mode calls, and will not use the MQ register. GNU CC assumes a generic processor model for scheduling purposes. Specifying @{b}-mcpu=rios1@{ub}, @{b}-mcpu=rios2@{ub}, @{b}-mcpu=rsc@{ub}, or @{b}-mcpu=power@{ub} also disables the @{b}new-mnemonics@{ub} option. Specifying @{b}-mcpu=601@{ub}, @{b}-mcpu=603@{ub}, @{b}-mcpu=604@{ub}, @{b}403@{ub}, or @{b}-mcpu=powerpc@{ub} also enables the @{b}new-mnemonics@{ub} option. @{b}-mfull-toc@{ub} @{b}-mno-fp-in-toc@{ub} @{b}-mno-sum-in-toc@{ub} @{b}-mminimal-toc@{ub} Modify generation of the TOC (Table Of Contents), which is created for every executable file. The @{b}-mfull-toc@{ub} option is selected by default. In that case, GNU CC will allocate at least one TOC entry for each unique non-automatic variable reference in your program. GNU CC will also place floating-point constants in the TOC. However, only 16,384 entries are available in the TOC. If you receive a linker error message that saying you have overflowed the available TOC space, you can reduce the amount of TOC space used with the @{b}-mno-fp-in-toc@{ub} and @{b}-mno-sum-in-toc@{ub} options. @{b}-mno-fp-in-toc@{ub} prevents GNU CC from putting floating-point constants in the TOC and @{b}-mno-sum-in-toc@{ub} forces GNU CC to generate code to calculate the sum of an address and a constant at run-time instead of putting that sum into the TOC. You may specify one or both of these options. Each causes GNU CC to produce very slightly slower and larger code at the expense of conserving TOC space. If you still run out of space in the TOC even when you specify both of these options, specify @{b}-mminimal-toc@{ub} instead. This option causes GNU CC to make only one TOC entry for every file. When you specify this option, GNU CC will produce code that is slower and larger but which uses extremely little TOC space. You may wish to use this option only on files that contain less frequently executed code. @{b}-msoft-float@{ub} @{b}-mhard-float@{ub} Generate code that does not use (uses) the floating-point register set. Software floating point emulation is provided if you use the @{b}-msoft-float@{ub} option, and pass the option to GNU CC when linking. @{b}-mmultiple@{ub} @{b}-mno-multiple@{ub} Generate code that uses (does not use) the load multiple word instructions and the store multiple word instructions. These instructions are generated by default on POWER systems, and not generated on PowerPC systems. Do not use @{b}-mmultiple@{ub} on little endian PowerPC systems, since those instructions do not work when the processor is in little endian mode. @{b}-mstring@{ub} @{b}-mno-string@{ub} Generate code that uses (does not use) the load string instructions and the store string word instructions to save multiple registers and do small block moves. These instructions are generated by default on POWER systems, anod not generated on PowerPC systems. Do not use @{b}-mstring@{ub} on little endian PowerPC systems, since those instructions do not work when the processor is in little endian mode. @{b}-mno-bit-align@{ub} @{b}-mbit-align@{ub} On System V.4 and embedded PowerPC systems do not (do) force structures and unions that contain bit fields to be aligned to the base type of the bit field. For example, by default a structure containing nothing but 8 @{b}unsigned@{ub} bitfields of length 1 would be aligned to a 4 byte boundary and have a size of 4 bytes. By using @{b}-mno-bit-align@{ub}, the structure would be aligned to a 1 byte boundary and be one byte in size. @{b}-mno-strict-align@{ub} @{b}-mstrict-align@{ub} On System V.4 and embedded PowerPC systems do not (do) assume that unaligned memory references will be handled by the system. @{b}-mrelocatable@{ub} @{b}-mno-relocatable@{ub} On embedded PowerPC systems generate code that allows (does not allow) the program to be relocated to a different address at runtime. @{b}-mno-toc@{ub} @{b}-mtoc@{ub} On System V.4 and embedded PowerPC systems do not (do) assume that register 2 contains a pointer to a global area pointing to the addresses used in the program. @{b}-mno-traceback@{ub} @{b}-mtraceback@{ub} On embedded PowerPC systems do not (do) generate a traceback tag before the start of the function. This tag can be used by the debugger to identify where the start of a function is. @{b}-mlittle@{ub} @{b}-mlittle-endian@{ub} On System V.4 and embedded PowerPC systems compile code for the processor in little endian mode. The @{b}-mlittle-endian@{ub} option is the same as @{b}-mlittle@{ub}. @{b}-mbig@{ub} @{b}-mbig-endian@{ub} On System V.4 and embedded PowerPC systems compile code for the processor in big endian mode. The @{b}-mbig-endian@{ub} option is the same as @{b}-mbig@{ub}. @{b}-mcall-sysv@{ub} On System V.4 and embedded PowerPC systems compile code using calling conventions that adheres to the March 1995 draft of the System V Application Binary Interface, PowerPC processor supplement. This is the default unless you configured GCC using @{b}powerpc-*-eabiaix@{ub}. @{b}-mcall-aix@{ub} On System V.4 and embedded PowerPC systems compile code using calling conventions that are similar to those used on AIX. This is the default if you configured GCC using @{b}powerpc-*-eabiaix@{ub}. @{b}-mprototype@{ub} @{b}-mno-prototype@{ub} On System V.4 and embedded PowerPC systems assume that all calls to variable argument functions are properly prototyped. Otherwise, the compiler must insert an instruction before every non prototyped call to set or clear bit 6 of the condition code register (@{I}CR@{ui}) to indicate whether floating point values were passed in the floating point registers in case the function takes a variable arguments. With @{b}-mprototype@{ub}, only calls to prototyped variable argument functions will set or clear the bit. @EndNode @Node "RT Options" "gcc.guide/RT Options" @Next "MIPS Options" @Prev "RS-6000 and PowerPC Options" @Toc "Submodel Options" IBM RT Options -------------- These @{b}-m@{ub} options are defined for the IBM RT PC: @{b}-min-line-mul@{ub} Use an in-line code sequence for integer multiplies. This is the default. @{b}-mcall-lib-mul@{ub} Call @{b}lmul$$@{ub} for integer multiples. @{b}-mfull-fp-blocks@{ub} Generate full-size floating point data blocks, including the minimum amount of scratch space recommended by IBM. This is the default. @{b}-mminimum-fp-blocks@{ub} Do not include extra scratch space in floating point data blocks. This results in smaller code, but slower execution, since scratch space must be allocated dynamically. @{b}-mfp-arg-in-fpregs@{ub} Use a calling sequence incompatible with the IBM calling convention in which floating point arguments are passed in floating point registers. Note that @{b}varargs.h@{ub} and @{b}stdargs.h@{ub} will not work with floating point operands if this option is specified. @{b}-mfp-arg-in-gregs@{ub} Use the normal calling convention for floating point arguments. This is the default. @{b}-mhc-struct-return@{ub} Return structures of more than one word in memory, rather than in a register. This provides compatibility with the MetaWare HighC (hc) compiler. Use the option @{b}-fpcc-struct-return@{ub} for compatibility with the Portable C Compiler (pcc). @{b}-mnohc-struct-return@{ub} Return some structures of more than one word in registers, when convenient. This is the default. For compatibility with the IBM-supplied compilers, use the option @{b}-fpcc-struct-return@{ub} or the option @{b}-mhc-struct-return@{ub}. @EndNode @Node "MIPS Options" "gcc.guide/MIPS Options" @Next "i386 Options" @Prev "RT Options" @Toc "Submodel Options" MIPS Options ------------ These @{b}-m@{ub} options are defined for the MIPS family of computers: @{b}-mcpu=@{I}CPU TYPE@{ui}@{ub} Assume the defaults for the machine type @{I}CPU TYPE@{ui} when scheduling instructions. The choices for @{I}CPU TYPE@{ui} are @{b}r2000@{ub}, @{b}r3000@{ub}, @{b}r4000@{ub}, @{b}r4400@{ub}, @{b}r4600@{ub}, and @{b}r6000@{ub}. While picking a specific @{I}CPU TYPE@{ui} will schedule things appropriately for that particular chip, the compiler will not generate any code that does not meet level 1 of the MIPS ISA (instruction set architecture) without the @{b}-mips2@{ub} or @{b}-mips3@{ub} switches being used. @{b}-mips1@{ub} Issue instructions from level 1 of the MIPS ISA. This is the default. @{b}r3000@{ub} is the default @{I}CPU TYPE@{ui} at this ISA level. @{b}-mips2@{ub} Issue instructions from level 2 of the MIPS ISA (branch likely, square root instructions). @{b}r6000@{ub} is the default @{I}CPU TYPE@{ui} at this ISA level. @{b}-mips3@{ub} Issue instructions from level 3 of the MIPS ISA (64 bit instructions). @{b}r4000@{ub} is the default @{I}CPU TYPE@{ui} at this ISA level. This option does not change the sizes of any of the C data types. @{b}-mfp32@{ub} Assume that 32 32-bit floating point registers are available. This is the default. @{b}-mfp64@{ub} Assume that 32 64-bit floating point registers are available. This is the default when the @{b}-mips3@{ub} option is used. @{b}-mgp32@{ub} Assume that 32 32-bit general purpose registers are available. This is the default. @{b}-mgp64@{ub} Assume that 32 64-bit general purpose registers are available. This is the default when the @{b}-mips3@{ub} option is used. @{b}-mint64@{ub} Types long, int, and pointer are 64 bits. This works only if @{b}-mips3@{ub} is also specified. @{b}-mlong64@{ub} Types long and pointer are 64 bits, and type int is 32 bits. This works only if @{b}-mips3@{ub} is also specified. @{b}-mmips-as@{ub} Generate code for the MIPS assembler, and invoke @{b}mips-tfile@{ub} to add normal debug information. This is the default for all platforms except for the OSF/1 reference platform, using the OSF/rose object format. If the either of the @{b}-gstabs@{ub} or @{b}-gstabs+@{ub} switches are used, the @{b}mips-tfile@{ub} program will encapsulate the stabs within MIPS ECOFF. @{b}-mgas@{ub} Generate code for the GNU assembler. This is the default on the OSF/1 reference platform, using the OSF/rose object format. @{b}-mrnames@{ub} @{b}-mno-rnames@{ub} The @{b}-mrnames@{ub} switch says to output code using the MIPS software names for the registers, instead of the hardware names (ie, @{I}A0@{ui} instead of @{I}$4@{ui}). The only known assembler that supports this option is the Algorithmics assembler. @{b}-mgpopt@{ub} @{b}-mno-gpopt@{ub} The @{b}-mgpopt@{ub} switch says to write all of the data declarations before the instructions in the text section, this allows the MIPS assembler to generate one word memory references instead of using two words for short global or static data items. This is on by default if optimization is selected. @{b}-mstats@{ub} @{b}-mno-stats@{ub} For each non-inline function processed, the @{b}-mstats@{ub} switch causes the compiler to emit one line to the standard error file to print statistics about the program (number of registers saved, stack size, etc.). @{b}-mmemcpy@{ub} @{b}-mno-memcpy@{ub} The @{b}-mmemcpy@{ub} switch makes all block moves call the appropriate string function (@{b}memcpy@{ub} or @{b}bcopy@{ub}) instead of possibly generating inline code. @{b}-mmips-tfile@{ub} @{b}-mno-mips-tfile@{ub} The @{b}-mno-mips-tfile@{ub} switch causes the compiler not postprocess the object file with the @{b}mips-tfile@{ub} program, after the MIPS assembler has generated it to add debug support. If @{b}mips-tfile@{ub} is not run, then no local variables will be available to the debugger. In addition, @{b}stage2@{ub} and @{b}stage3@{ub} objects will have the temporary file names passed to the assembler embedded in the object file, which means the objects will not compare the same. The @{b}-mno-mips-tfile@{ub} switch should only be used when there are bugs in the @{b}mips-tfile@{ub} program that prevents compilation. @{b}-msoft-float@{ub} Generate output containing library calls for floating point. @{i}Warning:@{ui} the requisite libraries are not part of GNU CC. Normally the facilities of the machine's usual C compiler are used, but this can't be done directly in cross-compilation. You must make your own arrangements to provide suitable library functions for cross-compilation. @{b}-mhard-float@{ub} Generate output containing floating point instructions. This is the default if you use the unmodified sources. @{b}-mabicalls@{ub} @{b}-mno-abicalls@{ub} Emit (or do not emit) the pseudo operations @{b}.abicalls@{ub}, @{b}.cpload@{ub}, and @{b}.cprestore@{ub} that some System V.4 ports use for position independent code. @{b}-mlong-calls@{ub} @{b}-mno-long-calls@{ub} Do all calls with the @{b}JALR@{ub} instruction, which requires loading up a function's address into a register before the call. You need to use this switch, if you call outside of the current 512 megabyte segment to functions that are not through pointers. @{b}-mhalf-pic@{ub} @{b}-mno-half-pic@{ub} Put pointers to extern references into the data section and load them up, rather than put the references in the text section. @{b}-membedded-pic@{ub} @{b}-mno-embedded-pic@{ub} Generate PIC code suitable for some embedded systems. All calls are made using PC relative address, and all data is addressed using the $gp register. This requires GNU as and GNU ld which do most of the work. @{b}-membedded-data@{ub} @{b}-mno-embedded-data@{ub} Allocate variables to the read-only data section first if possible, then next in the small data section if possible, otherwise in data. This gives slightly slower code than the default, but reduces the amount of RAM required when executing, and thus may be preferred for some embedded systems. @{b}-msingle-float@{ub} @{b}-mdouble-float@{ub} The @{b}-msingle-float@{ub} switch tells gcc to assume that the floating point coprocessor only supports single precision operations, as on the @{b}r4650@{ub} chip. The @{b}-mdouble-float@{ub} switch permits gcc to use double precision operations. This is the default. @{b}-mmad@{ub} @{b}-mno-mad@{ub} Permit use of the @{b}mad@{ub}, @{b}madu@{ub} and @{b}mul@{ub} instructions, as on the @{b}r4650@{ub} chip. @{b}-m4650@{ub} Turns on @{b}-msingle-float@{ub}, @{b}-mmad@{ub}, and, at least for now, @{b}-mcpu=r4650@{ub}. @{b}-EL@{ub} Compile code for the processor in little endian mode. The requisite libraries are assumed to exist. @{b}-EB@{ub} Compile code for the processor in big endian mode. The requisite libraries are assumed to exist. @{b}-G @{I}NUM@{ui}@{ub} Put global and static items less than or equal to @{I}NUM@{ui} bytes into the small data or bss sections instead of the normal data or bss section. This allows the assembler to emit one word memory reference instructions based on the global pointer (@{I}GP@{ui} or @{I}$28@{ui}), instead of the normal two words used. By default, @{I}NUM@{ui} is 8 when the MIPS assembler is used, and 0 when the GNU assembler is used. The @{b}-G @{I}NUM@{ui}@{ub} switch is also passed to the assembler and linker. All modules should be compiled with the same @{b}-G @{I}NUM@{ui}@{ub} value. @{b}-nocpp@{ub} Tell the MIPS assembler to not run it's preprocessor over user assembler files (with a @{b}.s@{ub} suffix) when assembling them. These options are defined by the macro @{b}TARGET_SWITCHES@{ub} in the machine description. The default for the options is also defined by that macro, which enables you to change the defaults. @EndNode @Node "i386 Options" "gcc.guide/i386 Options" @Next "HPPA Options" @Prev "MIPS Options" @Toc "Submodel Options" Intel 386 Options ----------------- These @{b}-m@{ub} options are defined for the i386 family of computers: @{b}-m486@{ub} @{b}-m386@{ub} Control whether or not code is optimized for a 486 instead of an 386. Code generated for an 486 will run on a 386 and vice versa. @{b}-mieee-fp@{ub} @{b}-mno-ieee-fp@{ub} Control whether or not the compiler uses IEEE floating point comparisons. These handle correctly the case where the result of a comparison is unordered. @{b}-msoft-float@{ub} Generate output containing library calls for floating point. @{i}Warning:@{ui} the requisite libraries are not part of GNU CC. Normally the facilities of the machine's usual C compiler are used, but this can't be done directly in cross-compilation. You must make your own arrangements to provide suitable library functions for cross-compilation. On machines where a function returns floating point results in the 80387 register stack, some floating point opcodes may be emitted even if @{b}-msoft-float@{ub} is used. @{b}-mno-fp-ret-in-387@{ub} Do not use the FPU registers for return values of functions. The usual calling convention has functions return values of types @{b}float@{ub} and @{b}double@{ub} in an FPU register, even if there is no FPU. The idea is that the operating system should emulate an FPU. The option @{b}-mno-fp-ret-in-387@{ub} causes such values to be returned in ordinary CPU registers instead. @{b}-mno-fancy-math-387@{ub} Some 387 emulators do not support the @{b}sin@{ub}, @{b}cos@{ub} and @{b}sqrt@{ub} instructions for the 387. Specify this option to avoid generating those instructions. This option is the default on FreeBSD. As of revision 2.6.1, these instructions are not generated unless you also use the @{b}-ffast-math@{ub} switch. @{b}-malign-double@{ub} @{b}-mno-align-double@{ub} Control whether GNU CC aligns @{b}double@{ub}, @{b}long double@{ub}, and @{b}long long@{ub} variables on a two word boundary or a one word boundary. Aligning @{b}double@{ub} variables on a two word boundary will produce code that runs somewhat faster on a @{b}Pentium@{ub} at the expense of more memory. @{i}Warning:@{ui} if you use the @{b}-malign-double@{ub} switch, structures containing the above types will be aligned differently than the published application binary interface specifications for the 386. @{b}-msvr3-shlib@{ub} @{b}-mno-svr3-shlib@{ub} Control whether GNU CC places uninitialized locals into @{b}bss@{ub} or @{b}data@{ub}. @{b}-msvr3-shlib@{ub} places these locals into @{b}bss@{ub}. These options are meaningful only on System V Release 3. @{b}-mno-wide-multiply@{ub} @{b}-mwide-multiply@{ub} Control whether GNU CC uses the @{b}mul@{ub} and @{b}imul@{ub} that produce 64 bit results in @{b}eax:edx@{ub} from 32 bit operands to do @{b}long long@{ub} multiplies and 32-bit division by constants. @{b}-mrtd@{ub} Use a different function-calling convention, in which functions that take a fixed number of arguments return with the @{b}ret@{ub} @{I}NUM@{ui} instruction, which pops their arguments while returning. This saves one instruction in the caller since there is no need to pop the arguments there. You can specify that an individual function is called with this calling sequence with the function attribute @{b}stdcall@{ub}. You can also override the @{b}-mrtd@{ub} option by using the function attribute @{b}cdecl@{ub}. See @{"Function Attributes" Link "Function Attributes"} @{i}Warning:@{ui} this calling convention is incompatible with the one normally used on Unix, so you cannot use it if you need to call libraries compiled with the Unix compiler. Also, you must provide function prototypes for all functions that take variable numbers of arguments (including @{b}printf@{ub}); otherwise incorrect code will be generated for calls to those functions. In addition, seriously incorrect code will result if you call a function with too many arguments. (Normally, extra arguments are harmlessly ignored.) @{b}-mreg-alloc=@{I}REGS@{ui}@{ub} Control the default allocation order of integer registers. The string @{I}REGS@{ui} is a series of letters specifying a register. The supported letters are: @{b}a@{ub} allocate EAX; @{b}b@{ub} allocate EBX; @{b}c@{ub} allocate ECX; @{b}d@{ub} allocate EDX; @{b}S@{ub} allocate ESI; @{b}D@{ub} allocate EDI; @{b}B@{ub} allocate EBP. @{b}-mregparm=@{I}NUM@{ui}@{ub} Control how many registers are used to pass integer arguments. By default, no registers are used to pass arguments, and at most 3 registers can be used. You can control this behavior for a specific function by using the function attribute @{b}regparm@{ub}. See @{"Function Attributes" Link "Function Attributes"} @{i}Warning:@{ui} if you use this switch, and @{I}NUM@{ui} is nonzero, then you must build all modules with the same value, including any libraries. This includes the system libraries and startup modules. @{b}-malign-loops=@{I}NUM@{ui}@{ub} Align loops to a 2 raised to a @{I}NUM@{ui} byte boundary. If @{b}-malign-loops@{ub} is not specified, the default is 2. @{b}-malign-jumps=@{I}NUM@{ui}@{ub} Align instructions that are only jumped to to a 2 raised to a @{I}NUM@{ui} byte boundary. If @{b}-malign-jumps@{ub} is not specified, the default is 2 if optimizing for a 386, and 4 if optimizing for a 486. @{b}-malign-functions=@{I}NUM@{ui}@{ub} Align the start of functions to a 2 raised to @{I}NUM@{ui} byte boundary. If @{b}-malign-jumps@{ub} is not specified, the default is 2 if optimizing for a 386, and 4 if optimizing for a 486. @EndNode @Node "HPPA Options" "gcc.guide/HPPA Options" @Next "Intel 960 Options" @Prev "i386 Options" @Toc "Submodel Options" HPPA Options ------------ These @{b}-m@{ub} options are defined for the HPPA family of computers: @{b}-mpa-risc-1-0@{ub} Generate code for a PA 1.0 processor. @{b}-mpa-risc-1-1@{ub} Generate code for a PA 1.1 processor. @{b}-mjump-in-delay@{ub} Fill delay slots of function calls with unconditional jump instructions by modifying the return pointer for the function call to be the target of the conditional jump. @{b}-mmillicode-long-calls@{ub} Generate code which assumes millicode routines can not be reached by the standard millicode call sequence, linker-generated long-calls, or linker-modified millicode calls. In practice this should only be needed for dynamicly linked executables with extremely large SHLIB_INFO sections. @{b}-mdisable-fpregs@{ub} Prevent floating point registers from being used in any manner. This is necessary for compiling kernels which perform lazy context switching of floating point registers. If you use this option and attempt to perform floating point operations, the compiler will abort. @{b}-mdisable-indexing@{ub} Prevent the compiler from using indexing address modes. This avoids some rather obscure problems when compiling MIG generated code under MACH. @{b}-mfast-indirect-calls@{ub} Generate code which performs faster indirect calls. Such code is suitable for kernels and for static linking. The fast indirect call code will fail miserably if it's part of a dynamically linked executable and in the presense of nested functions. @{b}-mportable-runtime@{ub} Use the portable calling conventions proposed by HP for ELF systems. @{b}-mgas@{ub} Enable the use of assembler directives only GAS understands. @{b}-mschedule=@{I}CPU TYPE@{ui}@{ub} Schedule code according to the constraints for the machine type @{I}CPU TYPE@{ui}. The choices for @{I}CPU TYPE@{ui} are @{b}700@{ub} for 7@{I}N@{ui}0 machines, @{b}7100@{ub} for 7@{I}N@{ui}5 machines, and @{b}7100@{ub} for 7@{I}N@{ui}2 machines. @{b}700@{ub} is the default for @{I}CPU TYPE@{ui}. Note the @{b}7100LC@{ub} scheduling information is incomplete and using @{b}7100LC@{ub} often leads to bad schedules. For now it's probably best to use @{b}7100@{ub} instead of @{b}7100LC@{ub} for the 7@{I}N@{ui}2 machines. @{b}-msoft-float@{ub} Generate output containing library calls for floating point. @{i}Warning:@{ui} the requisite libraries are not available for all HPPA targets. Normally the facilities of the machine's usual C compiler are used, but this cannot be done directly in cross-compilation. You must make your own arrangements to provide suitable library functions for cross-compilation. The embedded target @{b}hppa1.1-*-pro@{ub} does provide software floating point support. @{b}-msoft-float@{ub} changes the calling convention in the output file; therefore, it is only useful if you compile @{i}all@{ui} of a program with this option. In particular, you need to compile @{b}libgcc.a@{ub}, the library that comes with GNU CC, with @{b}-msoft-float@{ub} in order for this to work. @EndNode @Node "Intel 960 Options" "gcc.guide/Intel 960 Options" @Next "DEC Alpha Options" @Prev "HPPA Options" @Toc "Submodel Options" Intel 960 Options ----------------- These @{b}-m@{ub} options are defined for the Intel 960 implementations: @{b}-m@{I}CPU TYPE@{ui}@{ub} Assume the defaults for the machine type @{I}CPU TYPE@{ui} for some of the other options, including instruction scheduling, floating point support, and addressing modes. The choices for @{I}CPU TYPE@{ui} are @{b}ka@{ub}, @{b}kb@{ub}, @{b}mc@{ub}, @{b}ca@{ub}, @{b}cf@{ub}, @{b}sa@{ub}, and @{b}sb@{ub}. The default is @{b}kb@{ub}. @{b}-mnumerics@{ub} @{b}-msoft-float@{ub} The @{b}-mnumerics@{ub} option indicates that the processor does support floating-point instructions. The @{b}-msoft-float@{ub} option indicates that floating-point support should not be assumed. @{b}-mleaf-procedures@{ub} @{b}-mno-leaf-procedures@{ub} Do (or do not) attempt to alter leaf procedures to be callable with the @{b}bal@{ub} instruction as well as @{b}call@{ub}. This will result in more efficient code for explicit calls when the @{b}bal@{ub} instruction can be substituted by the assembler or linker, but less efficient code in other cases, such as calls via function pointers, or using a linker that doesn't support this optimization. @{b}-mtail-call@{ub} @{b}-mno-tail-call@{ub} Do (or do not) make additional attempts (beyond those of the machine-independent portions of the compiler) to optimize tail-recursive calls into branches. You may not want to do this because the detection of cases where this is not valid is not totally complete. The default is @{b}-mno-tail-call@{ub}. @{b}-mcomplex-addr@{ub} @{b}-mno-complex-addr@{ub} Assume (or do not assume) that the use of a complex addressing mode is a win on this implementation of the i960. Complex addressing modes may not be worthwhile on the K-series, but they definitely are on the C-series. The default is currently @{b}-mcomplex-addr@{ub} for all processors except the CB and CC. @{b}-mcode-align@{ub} @{b}-mno-code-align@{ub} Align code to 8-byte boundaries for faster fetching (or don't bother). Currently turned on by default for C-series implementations only. @{b}-mic-compat@{ub} @{b}-mic2.0-compat@{ub} @{b}-mic3.0-compat@{ub} Enable compatibility with iC960 v2.0 or v3.0. @{b}-masm-compat@{ub} @{b}-mintel-asm@{ub} Enable compatibility with the iC960 assembler. @{b}-mstrict-align@{ub} @{b}-mno-strict-align@{ub} Do not permit (do permit) unaligned accesses. @{b}-mold-align@{ub} Enable structure-alignment compatibility with Intel's gcc release version 1.3 (based on gcc 1.37). Currently this is buggy in that @{b}#pragma align 1@{ub} is always assumed as well, and cannot be turned off. @EndNode @Node "DEC Alpha Options" "gcc.guide/DEC Alpha Options" @Next "Clipper Options" @Prev "Intel 960 Options" @Toc "Submodel Options" DEC Alpha Options ----------------- These @{b}-m@{ub} options are defined for the DEC Alpha implementations: @{b}-mno-soft-float@{ub} @{b}-msoft-float@{ub} Use (do not use) the hardware floating-point instructions for floating-point operations. When @{b}-msoft-float@{ub} is specified, functions in @{b}libgcc1.c@{ub} will be used to perform floating-point operations. Unless they are replaced by routines that emulate the floating-point operations, or compiled in such a way as to call such emulations routines, these routines will issue floating-point operations. If you are compiling for an Alpha without floating-point operations, you must ensure that the library is built so as not to call them. Note that Alpha implementations without floating-point operations are required to have floating-point registers. @{b}-mfp-reg@{ub} @{b}-mno-fp-regs@{ub} Generate code that uses (does not use) the floating-point register set. @{b}-mno-fp-regs@{ub} implies @{b}-msoft-float@{ub}. If the floating-point register set is not used, floating point operands are passed in integer registers as if they were integers and floating-point results are passed in $0 instead of $f0. This is a non-standard calling sequence, so any function with a floating-point argument or return value called by code compiled with @{b}-mno-fp-regs@{ub} must also be compiled with that option. A typical use of this option is building a kernel that does not use, and hence need not save and restore, any floating-point registers. @EndNode @Node "Clipper Options" "gcc.guide/Clipper Options" @Next "H8-300 Options" @Prev "DEC Alpha Options" @Toc "Submodel Options" Clipper Options --------------- These @{b}-m@{ub} options are defined for the Clipper implementations: @{b}-mc300@{ub} Produce code for a C300 Clipper processor. This is the default. @{b}-mc400@{ub} Produce code for a C400 Clipper processor i.e. use floating point registers f8..f15. @EndNode @Node "H8-300 Options" "gcc.guide/H8-300 Options" @Next "System V Options" @Prev "Clipper Options" @Toc "Submodel Options" H8/300 Options -------------- These @{b}-m@{ub} options are defined for the H8/300 implementations: @{b}-mrelax@{ub} Shorten some address references at link time, when possible; uses the linker option @{b}-relax@{ub}. See @{"ld and the H8/300" Link "ld.guide/H8-300"}, for a fuller description. @{b}-mh@{ub} Generate code for the H8/300H. @EndNode @Node "System V Options" "gcc.guide/System V Options" @Prev "H8-300 Options" @Toc "Submodel Options" Options for System V -------------------- These additional options are available on System V Release 4 for compatibility with other compilers on those systems: @{b}-Qy@{ub} Identify the versions of each tool used by the compiler, in a @{b}.ident@{ub} assembler directive in the output. @{b}-Qn@{ub} Refrain from adding @{b}.ident@{ub} directives to the output file (this is the default). @{b}-YP,@{I}DIRS@{ui}@{ub} Search the directories @{I}DIRS@{ui}, and no others, for libraries specified with @{b}-l@{ub}. @{b}-Ym,@{I}DIR@{ui}@{ub} Look in the directory @{I}DIR@{ui} to find the M4 preprocessor. The assembler uses this option. @EndNode @Node "Code Gen Options" "gcc.guide/Code Gen Options" @Next "Environment Variables" @Prev "Submodel Options" @Toc "Invoking GCC" Options for Code Generation Conventions ======================================= These machine-independent options control the interface conventions used in code generation. Most of them have both positive and negative forms; the negative form of @{b}-ffoo@{ub} would be @{b}-fno-foo@{ub}. In the table below, only one of the forms is listed--the one which is not the default. You can figure out the other form by either removing @{b}no-@{ub} or adding it. @{b}-fpcc-struct-return@{ub} Return "short" @{b}struct@{ub} and @{b}union@{ub} values in memory like longer ones, rather than in registers. This convention is less efficient, but it has the advantage of allowing intercallability between GNU CC-compiled files and files compiled with other compilers. The precise convention for returning structures in memory depends on the target configuration macros. Short structures and unions are those whose size and alignment match that of some integer type. @{b}-freg-struct-return@{ub} Use the convention that @{b}struct@{ub} and @{b}union@{ub} values are returned in registers when possible. This is more efficient for small structures than @{b}-fpcc-struct-return@{ub}. If you specify neither @{b}-fpcc-struct-return@{ub} nor its contrary @{b}-freg-struct-return@{ub}, GNU CC defaults to whichever convention is standard for the target. If there is no standard convention, GNU CC defaults to @{b}-fpcc-struct-return@{ub}, except on targets where GNU CC is the principal compiler. In those cases, we can choose the standard, and we chose the more efficient register return alternative. @{b}-fshort-enums@{ub} Allocate to an @{b}enum@{ub} type only as many bytes as it needs for the declared range of possible values. Specifically, the @{b}enum@{ub} type will be equivalent to the smallest integer type which has enough room. @{b}-fshort-double@{ub} Use the same size for @{b}double@{ub} as for @{b}float@{ub}. @{b}-fshared-data@{ub} Requests that the data and non-@{b}const@{ub} variables of this compilation be shared data rather than private data. The distinction makes sense only on certain operating systems, where shared data is shared between processes running the same program, while private data exists in one copy per process. @{b}-fno-common@{ub} Allocate even uninitialized global variables in the bss section of the object file, rather than generating them as common blocks. This has the effect that if the same variable is declared (without @{b}extern@{ub}) in two different compilations, you will get an error when you link them. The only reason this might be useful is if you wish to verify that the program will work on other systems which always work this way. @{b}-fno-ident@{ub} Ignore the @{b}#ident@{ub} directive. @{b}-fno-gnu-linker@{ub} Do not output global initializations (such as C++ constructors and destructors) in the form used by the GNU linker (on systems where the GNU linker is the standard method of handling them). Use this option when you want to use a non-GNU linker, which also requires using the @{b}collect2@{ub} program to make sure the system linker includes constructors and destructors. (@{b}collect2@{ub} is included in the GNU CC distribution.) For systems which @{i}must@{ui} use @{b}collect2@{ub}, the compiler driver @{b}gcc@{ub} is configured to do this automatically. @{b}-finhibit-size-directive@{ub} Don't output a @{b}.size@{ub} assembler directive, or anything else that would cause trouble if the function is split in the middle, and the two halves are placed at locations far apart in memory. This option is used when compiling @{b}crtstuff.c@{ub}; you should not need to use it for anything else. @{b}-fverbose-asm@{ub} Put extra commentary information in the generated assembly code to make it more readable. This option is generally only of use to those who actually need to read the generated assembly code (perhaps while debugging the compiler itself). @{b}-fvolatile@{ub} Consider all memory references through pointers to be volatile. @{b}-fvolatile-global@{ub} Consider all memory references to extern and global data items to be volatile. @{b}-fpic@{ub} Generate position-independent code (PIC) suitable for use in a shared library, if supported for the target machine. Such code accesses all constant addresses through a global offset table (GOT). If the GOT size for the linked executable exceeds a machine-specific maximum size, you get an error message from the linker indicating that @{b}-fpic@{ub} does not work; in that case, recompile with @{b}-fPIC@{ub} instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k on the m68k and RS/6000. The 386 has no such limit.) Position-independent code requires special support, and therefore works only on certain machines. For the 386, GNU CC supports PIC for System V but not for the Sun 386i. Code generated for the IBM RS/6000 is always position-independent. The GNU assembler does not fully support PIC. Currently, you must use some other assembler in order for PIC to work. We would welcome volunteers to upgrade GAS to handle this; the first part of the job is to figure out what the assembler must do differently. @{b}-fPIC@{ub} If supported for the target machine, emit position-independent code, suitable for dynamic linking and avoiding any limit on the size of the global offset table. This option makes a difference on the m68k, m88k and the Sparc. Position-independent code requires special support, and therefore works only on certain machines. @{b}-ffixed-@{I}REG@{ui}@{ub} Treat the register named @{I}REG@{ui} as a fixed register; generated code should never refer to it (except perhaps as a stack pointer, frame pointer or in some other fixed role). @{I}REG@{ui} must be the name of a register. The register names accepted are machine-specific and are defined in the @{b}REGISTER_NAMES@{ub} macro in the machine description macro file. This flag does not have a negative form, because it specifies a three-way choice. @{b}-fcall-used-@{I}REG@{ui}@{ub} Treat the register named @{I}REG@{ui} as an allocatable register that is clobbered by function calls. It may be allocated for temporaries or variables that do not live across a call. Functions compiled this way will not save and restore the register @{I}REG@{ui}. Use of this flag for a register that has a fixed pervasive role in the machine's execution model, such as the stack pointer or frame pointer, will produce disastrous results. This flag does not have a negative form, because it specifies a three-way choice. @{b}-fcall-saved-@{I}REG@{ui}@{ub} Treat the register named @{I}REG@{ui} as an allocatable register saved by functions. It may be allocated even for temporaries or variables that live across a call. Functions compiled this way will save and restore the register @{I}REG@{ui} if they use it. Use of this flag for a register that has a fixed pervasive role in the machine's execution model, such as the stack pointer or frame pointer, will produce disastrous results. A different sort of disaster will result from the use of this flag for a register in which function values may be returned. This flag does not have a negative form, because it specifies a three-way choice. @{b}-fpack-struct@{ub} Pack all structure members together without holes. Usually you would not want to use this option, since it makes the code suboptimal, and the offsets of structure members won't agree with system libraries. @{b}+e0@{ub} @{b}+e1@{ub} Control whether virtual function definitions in classes are used to generate code, or only to define interfaces for their callers. (C++ only). These options are provided for compatibility with @{b}cfront@{ub} 1.x usage; the recommended alternative GNU C++ usage is in flux. See @{"Declarations and Definitions in One Header" Link "C++ Interface"}. With @{b}+e0@{ub}, virtual function definitions in classes are declared @{b}extern@{ub}; the declaration is used only as an interface specification, not to generate code for the virtual functions (in this compilation). With @{b}+e1@{ub}, G++ actually generates the code implementing virtual functions defined in the code, and makes them publicly visible. @EndNode @Node "Environment Variables" "gcc.guide/Environment Variables" @Next "Running Protoize" @Prev "Code Gen Options" @Toc "Invoking GCC" Environment Variables Affecting GNU CC ====================================== This section describes several environment variables that affect how GNU CC operates. They work by specifying directories or prefixes to use when searching for various kinds of files. Note that you can also specify places to search using options such as @{b}-B@{ub}, @{b}-I@{ub} and @{b}-L@{ub} (see @{"Directory Options" Link "Directory Options"}). These take precedence over places specified using environment variables, which in turn take precedence over those specified by the configuration of GNU CC. See @{"Driver" Link "Driver"}. @{b}TMPDIR@{ub} If @{b}TMPDIR@{ub} is set, it specifies the directory to use for temporary files. GNU CC uses temporary files to hold the output of one stage of compilation which is to be used as input to the next stage: for example, the output of the preprocessor, which is the input to the compiler proper. @{b}GCC_EXEC_PREFIX@{ub} If @{b}GCC_EXEC_PREFIX@{ub} is set, it specifies a prefix to use in the names of the subprograms executed by the compiler. No slash is added when this prefix is combined with the name of a subprogram, but you can specify a prefix that ends with a slash if you wish. If GNU CC cannot find the subprogram using the specified prefix, it tries looking in the usual places for the subprogram. The default value of @{b}GCC_EXEC_PREFIX@{ub} is @{b}@{I}PREFIX@{ui}/lib/gcc-lib/@{ub} where @{I}PREFIX@{ui} is the value of @{b}prefix@{ub} when you ran the @{b}configure@{ub} script. Other prefixes specified with @{b}-B@{ub} take precedence over this prefix. This prefix is also used for finding files such as @{b}crt0.o@{ub} that are used for linking. In addition, the prefix is used in an unusual way in finding the directories to search for header files. For each of the standard directories whose name normally begins with @{b}/usr/local/lib/gcc-lib@{ub} (more precisely, with the value of @{b}GCC_INCLUDE_DIR@{ub}), GNU CC tries replacing that beginning with the specified prefix to produce an alternate directory name. Thus, with @{b}-Bfoo/@{ub}, GNU CC will search @{b}foo/bar@{ub} where it would normally search @{b}/usr/local/lib/bar@{ub}. These alternate directories are searched first; the standard directories come next. @{b}COMPILER_PATH@{ub} The value of @{b}COMPILER_PATH@{ub} is a colon-separated list of directories, much like @{b}PATH@{ub}. GNU CC tries the directories thus specified when searching for subprograms, if it can't find the subprograms using @{b}GCC_EXEC_PREFIX@{ub}. @{b}LIBRARY_PATH@{ub} The value of @{b}LIBRARY_PATH@{ub} is a colon-separated list of directories, much like @{b}PATH@{ub}. When configured as a native compiler, GNU CC tries the directories thus specified when searching for special linker files, if it can't find them using @{b}GCC_EXEC_PREFIX@{ub}. Linking using GNU CC also uses these directories when searching for ordinary libraries for the @{b}-l@{ub} option (but directories specified with @{b}-L@{ub} come first). @{b}C_INCLUDE_PATH@{ub} @{b}CPLUS_INCLUDE_PATH@{ub} @{b}OBJC_INCLUDE_PATH@{ub} These environment variables pertain to particular languages. Each variable's value is a colon-separated list of directories, much like @{b}PATH@{ub}. When GNU CC searches for header files, it tries the directories listed in the variable for the language you are using, after the directories specified with @{b}-I@{ub} but before the standard header file directories. @{b}DEPENDENCIES_OUTPUT@{ub} If this variable is set, its value specifies how to output dependencies for Make based on the header files processed by the compiler. This output looks much like the output from the @{b}-M@{ub} option (see @{"Preprocessor Options" Link "Preprocessor Options"}), but it goes to a separate file, and is in addition to the usual results of compilation. The value of @{b}DEPENDENCIES_OUTPUT@{ub} can be just a file name, in which case the Make rules are written to that file, guessing the target name from the source file name. Or the value can have the form @{b}@{I}FILE@{ui} @{I}TARGET@{ui}@{ub}, in which case the rules are written to file @{I}FILE@{ui} using @{I}TARGET@{ui} as the target name. @EndNode @Node "Running Protoize" "gcc.guide/Running Protoize" @Prev "Environment Variables" @Toc "Invoking GCC" Running Protoize ================ The program @{b}protoize@{ub} is an optional part of GNU C. You can use it to add prototypes to a program, thus converting the program to ANSI C in one respect. The companion program @{b}unprotoize@{ub} does the reverse: it removes argument types from any prototypes that are found. When you run these programs, you must specify a set of source files as command line arguments. The conversion programs start out by compiling these files to see what functions they define. The information gathered about a file @{I}FOO@{ui} is saved in a file named @{b}@{I}FOO@{ui}.X@{ub}. After scanning comes actual conversion. The specified files are all eligible to be converted; any files they include (whether sources or just headers) are eligible as well. But not all the eligible files are converted. By default, @{b}protoize@{ub} and @{b}unprotoize@{ub} convert only source and header files in the current directory. You can specify additional directories whose files should be converted with the @{b}-d @{I}DIRECTORY@{ui}@{ub} option. You can also specify particular files to exclude with the @{b}-x @{I}FILE@{ui}@{ub} option. A file is converted if it is eligible, its directory name matches one of the specified directory names, and its name within the directory has not been excluded. Basic conversion with @{b}protoize@{ub} consists of rewriting most function definitions and function declarations to specify the types of the arguments. The only ones not rewritten are those for varargs functions. @{b}protoize@{ub} optionally inserts prototype declarations at the beginning of the source file, to make them available for any calls that precede the function's definition. Or it can insert prototype declarations with block scope in the blocks where undeclared functions are called. Basic conversion with @{b}unprotoize@{ub} consists of rewriting most function declarations to remove any argument types, and rewriting function definitions to the old-style pre-ANSI form. Both conversion programs print a warning for any function declaration or definition that they can't convert. You can suppress these warnings with @{b}-q@{ub}. The output from @{b}protoize@{ub} or @{b}unprotoize@{ub} replaces the original source file. The original file is renamed to a name ending with @{b}.save@{ub}. If the @{b}.save@{ub} file already exists, then the source file is simply discarded. @{b}protoize@{ub} and @{b}unprotoize@{ub} both depend on GNU CC itself to scan the program and collect information about the functions it uses. So neither of these programs will work until GNU CC is installed. Here is a table of the options you can use with @{b}protoize@{ub} and @{b}unprotoize@{ub}. Each option works with both programs unless otherwise stated. @{b}-B @{I}DIRECTORY@{ui}@{ub} Look for the file @{b}SYSCALLS.c.X@{ub} in @{I}DIRECTORY@{ui}, instead of the usual directory (normally @{b}/usr/local/lib@{ub}). This file contains prototype information about standard system functions. This option applies only to @{b}protoize@{ub}. @{b}-c @{I}COMPILATION-OPTIONS@{ui}@{ub} Use @{I}COMPILATION-OPTIONS@{ui} as the options when running @{b}gcc@{ub} to produce the @{b}.X@{ub} files. The special option @{b}-aux-info@{ub} is always passed in addition, to tell @{b}gcc@{ub} to write a @{b}.X@{ub} file. Note that the compilation options must be given as a single argument to @{b}protoize@{ub} or @{b}unprotoize@{ub}. If you want to specify several @{b}gcc@{ub} options, you must quote the entire set of compilation options to make them a single word in the shell. There are certain @{b}gcc@{ub} arguments that you cannot use, because they would produce the wrong kind of output. These include @{b}-g@{ub}, @{b}-O@{ub}, @{b}-c@{ub}, @{b}-S@{ub}, and @{b}-o@{ub} If you include these in the @{I}COMPILATION-OPTIONS@{ui}, they are ignored. @{b}-C@{ub} Rename files to end in @{b}.C@{ub} instead of @{b}.c@{ub}. This is convenient if you are converting a C program to C++. This option applies only to @{b}protoize@{ub}. @{b}-g@{ub} Add explicit global declarations. This means inserting explicit declarations at the beginning of each source file for each function that is called in the file and was not declared. These declarations precede the first function definition that contains a call to an undeclared function. This option applies only to @{b}protoize@{ub}. @{b}-i @{I}STRING@{ui}@{ub} Indent old-style parameter declarations with the string @{I}STRING@{ui}. This option applies only to @{b}protoize@{ub}. @{b}unprotoize@{ub} converts prototyped function definitions to old-style function definitions, where the arguments are declared between the argument list and the initial @{b}{@{ub}. By default, @{b}unprotoize@{ub} uses five spaces as the indentation. If you want to indent with just one space instead, use @{b}-i " "@{ub}. @{b}-k@{ub} Keep the @{b}.X@{ub} files. Normally, they are deleted after conversion is finished. @{b}-l@{ub} Add explicit local declarations. @{b}protoize@{ub} with @{b}-l@{ub} inserts a prototype declaration for each function in each block which calls the function without any declaration. This option applies only to @{b}protoize@{ub}. @{b}-n@{ub} Make no real changes. This mode just prints information about the conversions that would have been done without @{b}-n@{ub}. @{b}-N@{ub} Make no @{b}.save@{ub} files. The original files are simply deleted. Use this option with caution. @{b}-p @{I}PROGRAM@{ui}@{ub} Use the program @{I}PROGRAM@{ui} as the compiler. Normally, the name @{b}gcc@{ub} is used. @{b}-q@{ub} Work quietly. Most warnings are suppressed. @{b}-v@{ub} Print the version number, just like @{b}-v@{ub} for @{b}gcc@{ub}. If you need special compiler options to compile one of your program's source files, then you should generate that file's @{b}.X@{ub} file specially, by running @{b}gcc@{ub} on that source file with the appropriate options and the option @{b}-aux-info@{ub}. Then run @{b}protoize@{ub} on the entire set of files. @{b}protoize@{ub} will use the existing @{b}.X@{ub} file because it is newer than the source file. For example: gcc -Dfoo=bar file1.c -aux-info protoize *.c You need to include the special files along with the rest in the @{b}protoize@{ub} command, even though their @{b}.X@{ub} files already exist, because otherwise they won't get converted. See @{"Protoize Caveats" Link "Protoize Caveats"}, for more information on how to use @{b}protoize@{ub} successfully. @EndNode @Node "Installation" "gcc.guide/Installation" @Next "C Extensions" @Prev "Invoking GCC" @Toc "Main" Installing GNU CC ***************** @{" Configurations " Link "Configurations"} Configurations Supported by GNU CC. @{" Other Dir " Link "Other Dir"} Compiling in a separate directory (not where the source is). @{" Cross-Compiler " Link "Cross-Compiler"} Building and installing a cross-compiler. @{" Sun Install " Link "Sun Install"} See below for installation on the Sun. @{" VMS Install " Link "VMS Install"} See below for installation on VMS. @{" Collect2 " Link "Collect2"} How @{b}collect2@{ub} works; how it finds @{b}ld@{ub}. @{" Header Dirs " Link "Header Dirs"} Understanding the standard header file directories. Here is the procedure for installing GNU CC on a Unix system. See @{"VMS Install" Link "VMS Install"}, for VMS systems. In this section we assume you compile in the same directory that contains the source files; see @{"Other Dir" Link "Other Dir"}, to find out how to compile in a separate directory on Unix systems. You cannot install GNU C by itself on MSDOS; it will not compile under any MSDOS compiler except itself. You need to get the complete compilation package DJGPP, which includes binaries as well as sources, and includes all the necessary compilation tools and libraries. 1. If you have built GNU CC previously in the same directory for a different target machine, do @{b}make distclean@{ub} to delete all files that might be invalid. One of the files this deletes is @{b}Makefile@{ub}; if @{b}make distclean@{ub} complains that @{b}Makefile@{ub} does not exist, it probably means that the directory is already suitably clean. 2. On a System V release 4 system, make sure @{b}/usr/bin@{ub} precedes @{b}/usr/ucb@{ub} in @{b}PATH@{ub}. The @{b}cc@{ub} command in @{b}/usr/ucb@{ub} uses libraries which have bugs. 3. Specify the host, build and target machine configurations. You do this by running the file @{b}configure@{ub}. The @{i}build@{ui} machine is the system which you are using, the @{i}host@{ui} machine is the system where you want to run the resulting compiler (normally the build machine), and the @{i}target@{ui} machine is the system for which you want the compiler to generate code. If you are building a compiler to produce code for the machine it runs on (a native compiler), you normally do not need to specify any operands to @{b}configure@{ub}; it will try to guess the type of machine you are on and use that as the build, host and target machines. So you don't need to specify a configuration when building a native compiler unless @{b}configure@{ub} cannot figure out what your configuration is or guesses wrong. In those cases, specify the build machine's @{i}configuration name@{ui} with the @{b}--build@{ub} option; the host and target will default to be the same as the build machine. (If you are building a cross-compiler, see @{"Cross-Compiler" Link "Cross-Compiler"}.) Here is an example: ./configure --build=sparc-sun-sunos4.1 A configuration name may be canonical or it may be more or less abbreviated. A canonical configuration name has three parts, separated by dashes. It looks like this: @{b}@{I}CPU@{ui}-@{I}COMPANY@{ui}-@{I}SYSTEM@{ui}@{ub}. (The three parts may themselves contain dashes; @{b}configure@{ub} can figure out which dashes serve which purpose.) For example, @{b}m68k-sun-sunos4.1@{ub} specifies a Sun 3. You can also replace parts of the configuration by nicknames or aliases. For example, @{b}sun3@{ub} stands for @{b}m68k-sun@{ub}, so @{b}sun3-sunos4.1@{ub} is another way to specify a Sun 3. You can also use simply @{b}sun3-sunos@{ub}, since the version of SunOS is assumed by default to be version 4. @{b}sun3-bsd@{ub} also works, since @{b}configure@{ub} knows that the only BSD variant on a Sun 3 is SunOS. You can specify a version number after any of the system types, and some of the CPU types. In most cases, the version is irrelevant, and will be ignored. So you might as well specify the version if you know it. See @{"Configurations" Link "Configurations"}, for a list of supported configuration names and notes on many of the configurations. You should check the notes in that section before proceeding any further with the installation of GNU CC. There are four additional options you can specify independently to describe variant hardware and software configurations. These are @{b}--with-gnu-as@{ub}, @{b}--with-gnu-ld@{ub}, @{b}--with-stabs@{ub} and @{b}--nfp@{ub}. @{b}--with-gnu-as@{ub} If you will use GNU CC with the GNU assembler (GAS), you should declare this by using the @{b}--with-gnu-as@{ub} option when you run @{b}configure@{ub}. Using this option does not install GAS. It only modifies the output of GNU CC to work with GAS. Building and installing GAS is up to you. Conversely, if you @{i}do not@{ui} wish to use GAS and do not specify @{b}--with-gnu-as@{ub} when building GNU CC, it is up to you to make sure that GAS is not installed. GNU CC searches for a program named @{b}as@{ub} in various directories; if the program it finds is GAS, then it runs GAS. If you are not sure where GNU CC finds the assembler it is using, try specifying @{b}-v@{ub} when you run it. The systems where it makes a difference whether you use GAS are @{b}hppa1.0-@{I}ANY@{ui}-@{I}ANY@{ui}@{ub}, @{b}hppa1.1-@{I}ANY@{ui}-@{I}ANY@{ui}@{ub}, @{b}i386-@{I}ANY@{ui}-sysv@{ub}, @{b}i386-@{I}ANY@{ui}-isc@{ub}, @{b}i860-@{I}ANY@{ui}-bsd@{ub}, @{b}m68k-bull-sysv@{ub}, @{b}m68k-hp-hpux@{ub}, @{b}m68k-sony-bsd@{ub}, @{b}m68k-altos-sysv@{ub}, @{b}m68000-hp-hpux@{ub}, @{b}m68000-att-sysv@{ub}, @{b}@{I}ANY@{ui}-lynx-lynxos@{ub}, and @{b}mips-@{I}ANY@{ui}@{ub}). On any other system, @{b}--with-gnu-as@{ub} has no effect. On the systems listed above (except for the HP-PA, for ISC on the 386, and for @{b}mips-sgi-irix5.*@{ub}), if you use GAS, you should also use the GNU linker (and specify @{b}--with-gnu-ld@{ub}). @{b}--with-gnu-ld@{ub} Specify the option @{b}--with-gnu-ld@{ub} if you plan to use the GNU linker with GNU CC. This option does not cause the GNU linker to be installed; it just modifies the behavior of GNU CC to work with the GNU linker. Specifically, it inhibits the installation of @{b}collect2@{ub}, a program which otherwise serves as a front-end for the system's linker on most configurations. @{b}--with-stabs@{ub} On MIPS based systems and on Alphas, you must specify whether you want GNU CC to create the normal ECOFF debugging format, or to use BSD-style stabs passed through the ECOFF symbol table. The normal ECOFF debug format cannot fully handle languages other than C. BSD stabs format can handle other languages, but it only works with the GNU debugger GDB. Normally, GNU CC uses the ECOFF debugging format by default; if you prefer BSD stabs, specify @{b}--with-stabs@{ub} when you configure GNU CC. No matter which default you choose when you configure GNU CC, the user can use the @{b}-gcoff@{ub} and @{b}-gstabs+@{ub} options to specify explicitly the debug format for a particular compilation. @{b}--with-stabs@{ub} is meaningful on the ISC system on the 386, also, if @{b}--with-gas@{ub} is used. It selects use of stabs debugging information embedded in COFF output. This kind of debugging information supports C++ well; ordinary COFF debugging information does not. @{b}--with-stabs@{ub} is also meaningful on 386 systems running SVR4. It selects use of stabs debugging information embedded in ELF output. The C++ compiler currently (2.6.0) does not support the DWARF debugging information normally used on 386 SVR4 platforms; stabs provide a workable alternative. This requires gas and gdb, as the normal SVR4 tools can not generate or interpret stabs. @{b}--nfp@{ub} On certain systems, you must specify whether the machine has a floating point unit. These systems include @{b}m68k-sun-sunos@{I}N@{ui}@{ub} and @{b}m68k-isi-bsd@{ub}. On any other system, @{b}--nfp@{ub} currently has no effect, though perhaps there are other systems where it could usefully make a difference. The @{b}configure@{ub} script searches subdirectories of the source directory for other compilers that are to be integrated into GNU CC. The GNU compiler for C++, called G++ is in a subdirectory named @{b}cp@{ub}. @{b}configure@{ub} inserts rules into @{b}Makefile@{ub} to build all of those compilers. Here we spell out what files will be set up by @{b}configure@{ub}. Normally you need not be concerned with these files. @{b}*@{ub} A file named @{b}config.h@{ub} is created that contains a @{b}#include@{ub} of the top-level config file for the machine you will run the compiler on (see @{"Config" Link "Config"}). This file is responsible for defining information about the host machine. It includes @{b}tm.h@{ub}. The top-level config file is located in the subdirectory @{b}config@{ub}. Its name is always @{b}xm-@{I}SOMETHING@{ui}.h@{ub}; usually @{b}xm-@{I}MACHINE@{ui}.h@{ub}, but there are some exceptions. If your system does not support symbolic links, you might want to set up @{b}config.h@{ub} to contain a @{b}#include@{ub} command which refers to the appropriate file. @{b}*@{ub} A file named @{b}tconfig.h@{ub} is created which includes the top-level config file for your target machine. This is used for compiling certain programs to run on that machine. @{b}*@{ub} A file named @{b}tm.h@{ub} is created which includes the machine-description macro file for your target machine. It should be in the subdirectory @{b}config@{ub} and its name is often @{b}@{I}MACHINE@{ui}.h@{ub}. @{b}*@{ub} The command file @{b}configure@{ub} also constructs the file @{b}Makefile@{ub} by adding some text to the template file @{b}Makefile.in@{ub}. The additional text comes from files in the @{b}config@{ub} directory, named @{b}t-@{I}TARGET@{ui}@{ub} and @{b}x-@{I}HOST@{ui}@{ub}. If these files do not exist, it means nothing needs to be added for a given target or host. 4. The standard directory for installing GNU CC is @{b}/usr/local/lib@{ub}. If you want to install its files somewhere else, specify @{b}--prefix=@{I}DIR@{ui}@{ub} when you run @{b}configure@{ub}. Here @{I}DIR@{ui} is a directory name to use instead of @{b}/usr/local@{ub} for all purposes with one exception: the directory @{b}/usr/local/include@{ub} is searched for header files no matter where you install the compiler. To override this name, use the @{b}--local-prefix@{ub} option below. 5. Specify @{b}--local-prefix=@{I}DIR@{ui}@{ub} if you want the compiler to search directory @{b}@{I}DIR@{ui}/include@{ub} for locally installed header files @{i}instead@{ui} of @{b}/usr/local/include@{ub}. You should specify @{b}--local-prefix@{ub} @{i}only@{ui} if your site has a different convention (not @{b}/usr/local@{ub}) for where to put site-specific files. @{i}Do not@{ui} specify @{b}/usr@{ub} as the @{b}--local-prefix@{ub}! The directory you use for @{b}--local-prefix@{ub} @{i}must not@{ui} contain any of the system's standard header files. If it did contain them, certain programs would be miscompiled (including GNU Emacs, on certain targets), because this would override and nullify the header file corrections made by the @{b}fixincludes@{ub} script. 6. Make sure the Bison parser generator is installed. (This is unnecessary if the Bison output files @{b}c-parse.c@{ub} and @{b}cexp.c@{ub} are more recent than @{b}c-parse.y@{ub} and @{b}cexp.y@{ub} and you do not plan to change the @{b}.y@{ub} files.) Bison versions older than Sept 8, 1988 will produce incorrect output for @{b}c-parse.c@{ub}. 7. If you have chosen a configuration for GNU CC which requires other GNU tools (such as GAS or the GNU linker) instead of the standard system tools, install the required tools in the build directory under the names @{b}as@{ub}, @{b}ld@{ub} or whatever is appropriate. This will enable the compiler to find the proper tools for compilation of the program @{b}enquire@{ub}. Alternatively, you can do subsequent compilation using a value of the @{b}PATH@{ub} environment variable such that the necessary GNU tools come before the standard system tools. 8. Build the compiler. Just type @{b}make LANGUAGES=c@{ub} in the compiler directory. @{b}LANGUAGES=c@{ub} specifies that only the C compiler should be compiled. The makefile normally builds compilers for all the supported languages; currently, C, C++ and Objective C. However, C is the only language that is sure to work when you build with other non-GNU C compilers. In addition, building anything but C at this stage is a waste of time. In general, you can specify the languages to build by typing the argument @{b}LANGUAGES="@{I}LIST@{ui}"@{ub}, where @{I}LIST@{ui} is one or more words from the list @{b}c@{ub}, @{b}c++@{ub}, and @{b}objective-c@{ub}. If you have any additional GNU compilers as subdirectories of the GNU CC source directory, you may also specify their names in this list. Ignore any warnings you may see about "statement not reached" in @{b}insn-emit.c@{ub}; they are normal. Also, warnings about "unknown escape sequence" are normal in @{b}genopinit.c@{ub} and perhaps some other files. Likewise, you should ignore warnings about "constant is so large that it is unsigned" in @{b}insn-emit.c@{ub} and @{b}insn-recog.c@{ub} and a warning about a comparison always being zero in @{b}enquire.o@{ub}. Any other compilation errors may represent bugs in the port to your machine or operating system, and should be investigated and reported (see @{"Bugs" Link "Bugs"}). Some commercial compilers fail to compile GNU CC because they have bugs or limitations. For example, the Microsoft compiler is said to run out of macro space. Some Ultrix compilers run out of expression space; then you need to break up the statement where the problem happens. 9. If you are building a cross-compiler, stop here. See @{"Cross-Compiler" Link "Cross-Compiler"}. 10. Move the first-stage object files and executables into a subdirectory with this command: make stage1 The files are moved into a subdirectory named @{b}stage1@{ub}. Once installation is complete, you may wish to delete these files with @{b}rm -r stage1@{ub}. 11. If you have chosen a configuration for GNU CC which requires other GNU tools (such as GAS or the GNU linker) instead of the standard system tools, install the required tools in the @{b}stage1@{ub} subdirectory under the names @{b}as@{ub}, @{b}ld@{ub} or whatever is appropriate. This will enable the stage 1 compiler to find the proper tools in the following stage. Alternatively, you can do subsequent compilation using a value of the @{b}PATH@{ub} environment variable such that the necessary GNU tools come before the standard system tools. 12. Recompile the compiler with itself, with this command: make CC="stage1/xgcc -Bstage1/" CFLAGS="-g -O2" This is called making the stage 2 compiler. The command shown above builds compilers for all the supported languages. If you don't want them all, you can specify the languages to build by typing the argument @{b}LANGUAGES="@{I}LIST@{ui}"@{ub}. @{I}LIST@{ui} should contain one or more words from the list @{b}c@{ub}, @{b}c++@{ub}, @{b}objective-c@{ub}, and @{b}proto@{ub}. Separate the words with spaces. @{b}proto@{ub} stands for the programs @{b}protoize@{ub} and @{b}unprotoize@{ub}; they are not a separate language, but you use @{b}LANGUAGES@{ub} to enable or disable their installation. If you are going to build the stage 3 compiler, then you might want to build only the C language in stage 2. Once you have built the stage 2 compiler, if you are short of disk space, you can delete the subdirectory @{b}stage1@{ub}. On a 68000 or 68020 system lacking floating point hardware, unless you have selected a @{b}tm.h@{ub} file that expects by default that there is no such hardware, do this instead: make CC="stage1/xgcc -Bstage1/" CFLAGS="-g -O2 -msoft-float" 13. If you wish to test the compiler by compiling it with itself one more time, install any other necessary GNU tools (such as GAS or the GNU linker) in the @{b}stage2@{ub} subdirectory as you did in the @{b}stage1@{ub} subdirectory, then do this: make stage2 make CC="stage2/xgcc -Bstage2/" CFLAGS="-g -O2" This is called making the stage 3 compiler. Aside from the @{b}-B@{ub} option, the compiler options should be the same as when you made the stage 2 compiler. But the @{b}LANGUAGES@{ub} option need not be the same. The command shown above builds compilers for all the supported languages; if you don't want them all, you can specify the languages to build by typing the argument @{b}LANGUAGES="@{I}LIST@{ui}"@{ub}, as described above. If you do not have to install any additional GNU tools, you may use the command make bootstrap LANGUAGES=@{I}LANGUAGE-LIST@{ui} BOOT_CFLAGS=@{I}OPTION-LIST@{ui} instead of making @{b}stage1@{ub}, @{b}stage2@{ub}, and performing the two compiler builds. 14. Then compare the latest object files with the stage 2 object files--they ought to be identical, aside from time stamps (if any). On some systems, meaningful comparison of object files is impossible; they always appear "different." This is currently true on Solaris and some systems that use ELF object file format. On some versions of Irix on SGI machines and DEC Unix (OSF/1) on Alpha systems, you will not be able to compare the files without specifying @{b}-save-temps@{ub}; see the description of individual systems above to see if you get comparison failures. You may have similar problems on other systems. Use this command to compare the files: make compare This will mention any object files that differ between stage 2 and stage 3. Any difference, no matter how innocuous, indicates that the stage 2 compiler has compiled GNU CC incorrectly, and is therefore a potentially serious bug which you should investigate and report (see @{"Bugs" Link "Bugs"}). If your system does not put time stamps in the object files, then this is a faster way to compare them (using the Bourne shell): for file in *.o; do cmp $file stage2/$file done If you have built the compiler with the @{b}-mno-mips-tfile@{ub} option on MIPS machines, you will not be able to compare the files. 15. Install the compiler driver, the compiler's passes and run-time support with @{b}make install@{ub}. Use the same value for @{b}CC@{ub}, @{b}CFLAGS@{ub} and @{b}LANGUAGES@{ub} that you used when compiling the files that are being installed. One reason this is necessary is that some versions of Make have bugs and recompile files gratuitously when you do this step. If you use the same variable values, those files will be recompiled properly. For example, if you have built the stage 2 compiler, you can use the following command: make install CC="stage2/xgcc -Bstage2/" CFLAGS="-g -O" LANGUAGES="@{I}LIST@{ui}" This copies the files @{b}cc1@{ub}, @{b}cpp@{ub} and @{b}libgcc.a@{ub} to files @{b}cc1@{ub}, @{b}cpp@{ub} and @{b}libgcc.a@{ub} in the directory @{b}/usr/local/lib/gcc-lib/@{I}TARGET@{ui}/@{I}VERSION@{ui}@{ub}, which is where the compiler driver program looks for them. Here @{I}TARGET@{ui} is the target machine type specified when you ran @{b}configure@{ub}, and @{I}VERSION@{ui} is the version number of GNU CC. This naming scheme permits various versions and/or cross-compilers to coexist. This also copies the driver program @{b}xgcc@{ub} into @{b}/usr/local/bin/gcc@{ub}, so that it appears in typical execution search paths. On some systems, this command causes recompilation of some files. This is usually due to bugs in @{b}make@{ub}. You should either ignore this problem, or use GNU Make. @{i}Warning: there is a bug in @{b}alloca@{ub} in the Sun library. To avoid this bug, be sure to install the executables of GNU CC that were compiled by GNU CC. (That is, the executables from stage 2 or 3, not stage 1.) They use @{b}alloca@{ub} as a built-in function and never the one in the library.@{ui} (It is usually better to install GNU CC executables from stage 2 or 3, since they usually run faster than the ones compiled with some other compiler.) 16. If you're going to use C++, it's likely that you need to also install the libg++ distribution. It should be available from the same place where you got the GNU C distribution. Just as GNU C does not distribute a C runtime library, it also does not include a C++ run-time library. All I/O functionality, special class libraries, etc., are available in the libg++ distribution. @EndNode @Node "Configurations" "gcc.guide/Configurations" @Next "Other Dir" @Toc "Installation" Configurations Supported by GNU CC ================================== Here are the possible CPU types: 1750a, a29k, alpha, arm, c@{I}N@{ui}, clipper, dsp16xx, elxsi, h8300, hppa1.0, hppa1.1, i370, i386, i486, i586, i860, i960, m68000, m68k, m88k, mips, mipsel, mips64, mips64el, ns32k, powerpc, powerpcle, pyramid, romp, rs6000, sh, sparc, sparclite, sparc64, vax, we32k. Here are the recognized company names. As you can see, customary abbreviations are used rather than the longer official names. acorn, alliant, altos, apollo, att, bull, cbm, convergent, convex, crds, dec, dg, dolphin, elxsi, encore, harris, hitachi, hp, ibm, intergraph, isi, mips, motorola, ncr, next, ns, omron, plexus, sequent, sgi, sony, sun, tti, unicom, wrs. The company name is meaningful only to disambiguate when the rest of the information supplied is insufficient. You can omit it, writing just @{b}@{I}CPU@{ui}-@{I}SYSTEM@{ui}@{ub}, if it is not needed. For example, @{b}vax-ultrix4.2@{ub} is equivalent to @{b}vax-dec-ultrix4.2@{ub}. Here is a list of system types: 386bsd, aix, acis, amigaos, aos, aout, bosx, bsd, clix, coff, ctix, cxux, dgux, dynix, ebmon, ecoff, elf, esix, freebsd, hms, genix, gnu, gnu/linux, hiux, hpux, iris, irix, isc, luna, lynxos, mach, minix, msdos, mvs, netbsd, newsos, nindy, ns, osf, osfrose, ptx, riscix, riscos, rtu, sco, sim, solaris, sunos, sym, sysv, udi, ultrix, unicos, uniplus, unos, vms, vsta, vxworks, winnt, xenix. You can omit the system type; then @{b}configure@{ub} guesses the operating system from the CPU and company. You can add a version number to the system type; this may or may not make a difference. For example, you can write @{b}bsd4.3@{ub} or @{b}bsd4.4@{ub} to distinguish versions of BSD. In practice, the version number is most needed for @{b}sysv3@{ub} and @{b}sysv4@{ub}, which are often treated differently. If you specify an impossible combination such as @{b}i860-dg-vms@{ub}, then you may get an error message from @{b}configure@{ub}, or it may ignore part of the information and do the best it can with the rest. @{b}configure@{ub} always prints the canonical name for the alternative that it used. GNU CC does not support all possible alternatives. Often a particular model of machine has a name. Many machine names are recognized as aliases for CPU/company combinations. Thus, the machine name @{b}sun3@{ub}, mentioned above, is an alias for @{b}m68k-sun@{ub}. Sometimes we accept a company name as a machine name, when the name is popularly used for a particular machine. Here is a table of the known machine names: 3300, 3b1, 3b@{I}N@{ui}, 7300, altos3068, altos, apollo68, att-7300, balance, convex-c@{I}N@{ui}, crds, decstation-3100, decstation, delta, encore, fx2800, gmicro, hp7@{I}NN@{ui}, hp8@{I}NN@{ui}, hp9k2@{I}NN@{ui}, hp9k3@{I}NN@{ui}, hp9k7@{I}NN@{ui}, hp9k8@{I}NN@{ui}, iris4d, iris, isi68, m3230, magnum, merlin, miniframe, mmax, news-3600, news800, news, next, pbd, pc532, pmax, powerpc, powerpcle, ps2, risc-news, rtpc, sun2, sun386i, sun386, sun3, sun4, symmetry, tower-32, tower. Remember that a machine name specifies both the cpu type and the company name. If you want to install your own homemade configuration files, you can use @{b}local@{ub} as the company name to access them. If you use configuration @{b}@{I}CPU@{ui}-local@{ub}, the configuration name without the cpu prefix is used to form the configuration file names. Thus, if you specify @{b}m68k-local@{ub}, configuration uses files @{b}m68k.md@{ub}, @{b}local.h@{ub}, @{b}m68k.c@{ub}, @{b}xm-local.h@{ub}, @{b}t-local@{ub}, and @{b}x-local@{ub}, all in the directory @{b}config/m68k@{ub}. Here is a list of configurations that have special treatment or special things you must know: @{b}1750a-*-*@{ub} MIL-STD-1750A processors. Starting with GCC 2.6.1, the MIL-STD-1750A cross configuration no longer supports the Tektronix Assembler, but instead produces output for @{b}as1750@{ub}, an assembler/linker available under the GNU Public License for the 1750A. Contact @{i}kellogg@space.otn.dasa.de@{ui} for more details on obtaining @{b}as1750@{ub}. A similarly licensed simulator for the 1750A is available from same address. You should ignore a fatal error during the building of libgcc (libgcc is not yet implemented for the 1750A.) The @{b}as1750@{ub} assembler requires the file @{b}ms1750.inc@{ub}, which is found in the directory @{b}config/1750a@{ub}. GNU CC produced the same sections as the Fairchild F9450 C Compiler, namely: @{b}Normal@{ub} The program code section. @{b}Static@{ub} The read/write (RAM) data section. @{b}Konst@{ub} The read-only (ROM) constants section. @{b}Init@{ub} Initialization section (code to copy KREL to SREL). The smallest addressable unit is 16 bits (BITS_PER_UNIT is 16). This means that type `char' is represented with a 16-bit word per character. The 1750A's "Load/Store Upper/Lower Byte" instructions are not used by GNU CC. @{b}alpha-*-osf1@{ub} Systems using processors that implement the DEC Alpha architecture and are running the DEC Unix (OSF/1) operating system, for example the DEC Alpha AXP systems. (VMS on the Alpha is not currently supported by GNU CC.) GNU CC writes a @{b}.verstamp@{ub} directive to the assembler output file unless it is built as a cross-compiler. It gets the version to use from the system header file @{b}/usr/include/stamp.h@{ub}. If you install a new version of DEC Unix, you should rebuild GCC to pick up the new version stamp. Note that since the Alpha is a 64-bit architecture, cross-compilers from 32-bit machines will not generate code as efficient as that generated when the compiler is running on a 64-bit machine because many optimizations that depend on being able to represent a word on the target in an integral value on the host cannot be performed. Building cross-compilers on the Alpha for 32-bit machines has only been tested in a few cases and may not work properly. @{b}make compare@{ub} may fail on old versions of DEC Unix unless you add @{b}-save-temps@{ub} to @{b}CFLAGS@{ub}. On these systems, the name of the assembler input file is stored in the object file, and that makes comparison fail if it differs between the @{b}stage1@{ub} and @{b}stage2@{ub} compilations. The option @{b}-save-temps@{ub} forces a fixed name to be used for the assembler input file, instead of a randomly chosen name in @{b}/tmp@{ub}. Do not add @{b}-save-temps@{ub} unless the comparisons fail without that option. If you add @{b}-save-temps@{ub}, you will have to manually delete the @{b}.i@{ub} and @{b}.s@{ub} files after each series of compilations. GNU CC now supports both the native (ECOFF) debugging format used by DBX and GDB and an encapsulated STABS format for use only with GDB. See the discussion of the @{b}--with-stabs@{ub} option of @{b}configure@{ub} above for more information on these formats and how to select them. There is a bug in DEC's assembler that produces incorrect line numbers for ECOFF format when the @{b}.align@{ub} directive is used. To work around this problem, GNU CC will not emit such alignment directives while writing ECOFF format debugging information even if optimization is being performed. Unfortunately, this has the very undesirable side-effect that code addresses when @{b}-O@{ub} is specified are different depending on whether or not @{b}-g@{ub} is also specified. To avoid this behavior, specify @{b}-gstabs+@{ub} and use GDB instead of DBX. DEC is now aware of this problem with the assembler and hopes to provide a fix shortly. @{b}arm@{ub} Advanced RISC Machines ARM-family processors. These are often used in embedded applications. There are no standard Unix configurations. This configuration corresponds to the basic instruction sequences and will produce a.out format object modules. You may need to make a variant of the file @{b}arm.h@{ub} for your particular configuration. @{b}arm-*-riscix@{ub} The ARM2 or ARM3 processor running RISC iX, Acorn's port of BSD Unix. If you are running a version of RISC iX prior to 1.2 then you must specify the version number during configuration. Note that the assembler shipped with RISC iX does not support stabs debugging information; a new version of the assembler, with stabs support included, is now available from Acorn. @{b}a29k@{ub} AMD Am29k-family processors. These are normally used in embedded applications. There are no standard Unix configurations. This configuration corresponds to AMD's standard calling sequence and binary interface and is compatible with other 29k tools. You may need to make a variant of the file @{b}a29k.h@{ub} for your particular configuration. @{b}a29k-*-bsd@{ub} AMD Am29050 used in a system running a variant of BSD Unix. @{b}decstation-*@{ub} DECstations can support three different personalities: Ultrix, DEC OSF/1, and OSF/rose. To configure GCC for these platforms use the following configurations: @{b}decstation-ultrix@{ub} Ultrix configuration. @{b}decstation-osf1@{ub} Dec's version of OSF/1. @{b}decstation-osfrose@{ub} Open Software Foundation reference port of OSF/1 which uses the OSF/rose object file format instead of ECOFF. Normally, you would not select this configuration. The MIPS C compiler needs to be told to increase its table size for switch statements with the @{b}-Wf,-XNg1500@{ub} option in order to compile @{b}cp/parse.c@{ub}. If you use the @{b}-O2@{ub} optimization option, you also need to use @{b}-Olimit 3000@{ub}. Both of these options are automatically generated in the @{b}Makefile@{ub} that the shell script @{b}configure@{ub} builds. If you override the @{b}CC@{ub} make variable and use the MIPS compilers, you may need to add @{b}-Wf,-XNg1500 -Olimit 3000@{ub}. @{b}elxsi-elxsi-bsd@{ub} The Elxsi's C compiler has known limitations that prevent it from compiling GNU C. Please contact @{b}mrs@cygnus.com@{ub} for more details. @{b}dsp16xx@{ub} A port to the AT&T DSP1610 family of processors. @{b}h8300-*-*@{ub} The calling convention and structure layout has changed in release 2.6. All code must be recompiled. The calling convention now passes the first three arguments in function calls in registers. Structures are no longer a multiple of 2 bytes. @{b}hppa*-*-*@{ub} There are several variants of the HP-PA processor which run a variety of operating systems. GNU CC must be configured to use the correct processor type and operating system, or GNU CC will not function correctly. The easiest way to handle this problem is to @{i}not@{ui} specify a target when configuring GNU CC, the @{b}configure@{ub} script will try to automatically determine the right processor type and operating system. @{b}-g@{ub} does not work on HP-UX, since that system uses a peculiar debugging format which GNU CC does not know about. However, @{b}-g@{ub} will work if you also use GAS and GDB in conjunction with GCC. We highly recommend using GAS for all HP-PA configurations. You should be using GAS-2.6 (or later) along with GDB-4.16 (or later). These can be retrieved from all the traditional GNU ftp archive sites. GAS will need to be installed into a directory before @{b}/bin@{ub}, @{b}/usr/bin@{ub}, and @{b}/usr/ccs/bin@{ub} in your search path. You should install GAS before you build GNU CC. To enable debugging, you must configure GNU CC with the @{b}--with-gnu-as@{ub} option before building. @{b}i370-*-*@{ub} This port is very preliminary and has many known bugs. We hope to have a higher-quality port for this machine soon. @{b}i386-*-linuxoldld@{ub} Use this configuration to generate a.out binaries on Linux-based GNU systems, if you do not have gas/binutils version 2.5.2 or later installed. This is an obsolete configuration. @{b}i386-*-linuxaout@{ub} Use this configuration to generate a.out binaries on Linux-based GNU systems. This configuration is being superseded. You must use gas/binutils version 2.5.2 or later. @{b}i386-*-linux@{ub} Use this configuration to generate ELF binaries on Linux-based GNU systems. You must use gas/binutils version 2.5.2 or later. @{b}i386-*-sco@{ub} Compilation with RCC is recommended. Also, it may be a good idea to link with GNU malloc instead of the malloc that comes with the system. @{b}i386-*-sco3.2v4@{ub} Use this configuration for SCO release 3.2 version 4. @{b}i386-*-isc@{ub} It may be a good idea to link with GNU malloc instead of the malloc that comes with the system. In ISC version 4.1, @{b}sed@{ub} core dumps when building @{b}deduced.h@{ub}. Use the version of @{b}sed@{ub} from version 4.0. @{b}i386-*-esix@{ub} It may be good idea to link with GNU malloc instead of the malloc that comes with the system. @{b}i386-ibm-aix@{ub} You need to use GAS version 2.1 or later, and and LD from GNU binutils version 2.2 or later. @{b}i386-sequent-bsd@{ub} Go to the Berkeley universe before compiling. In addition, you probably need to create a file named @{b}string.h@{ub} containing just one line: @{b}#include @{ub}. @{b}i386-sequent-ptx1*@{ub} Sequent DYNIX/ptx 1.x. @{b}i386-sequent-ptx2*@{ub} Sequent DYNIX/ptx 2.x. @{b}i386-sun-sunos4@{ub} You may find that you need another version of GNU CC to begin bootstrapping with, since the current version when built with the system's own compiler seems to get an infinite loop compiling part of @{b}libgcc2.c@{ub}. GNU CC version 2 compiled with GNU CC (any version) seems not to have this problem. See @{"Sun Install" Link "Sun Install"}, for information on installing GNU CC on Sun systems. @{b}i[345]86-*-winnt3.5@{ub} This version requires a GAS that has not let been released. Until it is, you can get a prebuilt binary version via anonymous ftp from @{b}cs.washington.edu:pub/gnat@{ub} or @{b}cs.nyu.edu:pub/gnat@{ub}. You must also use the Microsoft header files from the Windows NT 3.5 SDK. Find these on the CDROM in the @{b}/mstools/h@{ub} directory dated 9/4/94. You must use a fixed version of Microsoft linker made especially for NT 3.5, which is also is available on the NT 3.5 SDK CDROM. If you do not have this linker, can you also use the linker from Visual C/C++ 1.0 or 2.0. Installing GNU CC for NT builds a wrapper linker, called @{b}ld.exe@{ub}, which mimics the behaviour of Unix @{b}ld@{ub} in the specification of libraries (@{b}-L@{ub} and @{b}-l@{ub}). @{b}ld.exe@{ub} looks for both Unix and Microsoft named libraries. For example, if you specify @{b}-lfoo@{ub}, @{b}ld.exe@{ub} will look first for @{b}libfoo.a@{ub} and then for @{b}foo.lib@{ub}. You may install GNU CC for Windows NT in one of two ways, depending on whether or not you have a Unix-like shell and various Unix-like utilities. 1. If you do not have a Unix-like shell and few Unix-like utilities, you will use a DOS style batch script called @{b}configure.bat@{ub}. Invoke it as @{b}configure winnt@{ub} from an MSDOS console window or from the program manager dialog box. @{b}configure.bat@{ub} assumes you have already installed and have in your path a Unix-like @{b}sed@{ub} program which is used to create a working @{b}Makefile@{ub} from @{b}Makefile.in@{ub}. @{b}Makefile@{ub} uses the Microsoft Nmake program maintenance utility and the Visual C/C++ V8.00 compiler to build GNU CC. You need only have the utilities @{b}sed@{ub} and @{b}touch@{ub} to use this installation method, which only automatically builds the compiler itself. You must then examine what @{b}fixinc.winnt@{ub} does, edit the header files by hand and build @{b}libgcc.a@{ub} manually. 2. The second type of installation assumes you are running a Unix-like shell, have a complete suite of Unix-like utilities in your path, and have a previous version of GNU CC already installed, either through building it via the above installation method or acquiring a pre-built binary. In this case, use the @{b}configure@{ub} script in the normal fashion. @{b}i860-intel-osf1@{ub} This is the Paragon. If you have version 1.0 of the operating system, see @{"Installation Problems" Link "Installation Problems"}, for special things you need to do to compensate for peculiarities in the system. @{b}*-lynx-lynxos@{ub} LynxOS 2.2 and earlier comes with GNU CC 1.x already installed as @{b}/bin/gcc@{ub}. You should compile with this instead of @{b}/bin/cc@{ub}. You can tell GNU CC to use the GNU assembler and linker, by specifying @{b}--with-gnu-as --with-gnu-ld@{ub} when configuring. These will produce COFF format object files and executables; otherwise GNU CC will use the installed tools, which produce a.out format executables. @{b}m68000-hp-bsd@{ub} HP 9000 series 200 running BSD. Note that the C compiler that comes with this system cannot compile GNU CC; contact @{b}law@cs.utah.edu@{ub} to get binaries of GNU CC for bootstrapping. @{b}m68k-altos@{ub} Altos 3068. You must use the GNU assembler, linker and debugger. Also, you must fix a kernel bug. Details in the file @{b}README.ALTOS@{ub}. @{b}m68k-att-sysv@{ub} AT&T 3b1, a.k.a. 7300 PC. Special procedures are needed to compile GNU CC with this machine's standard C compiler, due to bugs in that compiler. You can bootstrap it more easily with previous versions of GNU CC if you have them. Installing GNU CC on the 3b1 is difficult if you do not already have GNU CC running, due to bugs in the installed C compiler. However, the following procedure might work. We are unable to test it. 1. Comment out the @{b}#include "config.h"@{ub} line on line 37 of @{b}cccp.c@{ub} and do @{b}make cpp@{ub}. This makes a preliminary version of GNU cpp. 2. Save the old @{b}/lib/cpp@{ub} and copy the preliminary GNU cpp to that file name. 3. Undo your change in @{b}cccp.c@{ub}, or reinstall the original version, and do @{b}make cpp@{ub} again. 4. Copy this final version of GNU cpp into @{b}/lib/cpp@{ub}. 5. Replace every occurrence of @{b}obstack_free@{ub} in the file @{b}tree.c@{ub} with @{b}_obstack_free@{ub}. 6. Run @{b}make@{ub} to get the first-stage GNU CC. 7. Reinstall the original version of @{b}/lib/cpp@{ub}. 8. Now you can compile GNU CC with itself and install it in the normal fashion. @{b}m68k-bull-sysv@{ub} Bull DPX/2 series 200 and 300 with BOS-2.00.45 up to BOS-2.01. GNU CC works either with native assembler or GNU assembler. You can use GNU assembler with native coff generation by providing @{b}--with-gnu-as@{ub} to the configure script or use GNU assembler with dbx-in-coff encapsulation by providing @{b}--with-gnu-as --stabs@{ub}. For any problem with native assembler or for availability of the DPX/2 port of GAS, contact @{b}F.Pierresteguy@frcl.bull.fr@{ub}. @{b}m68k-crds-unox@{ub} Use @{b}configure unos@{ub} for building on Unos. The Unos assembler is named @{b}casm@{ub} instead of @{b}as@{ub}. For some strange reason linking @{b}/bin/as@{ub} to @{b}/bin/casm@{ub} changes the behavior, and does not work. So, when installing GNU CC, you should install the following script as @{b}as@{ub} in the subdirectory where the passes of GCC are installed: #!/bin/sh casm $* The default Unos library is named @{b}libunos.a@{ub} instead of @{b}libc.a@{ub}. To allow GNU CC to function, either change all references to @{b}-lc@{ub} in @{b}gcc.c@{ub} to @{b}-lunos@{ub} or link @{b}/lib/libc.a@{ub} to @{b}/lib/libunos.a@{ub}. When compiling GNU CC with the standard compiler, to overcome bugs in the support of @{b}alloca@{ub}, do not use @{b}-O@{ub} when making stage 2. Then use the stage 2 compiler with @{b}-O@{ub} to make the stage 3 compiler. This compiler will have the same characteristics as the usual stage 2 compiler on other systems. Use it to make a stage 4 compiler and compare that with stage 3 to verify proper compilation. (Perhaps simply defining @{b}ALLOCA@{ub} in @{b}x-crds@{ub} as described in the comments there will make the above paragraph superfluous. Please inform us of whether this works.) Unos uses memory segmentation instead of demand paging, so you will need a lot of memory. 5 Mb is barely enough if no other tasks are running. If linking @{b}cc1@{ub} fails, try putting the object files into a library and linking from that library. @{b}m68k-hp-hpux@{ub} HP 9000 series 300 or 400 running HP-UX. HP-UX version 8.0 has a bug in the assembler that prevents compilation of GNU CC. To fix it, get patch PHCO_4484 from HP. In addition, if you wish to use gas @{b}--with-gnu-as@{ub} you must use gas version 2.1 or later, and you must use the GNU linker version 2.1 or later. Earlier versions of gas relied upon a program which converted the gas output into the native HP/UX format, but that program has not been kept up to date. gdb does not understand that native HP/UX format, so you must use gas if you wish to use gdb. @{b}m68k-sun@{ub} Sun 3. We do not provide a configuration file to use the Sun FPA by default, because programs that establish signal handlers for floating point traps inherently cannot work with the FPA. See @{"Sun Install" Link "Sun Install"}, for information on installing GNU CC on Sun systems. @{b}m88k-*-svr3@{ub} Motorola m88k running the AT&T/Unisoft/Motorola V.3 reference port. These systems tend to use the Green Hills C, revision 1.8.5, as the standard C compiler. There are apparently bugs in this compiler that result in object files differences between stage 2 and stage 3. If this happens, make the stage 4 compiler and compare it to the stage 3 compiler. If the stage 3 and stage 4 object files are identical, this suggests you encountered a problem with the standard C compiler; the stage 3 and 4 compilers may be usable. It is best, however, to use an older version of GNU CC for bootstrapping if you have one. @{b}m88k-*-dgux@{ub} Motorola m88k running DG/UX. To build 88open BCS native or cross compilers on DG/UX, specify the configuration name as @{b}m88k-*-dguxbcs@{ub} and build in the 88open BCS software development environment. To build ELF native or cross compilers on DG/UX, specify @{b}m88k-*-dgux@{ub} and build in the DG/UX ELF development environment. You set the software development environment by issuing @{b}sde-target@{ub} command and specifying either @{b}m88kbcs@{ub} or @{b}m88kdguxelf@{ub} as the operand. If you do not specify a configuration name, @{b}configure@{ub} guesses the configuration based on the current software development environment. @{b}m88k-tektronix-sysv3@{ub} Tektronix XD88 running UTekV 3.2e. Do not turn on optimization while building stage1 if you bootstrap with the buggy Green Hills compiler. Also, The bundled LAI System V NFS is buggy so if you build in an NFS mounted directory, start from a fresh reboot, or avoid NFS all together. Otherwise you may have trouble getting clean comparisons between stages. @{b}mips-mips-bsd@{ub} MIPS machines running the MIPS operating system in BSD mode. It's possible that some old versions of the system lack the functions @{b}memcpy@{ub}, @{b}memcmp@{ub}, and @{b}memset@{ub}. If your system lacks these, you must remove or undo the definition of @{b}TARGET_MEM_FUNCTIONS@{ub} in @{b}mips-bsd.h@{ub}. The MIPS C compiler needs to be told to increase its table size for switch statements with the @{b}-Wf,-XNg1500@{ub} option in order to compile @{b}cp/parse.c@{ub}. If you use the @{b}-O2@{ub} optimization option, you also need to use @{b}-Olimit 3000@{ub}. Both of these options are automatically generated in the @{b}Makefile@{ub} that the shell script @{b}configure@{ub} builds. If you override the @{b}CC@{ub} make variable and use the MIPS compilers, you may need to add @{b}-Wf,-XNg1500 -Olimit 3000@{ub}. @{b}mips-mips-riscos*@{ub} The MIPS C compiler needs to be told to increase its table size for switch statements with the @{b}-Wf,-XNg1500@{ub} option in order to compile @{b}cp/parse.c@{ub}. If you use the @{b}-O2@{ub} optimization option, you also need to use @{b}-Olimit 3000@{ub}. Both of these options are automatically generated in the @{b}Makefile@{ub} that the shell script @{b}configure@{ub} builds. If you override the @{b}CC@{ub} make variable and use the MIPS compilers, you may need to add @{b}-Wf,-XNg1500 -Olimit 3000@{ub}. MIPS computers running RISC-OS can support four different personalities: default, BSD 4.3, System V.3, and System V.4 (older versions of RISC-OS don't support V.4). To configure GCC for these platforms use the following configurations: @{b}mips-mips-riscos@{b}rev@{ub}@{ub} Default configuration for RISC-OS, revision @{b}rev@{ub}. @{b}mips-mips-riscos@{b}rev@{ub}bsd@{ub} BSD 4.3 configuration for RISC-OS, revision @{b}rev@{ub}. @{b}mips-mips-riscos@{b}rev@{ub}sysv4@{ub} System V.4 configuration for RISC-OS, revision @{b}rev@{ub}. @{b}mips-mips-riscos@{b}rev@{ub}sysv@{ub} System V.3 configuration for RISC-OS, revision @{b}rev@{ub}. The revision @{b}rev@{ub} mentioned above is the revision of RISC-OS to use. You must reconfigure GCC when going from a RISC-OS revision 4 to RISC-OS revision 5. This has the effect of avoiding a linker bug (see @{"Installation Problems" Link "Installation Problems"}, for more details). @{b}mips-sgi-*@{ub} In order to compile GCC on an SGI running IRIX 4, the "c.hdr.lib" option must be installed from the CD-ROM supplied from Silicon Graphics. This is found on the 2nd CD in release 4.0.1. In order to compile GCC on an SGI running IRIX 5, the "compiler_dev.hdr" subsystem must be installed from the IDO CD-ROM supplied by Silicon Graphics. @{b}make compare@{ub} may fail on version 5 of IRIX unless you add @{b}-save-temps@{ub} to @{b}CFLAGS@{ub}. On these systems, the name of the assembler input file is stored in the object file, and that makes comparison fail if it differs between the @{b}stage1@{ub} and @{b}stage2@{ub} compilations. The option @{b}-save-temps@{ub} forces a fixed name to be used for the assembler input file, instead of a randomly chosen name in @{b}/tmp@{ub}. Do not add @{b}-save-temps@{ub} unless the comparisons fail without that option. If you do you @{b}-save-temps@{ub}, you will have to manually delete the @{b}.i@{ub} and @{b}.s@{ub} files after each series of compilations. The MIPS C compiler needs to be told to increase its table size for switch statements with the @{b}-Wf,-XNg1500@{ub} option in order to compile @{b}cp/parse.c@{ub}. If you use the @{b}-O2@{ub} optimization option, you also need to use @{b}-Olimit 3000@{ub}. Both of these options are automatically generated in the @{b}Makefile@{ub} that the shell script @{b}configure@{ub} builds. If you override the @{b}CC@{ub} make variable and use the MIPS compilers, you may need to add @{b}-Wf,-XNg1500 -Olimit 3000@{ub}. On Irix version 4.0.5F, and perhaps on some other versions as well, there is an assembler bug that reorders instructions incorrectly. To work around it, specify the target configuration @{b}mips-sgi-irix4loser@{ub}. This configuration inhibits assembler optimization. In a compiler configured with target @{b}mips-sgi-irix4@{ub}, you can turn off assembler optimization by using the @{b}-noasmopt@{ub} option. This compiler option passes the option @{b}-O0@{ub} to the assembler, to inhibit reordering. The @{b}-noasmopt@{ub} option can be useful for testing whether a problem is due to erroneous assembler reordering. Even if a problem does not go away with @{b}-noasmopt@{ub}, it may still be due to assembler reordering--perhaps GNU CC itself was miscompiled as a result. To enable debugging under Irix 5, you must use GNU as 2.5 or later, and use the @{b}--with-gnu-as@{ub} configure option when configuring gcc. GNU as is distributed as part of the binutils package. @{b}mips-sony-sysv@{ub} Sony MIPS NEWS. This works in NEWSOS 5.0.1, but not in 5.0.2 (which uses ELF instead of COFF). Support for 5.0.2 will probably be provided soon by volunteers. In particular, the linker does not like the code generated by GCC when shared libraries are linked in. @{b}ns32k-encore@{ub} Encore ns32000 system. Encore systems are supported only under BSD. @{b}ns32k-*-genix@{ub} National Semiconductor ns32000 system. Genix has bugs in @{b}alloca@{ub} and @{b}malloc@{ub}; you must get the compiled versions of these from GNU Emacs. @{b}ns32k-sequent@{ub} Go to the Berkeley universe before compiling. In addition, you probably need to create a file named @{b}string.h@{ub} containing just one line: @{b}#include @{ub}. @{b}ns32k-utek@{ub} UTEK ns32000 system ("merlin"). The C compiler that comes with this system cannot compile GNU CC; contact @{b}tektronix!reed!mason@{ub} to get binaries of GNU CC for bootstrapping. @{b}romp-*-aos@{ub} @{b}romp-*-mach@{ub} The only operating systems supported for the IBM RT PC are AOS and MACH. GNU CC does not support AIX running on the RT. We recommend you compile GNU CC with an earlier version of itself; if you compile GNU CC with @{b}hc@{ub}, the Metaware compiler, it will work, but you will get mismatches between the stage 2 and stage 3 compilers in various files. These errors are minor differences in some floating-point constants and can be safely ignored; the stage 3 compiler is correct. @{b}rs6000-*-aix@{ub} @{b}powerpc-*-aix@{ub} Various early versions of each release of the IBM XLC compiler will not bootstrap GNU CC. Symptoms include differences between the stage2 and stage3 object files, and errors when compiling @{b}libgcc.a@{ub} or @{b}enquire@{ub}. Known problematic releases include: xlc-1.2.1.8, xlc-1.3.0.0 (distributed with AIX 3.2.5), and xlc-1.3.0.19. Both xlc-1.2.1.28 and xlc-1.3.0.24 (PTF 432238) are known to produce working versions of GNU CC, but most other recent releases correctly bootstrap GNU CC. Also, releases of AIX prior to AIX 3.2.4 include a version of the IBM assembler which does not accept debugging directives: assembler updates are available as PTFs. Also, if you are using AIX 3.2.5 or greater and the GNU assembler, you must have a version modified after October 16th, 1995 in order for the GNU C compiler to build. See the file @{b}README.RS6000@{ub} for more details on of these problems. GNU CC does not yet support the 64-bit PowerPC instructions. Objective C does not work on this architecture because it makes assumptions that are incompatible with the calling conventions. AIX on the RS/6000 provides support (NLS) for environments outside of the United States. Compilers and assemblers use NLS to support locale-specific representations of various objects including floating-point numbers ("." vs "," for separating decimal fractions). There have been problems reported where the library linked with GNU CC does not produce the same floating-point formats that the assembler accepts. If you have this problem, set the LANG environment variable to "C" or "En_US". Due to changes in the way that GNU CC invokes the binder (linker) for AIX 4.1, you may now receive warnings of duplicate symbols from the link step that were not reported before. The assembly files generated by GNU CC for AIX have always included multiple symbol definitions for certain global variable and function declarations in the original program. The warnings should not prevent the linker from producing a correct library or runnable executable. @{b}powerpc-*-elf@{ub} @{b}powerpc-*-sysv4@{ub} PowerPC system in big endian mode, running System V.4. This configuration is currently under development. @{b}powerpc-*-eabiaix@{ub} Embedded PowerPC system in big endian mode with -mcall-aix selected as the default. This system is currently under development. @{b}powerpc-*-eabisim@{ub} Embedded PowerPC system in big endian mode for use in running under the PSIM simulator. This system is currently under development. @{b}powerpc-*-eabi@{ub} Embedded PowerPC system in big endian mode. This configuration is currently under development. @{b}powerpcle-*-elf@{ub} @{b}powerpcle-*-sysv4@{ub} PowerPC system in little endian mode, running System V.4. This configuration is currently under development. @{b}powerpcle-*-sysv4@{ub} Embedded PowerPC system in little endian mode. This system is currently under development. @{b}powerpcle-*-eabisim@{ub} Embedded PowerPC system in little endian mode for use in running under the PSIM simulator. This system is currently under development. @{b}powerpcle-*-eabi@{ub} Embedded PowerPC system in little endian mode. This configuration is currently under development. @{b}vax-dec-ultrix@{ub} Don't try compiling with Vax C (@{b}vcc@{ub}). It produces incorrect code in some cases (for example, when @{b}alloca@{ub} is used). Meanwhile, compiling @{b}cp/parse.c@{ub} with pcc does not work because of an internal table size limitation in that compiler. To avoid this problem, compile just the GNU C compiler first, and use it to recompile building all the languages that you want to run. @{b}sparc-sun-*@{ub} See @{"Sun Install" Link "Sun Install"}, for information on installing GNU CC on Sun systems. @{b}vax-dec-vms@{ub} See @{"VMS Install" Link "VMS Install"}, for details on how to install GNU CC on VMS. @{b}we32k-*-*@{ub} These computers are also known as the 3b2, 3b5, 3b20 and other similar names. (However, the 3b1 is actually a 68000; see @{"Configurations" Link "Configurations"}.) Don't use @{b}-g@{ub} when compiling with the system's compiler. The system's linker seems to be unable to handle such a large program with debugging information. The system's compiler runs out of capacity when compiling @{b}stmt.c@{ub} in GNU CC. You can work around this by building @{b}cpp@{ub} in GNU CC first, then use that instead of the system's preprocessor with the system's C compiler to compile @{b}stmt.c@{ub}. Here is how: mv /lib/cpp /lib/cpp.att cp cpp /lib/cpp.gnu echo '/lib/cpp.gnu -traditional ${1+"$@"}' > /lib/cpp chmod +x /lib/cpp The system's compiler produces bad code for some of the GNU CC optimization files. So you must build the stage 2 compiler without optimization. Then build a stage 3 compiler with optimization. That executable should work. Here are the necessary commands: make LANGUAGES=c CC=stage1/xgcc CFLAGS="-Bstage1/ -g" make stage2 make CC=stage2/xgcc CFLAGS="-Bstage2/ -g -O" You may need to raise the ULIMIT setting to build a C++ compiler, as the file @{b}cc1plus@{ub} is larger than one megabyte. @EndNode @Node "Other Dir" "gcc.guide/Other Dir" @Next "Cross-Compiler" @Prev "Configurations" @Toc "Installation" Compilation in a Separate Directory =================================== If you wish to build the object files and executables in a directory other than the one containing the source files, here is what you must do differently: 1. Make sure you have a version of Make that supports the @{b}VPATH@{ub} feature. (GNU Make supports it, as do Make versions on most BSD systems.) 2. If you have ever run @{b}configure@{ub} in the source directory, you must undo the configuration. Do this by running: make distclean 3. Go to the directory in which you want to build the compiler before running @{b}configure@{ub}: mkdir gcc-sun3 cd gcc-sun3 On systems that do not support symbolic links, this directory must be on the same file system as the source code directory. 4. Specify where to find @{b}configure@{ub} when you run it: ../gcc/configure ... This also tells @{b}configure@{ub} where to find the compiler sources; @{b}configure@{ub} takes the directory from the file name that was used to invoke it. But if you want to be sure, you can specify the source directory with the @{b}--srcdir@{ub} option, like this: ../gcc/configure --srcdir=../gcc @{I}OTHER OPTIONS@{ui} The directory you specify with @{b}--srcdir@{ub} need not be the same as the one that @{b}configure@{ub} is found in. Now, you can run @{b}make@{ub} in that directory. You need not repeat the configuration steps shown above, when ordinary source files change. You must, however, run @{b}configure@{ub} again when the configuration files change, if your system does not support symbolic links. @EndNode @Node "Cross-Compiler" "gcc.guide/Cross-Compiler" @Next "Sun Install" @Prev "Other Dir" @Toc "Installation" Building and Installing a Cross-Compiler ======================================== GNU CC can function as a cross-compiler for many machines, but not all. @{b}*@{ub} Cross-compilers for the Mips as target using the Mips assembler currently do not work, because the auxiliary programs @{b}mips-tdump.c@{ub} and @{b}mips-tfile.c@{ub} can't be compiled on anything but a Mips. It does work to cross compile for a Mips if you use the GNU assembler and linker. @{b}*@{ub} Cross-compilers between machines with different floating point formats have not all been made to work. GNU CC now has a floating point emulator with which these can work, but each target machine description needs to be updated to take advantage of it. @{b}*@{ub} Cross-compilation between machines of different word sizes is somewhat problematic and sometimes does not work. Since GNU CC generates assembler code, you probably need a cross-assembler that GNU CC can run, in order to produce object files. If you want to link on other than the target machine, you need a cross-linker as well. You also need header files and libraries suitable for the target machine that you can install on the host machine. @{" Steps of Cross " Link "Steps of Cross"} Using a cross-compiler involves several steps that may be carried out on different machines. @{" Configure Cross " Link "Configure Cross"} Configuring a cross-compiler. @{" Tools and Libraries " Link "Tools and Libraries"} Where to put the linker and assembler, and the C library. @{" Cross Headers " Link "Cross Headers"} Finding and installing header files for a cross-compiler. @{" Cross Runtime " Link "Cross Runtime"} Supplying arithmetic runtime routines (@{b}libgcc1.a@{ub}). @{" Build Cross " Link "Build Cross"} Actually compiling the cross-compiler. @EndNode @Node "Steps of Cross" "gcc.guide/Steps of Cross" @Next "Configure Cross" @Toc "Cross-Compiler" Steps of Cross-Compilation -------------------------- To compile and run a program using a cross-compiler involves several steps: @{b}*@{ub} Run the cross-compiler on the host machine to produce assembler files for the target machine. This requires header files for the target machine. @{b}*@{ub} Assemble the files produced by the cross-compiler. You can do this either with an assembler on the target machine, or with a cross-assembler on the host machine. @{b}*@{ub} Link those files to make an executable. You can do this either with a linker on the target machine, or with a cross-linker on the host machine. Whichever machine you use, you need libraries and certain startup files (typically @{b}crt....o@{ub}) for the target machine. It is most convenient to do all of these steps on the same host machine, since then you can do it all with a single invocation of GNU CC. This requires a suitable cross-assembler and cross-linker. For some targets, the GNU assembler and linker are available. @EndNode @Node "Configure Cross" "gcc.guide/Configure Cross" @Next "Tools and Libraries" @Prev "Steps of Cross" @Toc "Cross-Compiler" Configuring a Cross-Compiler ---------------------------- To build GNU CC as a cross-compiler, you start out by running @{b}configure@{ub}. Use the @{b}--target=@{I}TARGET@{ui}@{ub} to specify the target type. If @{b}configure@{ub} was unable to correctly identify the system you are running on, also specify the @{b}--build=@{I}BUILD@{ui}@{ub} option. For example, here is how to configure for a cross-compiler that produces code for an HP 68030 system running BSD on a system that @{b}configure@{ub} can correctly identify: ./configure --target=m68k-hp-bsd4.3 @EndNode @Node "Tools and Libraries" "gcc.guide/Tools and Libraries" @Next "Cross Headers" @Prev "Configure Cross" @Toc "Cross-Compiler" Tools and Libraries for a Cross-Compiler ---------------------------------------- If you have a cross-assembler and cross-linker available, you should install them now. Put them in the directory @{b}/usr/local/@{I}TARGET@{ui}/bin@{ub}. Here is a table of the tools you should put in this directory: @{b}as@{ub} This should be the cross-assembler. @{b}ld@{ub} This should be the cross-linker. @{b}ar@{ub} This should be the cross-archiver: a program which can manipulate archive files (linker libraries) in the target machine's format. @{b}ranlib@{ub} This should be a program to construct a symbol table in an archive file. The installation of GNU CC will find these programs in that directory, and copy or link them to the proper place to for the cross-compiler to find them when run later. The easiest way to provide these files is to build the Binutils package and GAS. Configure them with the same @{b}--host@{ub} and @{b}--target@{ub} options that you use for configuring GNU CC, then build and install them. They install their executables automatically into the proper directory. Alas, they do not support all the targets that GNU CC supports. If you want to install libraries to use with the cross-compiler, such as a standard C library, put them in the directory @{b}/usr/local/@{I}TARGET@{ui}/lib@{ub}; installation of GNU CC copies all all the files in that subdirectory into the proper place for GNU CC to find them and link with them. Here's an example of copying some libraries from a target machine: ftp @{I}TARGET-MACHINE@{ui} lcd /usr/local/@{I}TARGET@{ui}/lib cd /lib get libc.a cd /usr/lib get libg.a get libm.a quit The precise set of libraries you'll need, and their locations on the target machine, vary depending on its operating system. Many targets require "start files" such as @{b}crt0.o@{ub} and @{b}crtn.o@{ub} which are linked into each executable; these too should be placed in @{b}/usr/local/@{I}TARGET@{ui}/lib@{ub}. There may be several alternatives for @{b}crt0.o@{ub}, for use with profiling or other compilation options. Check your target's definition of @{b}STARTFILE_SPEC@{ub} to find out what start files it uses. Here's an example of copying these files from a target machine: ftp @{I}TARGET-MACHINE@{ui} lcd /usr/local/@{I}TARGET@{ui}/lib prompt cd /lib mget *crt*.o cd /usr/lib mget *crt*.o quit @EndNode @Node "Cross Runtime" "gcc.guide/Cross Runtime" @Next "Build Cross" @Prev "Cross Headers" @Toc "Cross-Compiler" @{b}libgcc.a@{ub} and Cross-Compilers ---------------------------- Code compiled by GNU CC uses certain runtime support functions implicitly. Some of these functions can be compiled successfully with GNU CC itself, but a few cannot be. These problem functions are in the source file @{b}libgcc1.c@{ub}; the library made from them is called @{b}libgcc1.a@{ub}. When you build a native compiler, these functions are compiled with some other compiler-the one that you use for bootstrapping GNU CC. Presumably it knows how to open code these operations, or else knows how to call the run-time emulation facilities that the machine comes with. But this approach doesn't work for building a cross-compiler. The compiler that you use for building knows about the host system, not the target system. So, when you build a cross-compiler you have to supply a suitable library @{b}libgcc1.a@{ub} that does the job it is expected to do. To compile @{b}libgcc1.c@{ub} with the cross-compiler itself does not work. The functions in this file are supposed to implement arithmetic operations that GNU CC does not know how to open code for your target machine. If these functions are compiled with GNU CC itself, they will compile into infinite recursion. On any given target, most of these functions are not needed. If GNU CC can open code an arithmetic operation, it will not call these functions to perform the operation. It is possible that on your target machine, none of these functions is needed. If so, you can supply an empty library as @{b}libgcc1.a@{ub}. Many targets need library support only for multiplication and division. If you are linking with a library that contains functions for multiplication and division, you can tell GNU CC to call them directly by defining the macros @{b}MULSI3_LIBCALL@{ub}, and the like. These macros need to be defined in the target description macro file. For some targets, they are defined already. This may be sufficient to avoid the need for libgcc1.a; if so, you can supply an empty library. Some targets do not have floating point instructions; they need other functions in @{b}libgcc1.a@{ub}, which do floating arithmetic. Recent versions of GNU CC have a file which emulates floating point. With a certain amount of work, you should be able to construct a floating point emulator that can be used as @{b}libgcc1.a@{ub}. Perhaps future versions will contain code to do this automatically and conveniently. That depends on whether someone wants to implement it. Some embedded targets come with all the necessary @{b}libgcc1.a@{ub} routines written in C or assembler. These targets build @{b}libgcc1.a@{ub} automatically and you do not need to do anything special for them. Other embedded targets do not need any @{b}libgcc1.a@{ub} routines since all the necessary operations are supported by the hardware. If your target system has another C compiler, you can configure GNU CC as a native compiler on that machine, build just @{b}libgcc1.a@{ub} with @{b}make libgcc1.a@{ub} on that machine, and use the resulting file with the cross-compiler. To do this, execute the following on the target machine: cd @{I}TARGET-BUILD-DIR@{ui} ./configure --host=sparc --target=sun3 make libgcc1.a And then this on the host machine: ftp @{I}TARGET-MACHINE@{ui} binary cd @{I}TARGET-BUILD-DIR@{ui} get libgcc1.a quit Another way to provide the functions you need in @{b}libgcc1.a@{ub} is to define the appropriate @{b}perform_...@{ub} macros for those functions. If these definitions do not use the C arithmetic operators that they are meant to implement, you should be able to compile them with the cross-compiler you are building. (If these definitions already exist for your target file, then you are all set.) To build @{b}libgcc1.a@{ub} using the perform macros, use @{b}LIBGCC1=libgcc1.a OLDCC=./xgcc@{ub} when building the compiler. Otherwise, you should place your replacement library under the name @{b}libgcc1.a@{ub} in the directory in which you will build the cross-compiler, before you run @{b}make@{ub}. @EndNode @Node "Cross Headers" "gcc.guide/Cross Headers" @Next "Cross Runtime" @Prev "Tools and Libraries" @Toc "Cross-Compiler" Cross-Compilers and Header Files -------------------------------- If you are cross-compiling a standalone program or a program for an embedded system, then you may not need any header files except the few that are part of GNU CC (and those of your program). However, if you intend to link your program with a standard C library such as @{b}libc.a@{ub}, then you probably need to compile with the header files that go with the library you use. The GNU C compiler does not come with these files, because (1) they are system-specific, and (2) they belong in a C library, not in a compiler. If the GNU C library supports your target machine, then you can get the header files from there (assuming you actually use the GNU library when you link your program). If your target machine comes with a C compiler, it probably comes with suitable header files also. If you make these files accessible from the host machine, the cross-compiler can use them also. Otherwise, you're on your own in finding header files to use when cross-compiling. When you have found suitable header files, put them in @{b}/usr/local/@{I}TARGET@{ui}/include@{ub}, before building the cross compiler. Then installation will run fixincludes properly and install the corrected versions of the header files where the compiler will use them. Provide the header files before you build the cross-compiler, because the build stage actually runs the cross-compiler to produce parts of @{b}libgcc.a@{ub}. (These are the parts that @{i}can@{ui} be compiled with GNU CC.) Some of them need suitable header files. Here's an example showing how to copy the header files from a target machine. On the target machine, do this: (cd /usr/include; tar cf - .) > tarfile Then, on the host machine, do this: ftp @{I}TARGET-MACHINE@{ui} lcd /usr/local/@{I}TARGET@{ui}/include get tarfile quit tar xf tarfile @EndNode @Node "Build Cross" "gcc.guide/Build Cross" @Prev "Cross Runtime" @Toc "Cross-Compiler" Actually Building the Cross-Compiler ------------------------------------ Now you can proceed just as for compiling a single-machine compiler through the step of building stage 1. If you have not provided some sort of @{b}libgcc1.a@{ub}, then compilation will give up at the point where it needs that file, printing a suitable error message. If you do provide @{b}libgcc1.a@{ub}, then building the compiler will automatically compile and link a test program called @{b}libgcc1-test@{ub}; if you get errors in the linking, it means that not all of the necessary routines in @{b}libgcc1.a@{ub} are available. You must provide the header file @{b}float.h@{ub}. One way to do this is to compile @{b}enquire@{ub} and run it on your target machine. The job of @{b}enquire@{ub} is to run on the target machine and figure out by experiment the nature of its floating point representation. @{b}enquire@{ub} records its findings in the header file @{b}float.h@{ub}. If you can't produce this file by running @{b}enquire@{ub} on the target machine, then you will need to come up with a suitable @{b}float.h@{ub} in some other way (or else, avoid using it in your programs). Do not try to build stage 2 for a cross-compiler. It doesn't work to rebuild GNU CC as a cross-compiler using the cross-compiler, because that would produce a program that runs on the target machine, not on the host. For example, if you compile a 386-to-68030 cross-compiler with itself, the result will not be right either for the 386 (because it was compiled into 68030 code) or for the 68030 (because it was configured for a 386 as the host). If you want to compile GNU CC into 68030 code, whether you compile it on a 68030 or with a cross-compiler on a 386, you must specify a 68030 as the host when you configure it. To install the cross-compiler, use @{b}make install@{ub}, as usual. @EndNode @Node "Sun Install" "gcc.guide/Sun Install" @Next "VMS Install" @Prev "Cross-Compiler" @Toc "Installation" Installing GNU CC on the Sun ============================ On Solaris (version 2.1), do not use the linker or other tools in @{b}/usr/ucb@{ub} to build GNU CC. Use @{b}/usr/ccs/bin@{ub}. Make sure the environment variable @{b}FLOAT_OPTION@{ub} is not set when you compile @{b}libgcc.a@{ub}. If this option were set to @{b}f68881@{ub} when @{b}libgcc.a@{ub} is compiled, the resulting code would demand to be linked with a special startup file and would not link properly without special pains. There is a bug in @{b}alloca@{ub} in certain versions of the Sun library. To avoid this bug, install the binaries of GNU CC that were compiled by GNU CC. They use @{b}alloca@{ub} as a built-in function and never the one in the library. Some versions of the Sun compiler crash when compiling GNU CC. The problem is a segmentation fault in cpp. This problem seems to be due to the bulk of data in the environment variables. You may be able to avoid it by using the following command to compile GNU CC with Sun CC: make CC="TERMCAP=x OBJS=x LIBFUNCS=x STAGESTUFF=x cc" @EndNode @Node "VMS Install" "gcc.guide/VMS Install" @Next "Collect2" @Prev "Sun Install" @Toc "Installation" Installing GNU CC on VMS ======================== The VMS version of GNU CC is distributed in a backup saveset containing both source code and precompiled binaries. To install the @{b}gcc@{ub} command so you can use the compiler easily, in the same manner as you use the VMS C compiler, you must install the VMS CLD file for GNU CC as follows: 1. Define the VMS logical names @{b}GNU_CC@{ub} and @{b}GNU_CC_INCLUDE@{ub} to point to the directories where the GNU CC executables (@{b}gcc-cpp.exe@{ub}, @{b}gcc-cc1.exe@{ub}, etc.) and the C include files are kept respectively. This should be done with the commands: $ assign /system /translation=concealed - disk:[gcc.] gnu_cc $ assign /system /translation=concealed - disk:[gcc.include.] gnu_cc_include with the appropriate disk and directory names. These commands can be placed in your system startup file so they will be executed whenever the machine is rebooted. You may, if you choose, do this via the @{b}GCC_INSTALL.COM@{ub} script in the @{b}[GCC]@{ub} directory. 2. Install the @{b}GCC@{ub} command with the command line: $ set command /table=sys$common:[syslib]dcltables - /output=sys$common:[syslib]dcltables gnu_cc:[000000]gcc $ install replace sys$common:[syslib]dcltables 3. To install the help file, do the following: $ library/help sys$library:helplib.hlb gcc.hlp Now you can invoke the compiler with a command like @{b}gcc /verbose file.c@{ub}, which is equivalent to the command @{b}gcc -v -c file.c@{ub} in Unix. If you wish to use GNU C++ you must first install GNU CC, and then perform the following steps: 1. Define the VMS logical name @{b}GNU_GXX_INCLUDE@{ub} to point to the directory where the preprocessor will search for the C++ header files. This can be done with the command: $ assign /system /translation=concealed - disk:[gcc.gxx_include.] gnu_gxx_include with the appropriate disk and directory name. If you are going to be using libg++, this is where the libg++ install procedure will install the libg++ header files. 2. Obtain the file @{b}gcc-cc1plus.exe@{ub}, and place this in the same directory that @{b}gcc-cc1.exe@{ub} is kept. The GNU C++ compiler can be invoked with a command like @{b}gcc /plus /verbose file.cc@{ub}, which is equivalent to the command @{b}g++ -v -c file.cc@{ub} in Unix. We try to put corresponding binaries and sources on the VMS distribution tape. But sometimes the binaries will be from an older version than the sources, because we don't always have time to update them. (Use the @{b}/version@{ub} option to determine the version number of the binaries and compare it with the source file @{b}version.c@{ub} to tell whether this is so.) In this case, you should use the binaries you get to recompile the sources. If you must recompile, here is how: 1. Execute the command procedure @{b}vmsconfig.com@{ub} to set up the files @{b}tm.h@{ub}, @{b}config.h@{ub}, @{b}aux-output.c@{ub}, and @{b}md.@{ub}, and to create files @{b}tconfig.h@{ub} and @{b}hconfig.h@{ub}. This procedure also creates several linker option files used by @{b}make-cc1.com@{ub} and a data file used by @{b}make-l2.com@{ub}. $ @vmsconfig.com 2. Setup the logical names and command tables as defined above. In addition, define the VMS logical name @{b}GNU_BISON@{ub} to point at the to the directories where the Bison executable is kept. This should be done with the command: $ assign /system /translation=concealed - disk:[bison.] gnu_bison You may, if you choose, use the @{b}INSTALL_BISON.COM@{ub} script in the @{b}[BISON]@{ub} directory. 3. Install the @{b}BISON@{ub} command with the command line: $ set command /table=sys$common:[syslib]dcltables - /output=sys$common:[syslib]dcltables - gnu_bison:[000000]bison $ install replace sys$common:[syslib]dcltables 4. Type @{b} @make-gcc@{ub} to recompile everything (alternatively, submit the file @{b}make-gcc.com@{ub} to a batch queue). If you wish to build the GNU C++ compiler as well as the GNU CC compiler, you must first edit @{b}make-gcc.com@{ub} and follow the instructions that appear in the comments. 5. In order to use GCC, you need a library of functions which GCC compiled code will call to perform certain tasks, and these functions are defined in the file @{b}libgcc2.c@{ub}. To compile this you should use the command procedure @{b}make-l2.com@{ub}, which will generate the library @{b}libgcc2.olb@{ub}. @{b}libgcc2.olb@{ub} should be built using the compiler built from the same distribution that @{b}libgcc2.c@{ub} came from, and @{b}make-gcc.com@{ub} will automatically do all of this for you. To install the library, use the following commands: $ library gnu_cc:[000000]gcclib/delete=(new,eprintf) $ library gnu_cc:[000000]gcclib/delete=L_* $ library libgcc2/extract=*/output=libgcc2.obj $ library gnu_cc:[000000]gcclib libgcc2.obj The first command simply removes old modules that will be replaced with modules from @{b}libgcc2@{ub} under different module names. The modules @{b}new@{ub} and @{b}eprintf@{ub} may not actually be present in your @{b}gcclib.olb@{ub}--if the VMS librarian complains about those modules not being present, simply ignore the message and continue on with the next command. The second command removes the modules that came from the previous version of the library @{b}libgcc2.c@{ub}. Whenever you update the compiler on your system, you should also update the library with the above procedure. 6. You may wish to build GCC in such a way that no files are written to the directory where the source files reside. An example would be the when the source files are on a read-only disk. In these cases, execute the following DCL commands (substituting your actual path names): $ assign dua0:[gcc.build_dir.]/translation=concealed, - dua1:[gcc.source_dir.]/translation=concealed gcc_build $ set default gcc_build:[000000] where the directory @{b}dua1:[gcc.source_dir]@{ub} contains the source code, and the directory @{b}dua0:[gcc.build_dir]@{ub} is meant to contain all of the generated object files and executables. Once you have done this, you can proceed building GCC as described above. (Keep in mind that @{b}gcc_build@{ub} is a rooted logical name, and thus the device names in each element of the search list must be an actual physical device name rather than another rooted logical name). 7. @{i}If you are building GNU CC with a previous version of GNU CC, you also should check to see that you have the newest version of the assembler@{ui}. In particular, GNU CC version 2 treats global constant variables slightly differently from GNU CC version 1, and GAS version 1.38.1 does not have the patches required to work with GCC version 2. If you use GAS 1.38.1, then @{b}extern const@{ub} variables will not have the read-only bit set, and the linker will generate warning messages about mismatched psect attributes for these variables. These warning messages are merely a nuisance, and can safely be ignored. If you are compiling with a version of GNU CC older than 1.33, specify @{b}/DEFINE=("inline=")@{ub} as an option in all the compilations. This requires editing all the @{b}gcc@{ub} commands in @{b}make-cc1.com@{ub}. (The older versions had problems supporting @{b}inline@{ub}.) Once you have a working 1.33 or newer GNU CC, you can change this file back. 8. If you want to build GNU CC with the VAX C compiler, you will need to make minor changes in @{b}make-cccp.com@{ub} and @{b}make-cc1.com@{ub} to choose alternate definitions of @{b}CC@{ub}, @{b}CFLAGS@{ub}, and @{b}LIBS@{ub}. See comments in those files. However, you must also have a working version of the GNU assembler (GNU as, aka GAS) as it is used as the back-end for GNU CC to produce binary object modules and is not included in the GNU CC sources. GAS is also needed to compile @{b}libgcc2@{ub} in order to build @{b}gcclib@{ub} (see above); @{b}make-l2.com@{ub} expects to be able to find it operational in @{b}gnu_cc:[000000]gnu-as.exe@{ub}. To use GNU CC on VMS, you need the VMS driver programs @{b}gcc.exe@{ub}, @{b}gcc.com@{ub}, and @{b}gcc.cld@{ub}. They are distributed with the VMS binaries (@{b}gcc-vms@{ub}) rather than the GNU CC sources. GAS is also included in @{b}gcc-vms@{ub}, as is Bison. Once you have successfully built GNU CC with VAX C, you should use the resulting compiler to rebuild itself. Before doing this, be sure to restore the @{b}CC@{ub}, @{b}CFLAGS@{ub}, and @{b}LIBS@{ub} definitions in @{b}make-cccp.com@{ub} and @{b}make-cc1.com@{ub}. The second generation compiler will be able to take advantage of many optimizations that must be suppressed when building with other compilers. Under previous versions of GNU CC, the generated code would occasionally give strange results when linked with the sharable @{b}VAXCRTL@{ub} library. Now this should work. Even with this version, however, GNU CC itself should not be linked with the sharable @{b}VAXCRTL@{ub}. The version of @{b}qsort@{ub} in @{b}VAXCRTL@{ub} has a bug (known to be present in VMS versions V4.6 through V5.5) which causes the compiler to fail. The executables are generated by @{b}make-cc1.com@{ub} and @{b}make-cccp.com@{ub} use the object library version of @{b}VAXCRTL@{ub} in order to make use of the @{b}qsort@{ub} routine in @{b}gcclib.olb@{ub}. If you wish to link the compiler executables with the shareable image version of @{b}VAXCRTL@{ub}, you should edit the file @{b}tm.h@{ub} (created by @{b}vmsconfig.com@{ub}) to define the macro @{b}QSORT_WORKAROUND@{ub}. @{b}QSORT_WORKAROUND@{ub} is always defined when GNU CC is compiled with VAX C, to avoid a problem in case @{b}gcclib.olb@{ub} is not yet available. @EndNode @Node "Collect2" "gcc.guide/Collect2" @Next "Header Dirs" @Prev "VMS Install" @Toc "Installation" @{b}collect2@{ub} ======== Many target systems do not have support in the assembler and linker for "constructors"--initialization functions to be called before the official "start" of @{b}main@{ub}. On such systems, GNU CC uses a utility called @{b}collect2@{ub} to arrange to call these functions at start time. The program @{b}collect2@{ub} works by linking the program once and looking through the linker output file for symbols with particular names indicating they are constructor functions. If it finds any, it creates a new temporary @{b}.c@{ub} file containing a table of them, compiles it, and links the program a second time including that file. The actual calls to the constructors are carried out by a subroutine called @{b}__main@{ub}, which is called (automatically) at the beginning of the body of @{b}main@{ub} (provided @{b}main@{ub} was compiled with GNU CC). Calling @{b}__main@{ub} is necessary, even when compiling C code, to allow linking C and C++ object code together. (If you use @{b}-nostdlib@{ub}, you get an unresolved reference to @{b}__main@{ub}, since it's defined in the standard GCC library. Include @{b}-lgcc@{ub} at the end of your compiler command line to resolve this reference.) The program @{b}collect2@{ub} is installed as @{b}ld@{ub} in the directory where the passes of the compiler are installed. When @{b}collect2@{ub} needs to find the @{i}real@{ui} @{b}ld@{ub}, it tries the following file names: @{b}*@{ub} @{b}real-ld@{ub} in the directories listed in the compiler's search directories. @{b}*@{ub} @{b}real-ld@{ub} in the directories listed in the environment variable @{b}PATH@{ub}. @{b}*@{ub} The file specified in the @{b}REAL_LD_FILE_NAME@{ub} configuration macro, if specified. @{b}*@{ub} @{b}ld@{ub} in the compiler's search directories, except that @{b}collect2@{ub} will not execute itself recursively. @{b}*@{ub} @{b}ld@{ub} in @{b}PATH@{ub}. "The compiler's search directories" means all the directories where @{b}gcc@{ub} searches for passes of the compiler. This includes directories that you specify with @{b}-B@{ub}. Cross-compilers search a little differently: @{b}*@{ub} @{b}real-ld@{ub} in the compiler's search directories. @{b}*@{ub} @{b}@{I}TARGET@{ui}-real-ld@{ub} in @{b}PATH@{ub}. @{b}*@{ub} The file specified in the @{b}REAL_LD_FILE_NAME@{ub} configuration macro, if specified. @{b}*@{ub} @{b}ld@{ub} in the compiler's search directories. @{b}*@{ub} @{b}@{I}TARGET@{ui}-ld@{ub} in @{b}PATH@{ub}. @{b}collect2@{ub} explicitly avoids running @{b}ld@{ub} using the file name under which @{b}collect2@{ub} itself was invoked. In fact, it remembers up a list of such names--in case one copy of @{b}collect2@{ub} finds another copy (or version) of @{b}collect2@{ub} installed as @{b}ld@{ub} in a second place in the search path. @{b}collect2@{ub} searches for the utilities @{b}nm@{ub} and @{b}strip@{ub} using the same algorithm as above for @{b}ld@{ub}. @EndNode @Node "Header Dirs" "gcc.guide/Header Dirs" @Prev "Collect2" @Toc "Installation" Standard Header File Directories ================================ @{b}GCC_INCLUDE_DIR@{ub} means the same thing for native and cross. It is where GNU CC stores its private include files, and also where GNU CC stores the fixed include files. A cross compiled GNU CC runs @{b}fixincludes@{ub} on the header files in @{b}$(tooldir)/include@{ub}. (If the cross compilation header files need to be fixed, they must be installed before GNU CC is built. If the cross compilation header files are already suitable for ANSI C and GNU CC, nothing special need be done). @{b}GPLUS_INCLUDE_DIR@{ub} means the same thing for native and cross. It is where @{b}g++@{ub} looks first for header files. @{b}libg++@{ub} installs only target independent header files in that directory. @{b}LOCAL_INCLUDE_DIR@{ub} is used only for a native compiler. It is normally @{b}/usr/local/include@{ub}. GNU CC searches this directory so that users can install header files in @{b}/usr/local/include@{ub}. @{b}CROSS_INCLUDE_DIR@{ub} is used only for a cross compiler. GNU CC doesn't install anything there. @{b}TOOL_INCLUDE_DIR@{ub} is used for both native and cross compilers. It is the place for other packages to install header files that GNU CC will use. For a cross-compiler, this is the equivalent of @{b}/usr/include@{ub}. When you build a cross-compiler, @{b}fixincludes@{ub} processes any header files in this directory. @EndNode @Node "C Extensions" "gcc.guide/C Extensions" @Next "C++ Extensions" @Prev "Installation" @Toc "Main" Extensions to the C Language Family *********************************** GNU C provides several language features not found in ANSI standard C. (The @{b}-pedantic@{ub} option directs GNU CC to print a warning message if any of these features is used.) To test for the availability of these features in conditional compilation, check for a predefined macro @{b}__GNUC__@{ub}, which is always defined under GNU CC. These extensions are available in C and Objective C. Most of them are also available in C++. See @{"Extensions to the C++ Language" Link "C++ Extensions"}, for extensions that apply @{i}only@{ui} to C++. @{" Statement Exprs " Link "Statement Exprs"} Putting statements and declarations inside expressions. @{" Local Labels " Link "Local Labels"} Labels local to a statement-expression. @{" Labels as Values " Link "Labels as Values"} Getting pointers to labels, and computed gotos. @{" Nested Functions " Link "Nested Functions"} As in Algol and Pascal, lexical scoping of functions. @{" Constructing Calls " Link "Constructing Calls"} Dispatching a call to another function. @{" Naming Types " Link "Naming Types"} Giving a name to the type of some expression. @{" Typeof " Link "Typeof"} @{b}typeof@{ub}: referring to the type of an expression. @{" Lvalues " Link "Lvalues"} Using @{b}?:@{ub}, @{b},@{ub} and casts in lvalues. @{" Conditionals " Link "Conditionals"} Omitting the middle operand of a @{b}?:@{ub} expression. @{" Long Long " Link "Long Long"} Double-word integers--@{b}long long int@{ub}. @{" Complex " Link "Complex"} Data types for complex numbers. @{" Zero Length " Link "Zero Length"} Zero-length arrays. @{" Variable Length " Link "Variable Length"} Arrays whose length is computed at run time. @{" Macro Varargs " Link "Macro Varargs"} Macros with variable number of arguments. @{" Subscripting " Link "Subscripting"} Any array can be subscripted, even if not an lvalue. @{" Pointer Arith " Link "Pointer Arith"} Arithmetic on @{b}void@{ub}-pointers and function pointers. @{" Initializers " Link "Initializers"} Non-constant initializers. @{" Constructors " Link "Constructors"} Constructor expressions give structures, unions or arrays as values. @{" Labeled Elements " Link "Labeled Elements"} Labeling elements of initializers. @{" Cast to Union " Link "Cast to Union"} Casting to union type from any member of the union. @{" Case Ranges " Link "Case Ranges"} `case 1 ... 9' and such. @{" Function Attributes " Link "Function Attributes"} Declaring that functions have no side effects, or that they can never return. @{" Function Prototypes " Link "Function Prototypes"} Prototype declarations and old-style definitions. @{" C++ Comments " Link "C++ Comments"} C++ comments are recognized. @{" Dollar Signs " Link "Dollar Signs"} Dollar sign is allowed in identifiers. @{" Character Escapes " Link "Character Escapes"} @{b}\\e@{ub} stands for the character . @{" Variable Attributes " Link "Variable Attributes"} Specifying attributes of variables. @{" Type Attributes " Link "Type Attributes"} Specifying attributes of types. @{" Alignment " Link "Alignment"} Inquiring about the alignment of a type or variable. @{" Inline " Link "Inline"} Defining inline functions (as fast as macros). @{" Extended Asm " Link "Extended Asm"} Assembler instructions with C expressions as operands. (With them you can define "built-in" functions.) @{" Asm Labels " Link "Asm Labels"} Specifying the assembler name to use for a C symbol. @{" Explicit Reg Vars " Link "Explicit Reg Vars"} Defining variables residing in specified registers. @{" Alternate Keywords " Link "Alternate Keywords"} @{b}__const__@{ub}, @{b}__asm__@{ub}, etc., for header files. @{" Incomplete Enums " Link "Incomplete Enums"} @{b}enum foo;@{ub}, with details to follow. @{" Function Names " Link "Function Names"} Printable strings which are the name of the current function. @EndNode @Node "Statement Exprs" "gcc.guide/Statement Exprs" @Next "Local Labels" @Toc "C Extensions" Statements and Declarations in Expressions ========================================== A compound statement enclosed in parentheses may appear as an expression in GNU C. This allows you to use loops, switches, and local variables within an expression. Recall that a compound statement is a sequence of statements surrounded by braces; in this construct, parentheses go around the braces. For example: ({ int y = foo (); int z; if (y > 0) z = y; else z = - y; z; }) is a valid (though slightly more complex than necessary) expression for the absolute value of @{b}foo ()@{ub}. The last thing in the compound statement should be an expression followed by a semicolon; the value of this subexpression serves as the value of the entire construct. (If you use some other kind of statement last within the braces, the construct has type @{b}void@{ub}, and thus effectively no value.) This feature is especially useful in making macro definitions "safe" (so that they evaluate each operand exactly once). For example, the "maximum" function is commonly defined as a macro in standard C as follows: #define max(a,b) ((a) > (b) ? (a) : (b)) But this definition computes either @{I}A@{ui} or @{I}B@{ui} twice, with bad results if the operand has side effects. In GNU C, if you know the type of the operands (here let's assume @{b}int@{ub}), you can define the macro safely as follows: #define maxint(a,b) \\ ({int _a = (a), _b = (b); _a > _b ? _a : _b; }) Embedded statements are not allowed in constant expressions, such as the value of an enumeration constant, the width of a bit field, or the initial value of a static variable. If you don't know the type of the operand, you can still do this, but you must use @{b}typeof@{ub} (see @{"Typeof" Link "Typeof"}) or type naming (see @{"Naming Types" Link "Naming Types"}). @EndNode @Node "Local Labels" "gcc.guide/Local Labels" @Next "Labels as Values" @Prev "Statement Exprs" @Toc "C Extensions" Locally Declared Labels ======================= Each statement expression is a scope in which @{i}local labels@{ui} can be declared. A local label is simply an identifier; you can jump to it with an ordinary @{b}goto@{ub} statement, but only from within the statement expression it belongs to. A local label declaration looks like this: __label__ @{I}LABEL@{ui}; or __label__ @{I}LABEL1@{ui}, @{I}LABEL2@{ui}, ...; Local label declarations must come at the beginning of the statement expression, right after the @{b} ({@{ub}, before any ordinary declarations. The label declaration defines the label @{i}name@{ui}, but does not define the label itself. You must do this in the usual way, with @{b}@{I}LABEL@{ui}:@{ub}, within the statements of the statement expression. The local label feature is useful because statement expressions are often used in macros. If the macro contains nested loops, a @{b}goto@{ub} can be useful for breaking out of them. However, an ordinary label whose scope is the whole function cannot be used: if the macro can be expanded several times in one function, the label will be multiply defined in that function. A local label avoids this problem. For example: #define SEARCH(array, target) \\ ({ \\ __label__ found; \\ typeof (target) _SEARCH_target = (target); \\ typeof (*(array)) *_SEARCH_array = (array); \\ int i, j; \\ int value; \\ for (i = 0; i < max; i++) \\ for (j = 0; j < max; j++) \\ if (_SEARCH_array[i][j] == _SEARCH_target) \\ { value = i; goto found; } \\ value = -1; \\ found: \\ value; \\ }) @EndNode @Node "Labels as Values" "gcc.guide/Labels as Values" @Next "Nested Functions" @Prev "Local Labels" @Toc "C Extensions" Labels as Values ================ You can get the address of a label defined in the current function (or a containing function) with the unary operator @{b}&&@{ub}. The value has type @{b}void *@{ub}. This value is a constant and can be used wherever a constant of that type is valid. For example: void *ptr; ... ptr = &&foo; To use these values, you need to be able to jump to one. This is done with the computed goto statement(1), @{b}goto *@{I}EXP@{ui};@{ub}. For example, goto *ptr; Any expression of type @{b}void *@{ub} is allowed. One way of using these constants is in initializing a static array that will serve as a jump table: static void *array[] = { &&foo, &&bar, &&hack }; Then you can select a label with indexing, like this: goto *array[i]; Note that this does not check whether the subscript is in bounds--array indexing in C never does that. Such an array of label values serves a purpose much like that of the @{b}switch@{ub} statement. The @{b}switch@{ub} statement is cleaner, so use that rather than an array unless the problem does not fit a @{b}switch@{ub} statement very well. Another use of label values is in an interpreter for threaded code. The labels within the interpreter function can be stored in the threaded code for super-fast dispatching. You can use this mechanism to jump to code in a different function. If you do that, totally unpredictable things will happen. The best way to avoid this is to store the label address only in automatic variables and never pass it as an argument. ---------- Footnotes ---------- (1) The analogous feature in Fortran is called an assigned goto, but that name seems inappropriate in C, where one can do more than simply store label addresses in label variables. @EndNode @Node "Nested Functions" "gcc.guide/Nested Functions" @Next "Constructing Calls" @Prev "Labels as Values" @Toc "C Extensions" Nested Functions ================ A @{i}nested function@{ui} is a function defined inside another function. (Nested functions are not supported for GNU C++.) The nested function's name is local to the block where it is defined. For example, here we define a nested function named @{b}square@{ub}, and call it twice: foo (double a, double b) { double square (double z) { return z * z; } return square (a) + square (b); } The nested function can access all the variables of the containing function that are visible at the point of its definition. This is called @{i}lexical scoping@{ui}. For example, here we show a nested function which uses an inherited variable named @{b}offset@{ub}: bar (int *array, int offset, int size) { int access (int *array, int index) { return array[index + offset]; } int i; ... for (i = 0; i < size; i++) ... access (array, i) ... } Nested function definitions are permitted within functions in the places where variable definitions are allowed; that is, in any block, before the first statement in the block. It is possible to call the nested function from outside the scope of its name by storing its address or passing the address to another function: hack (int *array, int size) { void store (int index, int value) { array[index] = value; } intermediate (store, size); } Here, the function @{b}intermediate@{ub} receives the address of @{b}store@{ub} as an argument. If @{b}intermediate@{ub} calls @{b}store@{ub}, the arguments given to @{b}store@{ub} are used to store into @{b}array@{ub}. But this technique works only so long as the containing function (@{b}hack@{ub}, in this example) does not exit. If you try to call the nested function through its address after the containing function has exited, all hell will break loose. If you try to call it after a containing scope level has exited, and if it refers to some of the variables that are no longer in scope, you may be lucky, but it's not wise to take the risk. If, however, the nested function does not refer to anything that has gone out of scope, you should be safe. GNU CC implements taking the address of a nested function using a technique called @{i}trampolines@{ui}. A paper describing them is available from @{b}maya.idiap.ch@{ub} in directory @{b}pub/tmb@{ub}, file @{b}usenix88-lexic.ps.Z@{ub}. A nested function can jump to a label inherited from a containing function, provided the label was explicitly declared in the containing function (see @{"Local Labels" Link "Local Labels"}). Such a jump returns instantly to the containing function, exiting the nested function which did the @{b}goto@{ub} and any intermediate functions as well. Here is an example: bar (int *array, int offset, int size) { __label__ failure; int access (int *array, int index) { if (index > size) goto failure; return array[index + offset]; } int i; ... for (i = 0; i < size; i++) ... access (array, i) ... ... return 0; /* @{i}Control comes here from @{b}access@{ub} if it detects an error.@{ui} */ failure: return -1; } A nested function always has internal linkage. Declaring one with @{b}extern@{ub} is erroneous. If you need to declare the nested function before its definition, use @{b}auto@{ub} (which is otherwise meaningless for function declarations). bar (int *array, int offset, int size) { __label__ failure; auto int access (int *, int); ... int access (int *array, int index) { if (index > size) goto failure; return array[index + offset]; } ... } @EndNode @Node "Constructing Calls" "gcc.guide/Constructing Calls" @Next "Naming Types" @Prev "Nested Functions" @Toc "C Extensions" Constructing Function Calls =========================== Using the built-in functions described below, you can record the arguments a function received, and call another function with the same arguments, without knowing the number or types of the arguments. You can also record the return value of that function call, and later return that value, without knowing what data type the function tried to return (as long as your caller expects that data type). @{b}__builtin_apply_args ()@{ub} This built-in function returns a pointer of type @{b}void *@{ub} to data describing how to perform a call with the same arguments as were passed to the current function. The function saves the arg pointer register, structure value address, and all registers that might be used to pass arguments to a function into a block of memory allocated on the stack. Then it returns the address of that block. @{b}__builtin_apply (@{I}FUNCTION@{ui}, @{I}ARGUMENTS@{ui}, @{I}SIZE@{ui})@{ub} This built-in function invokes @{I}FUNCTION@{ui} (type @{b}void (*)()@{ub}) with a copy of the parameters described by @{I}ARGUMENTS@{ui} (type @{b}void *@{ub}) and @{I}SIZE@{ui} (type @{b}int@{ub}). The value of @{I}ARGUMENTS@{ui} should be the value returned by @{b}__builtin_apply_args@{ub}. The argument @{I}SIZE@{ui} specifies the size of the stack argument data, in bytes. This function returns a pointer of type @{b}void *@{ub} to data describing how to return whatever value was returned by @{I}FUNCTION@{ui}. The data is saved in a block of memory allocated on the stack. It is not always simple to compute the proper value for @{I}SIZE@{ui}. The value is used by @{b}__builtin_apply@{ub} to compute the amount of data that should be pushed on the stack and copied from the incoming argument area. @{b}__builtin_return (@{I}RESULT@{ui})@{ub} This built-in function returns the value described by @{I}RESULT@{ui} from the containing function. You should specify, for @{I}RESULT@{ui}, a value returned by @{b}__builtin_apply@{ub}. @EndNode @Node "Naming Types" "gcc.guide/Naming Types" @Next "Typeof" @Prev "Constructing Calls" @Toc "C Extensions" Naming an Expression's Type =========================== You can give a name to the type of an expression using a @{b}typedef@{ub} declaration with an initializer. Here is how to define @{I}NAME@{ui} as a type name for the type of @{I}EXP@{ui}: typedef @{I}NAME@{ui} = @{I}EXP@{ui}; This is useful in conjunction with the statements-within-expressions feature. Here is how the two together can be used to define a safe "maximum" macro that operates on any arithmetic type: #define max(a,b) \\ ({typedef _ta = (a), _tb = (b); \\ _ta _a = (a); _tb _b = (b); \\ _a > _b ? _a : _b; }) The reason for using names that start with underscores for the local variables is to avoid conflicts with variable names that occur within the expressions that are substituted for @{b}a@{ub} and @{b}b@{ub}. Eventually we hope to design a new form of declaration syntax that allows you to declare variables whose scopes start only after their initializers; this will be a more reliable way to prevent such conflicts. @EndNode @Node "Typeof" "gcc.guide/Typeof" @Next "Lvalues" @Prev "Naming Types" @Toc "C Extensions" Referring to a Type with @{b}typeof@{ub} =============================== Another way to refer to the type of an expression is with @{b}typeof@{ub}. The syntax of using of this keyword looks like @{b}sizeof@{ub}, but the construct acts semantically like a type name defined with @{b}typedef@{ub}. There are two ways of writing the argument to @{b}typeof@{ub}: with an expression or with a type. Here is an example with an expression: typeof (x[0](1)) This assumes that @{b}x@{ub} is an array of functions; the type described is that of the values of the functions. Here is an example with a typename as the argument: typeof (int *) Here the type described is that of pointers to @{b}int@{ub}. If you are writing a header file that must work when included in ANSI C programs, write @{b}__typeof__@{ub} instead of @{b}typeof@{ub}. See @{"Alternate Keywords" Link "Alternate Keywords"}. A @{b}typeof@{ub}-construct can be used anywhere a typedef name could be used. For example, you can use it in a declaration, in a cast, or inside of @{b}sizeof@{ub} or @{b}typeof@{ub}. @{b}*@{ub} This declares @{b}y@{ub} with the type of what @{b}x@{ub} points to. typeof (*x) y; @{b}*@{ub} This declares @{b}y@{ub} as an array of such values. typeof (*x) y[4]; @{b}*@{ub} This declares @{b}y@{ub} as an array of pointers to characters: typeof (typeof (char *)[4]) y; It is equivalent to the following traditional C declaration: char *y[4]; To see the meaning of the declaration using @{b}typeof@{ub}, and why it might be a useful way to write, let's rewrite it with these macros: #define pointer(T) typeof(T *) #define array(T, N) typeof(T [N]) Now the declaration can be rewritten this way: array (pointer (char), 4) y; Thus, @{b}array (pointer (char), 4)@{ub} is the type of arrays of 4 pointers to @{b}char@{ub}. @EndNode @Node "Lvalues" "gcc.guide/Lvalues" @Next "Conditionals" @Prev "Typeof" @Toc "C Extensions" Generalized Lvalues =================== Compound expressions, conditional expressions and casts are allowed as lvalues provided their operands are lvalues. This means that you can take their addresses or store values into them. Standard C++ allows compound expressions and conditional expressions as lvalues, and permits casts to reference type, so use of this extension is deprecated for C++ code. For example, a compound expression can be assigned, provided the last expression in the sequence is an lvalue. These two expressions are equivalent: (a, b) += 5 a, (b += 5) Similarly, the address of the compound expression can be taken. These two expressions are equivalent: &(a, b) a, &b A conditional expression is a valid lvalue if its type is not void and the true and false branches are both valid lvalues. For example, these two expressions are equivalent: (a ? b : c) = 5 (a ? b = 5 : (c = 5)) A cast is a valid lvalue if its operand is an lvalue. A simple assignment whose left-hand side is a cast works by converting the right-hand side first to the specified type, then to the type of the inner left-hand side expression. After this is stored, the value is converted back to the specified type to become the value of the assignment. Thus, if @{b}a@{ub} has type @{b}char *@{ub}, the following two expressions are equivalent: (int)a = 5 (int)(a = (char *)(int)5) An assignment-with-arithmetic operation such as @{b}+=@{ub} applied to a cast performs the arithmetic using the type resulting from the cast, and then continues as in the previous case. Therefore, these two expressions are equivalent: (int)a += 5 (int)(a = (char *)(int) ((int)a + 5)) You cannot take the address of an lvalue cast, because the use of its address would not work out coherently. Suppose that @{b}&(int)f@{ub} were permitted, where @{b}f@{ub} has type @{b}float@{ub}. Then the following statement would try to store an integer bit-pattern where a floating point number belongs: *&(int)f = 1; This is quite different from what @{b}(int)f = 1@{ub} would do--that would convert 1 to floating point and store it. Rather than cause this inconsistency, we think it is better to prohibit use of @{b}&@{ub} on a cast. If you really do want an @{b}int *@{ub} pointer with the address of @{b}f@{ub}, you can simply write @{b}(int *)&f@{ub}. @EndNode @Node "Conditionals" "gcc.guide/Conditionals" @Next "Long Long" @Prev "Lvalues" @Toc "C Extensions" Conditionals with Omitted Operands ================================== The middle operand in a conditional expression may be omitted. Then if the first operand is nonzero, its value is the value of the conditional expression. Therefore, the expression x ? : y has the value of @{b}x@{ub} if that is nonzero; otherwise, the value of @{b}y@{ub}. This example is perfectly equivalent to x ? x : y In this simple case, the ability to omit the middle operand is not especially useful. When it becomes useful is when the first operand does, or may (if it is a macro argument), contain a side effect. Then repeating the operand in the middle would perform the side effect twice. Omitting the middle operand uses the value already computed without the undesirable effects of recomputing it. @EndNode @Node "Long Long" "gcc.guide/Long Long" @Next "Complex" @Prev "Conditionals" @Toc "C Extensions" Double-Word Integers ==================== GNU C supports data types for integers that are twice as long as @{b}long int@{ub}. Simply write @{b}long long int@{ub} for a signed integer, or @{b}unsigned long long int@{ub} for an unsigned integer. To make an integer constant of type @{b}long long int@{ub}, add the suffix @{b}LL@{ub} to the integer. To make an integer constant of type @{b}unsigned long long int@{ub}, add the suffix @{b}ULL@{ub} to the integer. You can use these types in arithmetic like any other integer types. Addition, subtraction, and bitwise boolean operations on these types are open-coded on all types of machines. Multiplication is open-coded if the machine supports fullword-to-doubleword a widening multiply instruction. Division and shifts are open-coded only on machines that provide special support. The operations that are not open-coded use special library routines that come with GNU CC. There may be pitfalls when you use @{b}long long@{ub} types for function arguments, unless you declare function prototypes. If a function expects type @{b}int@{ub} for its argument, and you pass a value of type @{b}long long int@{ub}, confusion will result because the caller and the subroutine will disagree about the number of bytes for the argument. Likewise, if the function expects @{b}long long int@{ub} and you pass @{b}int@{ub}. The best way to avoid such problems is to use prototypes. @EndNode @Node "Complex" "gcc.guide/Complex" @Next "Zero Length" @Prev "Long Long" @Toc "C Extensions" Complex Numbers =============== GNU C supports complex data types. You can declare both complex integer types and complex floating types, using the keyword @{b}__complex__@{ub}. For example, @{b}__complex__ double x;@{ub} declares @{b}x@{ub} as a variable whose real part and imaginary part are both of type @{b}double@{ub}. @{b}__complex__ short int y;@{ub} declares @{b}y@{ub} to have real and imaginary parts of type @{b}short int@{ub}; this is not likely to be useful, but it shows that the set of complex types is complete. To write a constant with a complex data type, use the suffix @{b}i@{ub} or @{b}j@{ub} (either one; they are equivalent). For example, @{b}2.5fi@{ub} has type @{b}__complex__ float@{ub} and @{b}3i@{ub} has type @{b}__complex__ int@{ub}. Such a constant always has a pure imaginary value, but you can form any complex value you like by adding one to a real constant. To extract the real part of a complex-valued expression @{I}EXP@{ui}, write @{b}__real__ @{I}EXP@{ui}@{ub}. Likewise, use @{b}__imag__@{ub} to extract the imaginary part. The operator @{b}~@{ub} performs complex conjugation when used on a value with a complex type. GNU CC can allocate complex automatic variables in a noncontiguous fashion; it's even possible for the real part to be in a register while the imaginary part is on the stack (or vice-versa). None of the supported debugging info formats has a way to represent noncontiguous allocation like this, so GNU CC describes a noncontiguous complex variable as if it were two separate variables of noncomplex type. If the variable's actual name is @{b}foo@{ub}, the two fictitious variables are named @{b}foo$real@{ub} and @{b}foo$imag@{ub}. You can examine and set these two fictitious variables with your debugger. A future version of GDB will know how to recognize such pairs and treat them as a single variable with a complex type. @EndNode @Node "Zero Length" "gcc.guide/Zero Length" @Next "Variable Length" @Prev "Complex" @Toc "C Extensions" Arrays of Length Zero ===================== Zero-length arrays are allowed in GNU C. They are very useful as the last element of a structure which is really a header for a variable-length object: struct line { int length; char contents[0]; }; { struct line *thisline = (struct line *) malloc (sizeof (struct line) + this_length); thisline->length = this_length; } In standard C, you would have to give @{b}contents@{ub} a length of 1, which means either you waste space or complicate the argument to @{b}malloc@{ub}. @EndNode @Node "Variable Length" "gcc.guide/Variable Length" @Next "Macro Varargs" @Prev "Zero Length" @Toc "C Extensions" Arrays of Variable Length ========================= Variable-length automatic arrays are allowed in GNU C. These arrays are declared like any other automatic arrays, but with a length that is not a constant expression. The storage is allocated at the point of declaration and deallocated when the brace-level is exited. For example: FILE * concat_fopen (char *s1, char *s2, char *mode) { char str[strlen (s1) + strlen (s2) + 1]; strcpy (str, s1); strcat (str, s2); return fopen (str, mode); } Jumping or breaking out of the scope of the array name deallocates the storage. Jumping into the scope is not allowed; you get an error message for it. You can use the function @{b}alloca@{ub} to get an effect much like variable-length arrays. The function @{b}alloca@{ub} is available in many other C implementations (but not in all). On the other hand, variable-length arrays are more elegant. There are other differences between these two methods. Space allocated with @{b}alloca@{ub} exists until the containing @{i}function@{ui} returns. The space for a variable-length array is deallocated as soon as the array name's scope ends. (If you use both variable-length arrays and @{b}alloca@{ub} in the same function, deallocation of a variable-length array will also deallocate anything more recently allocated with @{b}alloca@{ub}.) You can also use variable-length arrays as arguments to functions: struct entry tester (int len, char data[len][len]) { ... } The length of an array is computed once when the storage is allocated and is remembered for the scope of the array in case you access it with @{b}sizeof@{ub}. If you want to pass the array first and the length afterward, you can use a forward declaration in the parameter list--another GNU extension. struct entry tester (int len; char data[len][len], int len) { ... } The @{b}int len@{ub} before the semicolon is a @{i}parameter forward declaration@{ui}, and it serves the purpose of making the name @{b}len@{ub} known when the declaration of @{b}data@{ub} is parsed. You can write any number of such parameter forward declarations in the parameter list. They can be separated by commas or semicolons, but the last one must end with a semicolon, which is followed by the "real" parameter declarations. Each forward declaration must match a "real" declaration in parameter name and data type. @EndNode @Node "Macro Varargs" "gcc.guide/Macro Varargs" @Next "Subscripting" @Prev "Variable Length" @Toc "C Extensions" Macros with Variable Numbers of Arguments ========================================= In GNU C, a macro can accept a variable number of arguments, much as a function can. The syntax for defining the macro looks much like that used for a function. Here is an example: #define eprintf(format, args...) \\ fprintf (stderr, format , ## args) Here @{b}args@{ub} is a @{i}rest argument@{ui}: it takes in zero or more arguments, as many as the call contains. All of them plus the commas between them form the value of @{b}args@{ub}, which is substituted into the macro body where @{b}args@{ub} is used. Thus, we have this expansion: eprintf ("%s:%d: ", input_file_name, line_number) ==> fprintf (stderr, "%s:%d: " , input_file_name, line_number) Note that the comma after the string constant comes from the definition of @{b}eprintf@{ub}, whereas the last comma comes from the value of @{b}args@{ub}. The reason for using @{b}##@{ub} is to handle the case when @{b}args@{ub} matches no arguments at all. In this case, @{b}args@{ub} has an empty value. In this case, the second comma in the definition becomes an embarrassment: if it got through to the expansion of the macro, we would get something like this: fprintf (stderr, "success!\\n" , ) which is invalid C syntax. @{b}##@{ub} gets rid of the comma, so we get the following instead: fprintf (stderr, "success!\\n") This is a special feature of the GNU C preprocessor: @{b}##@{ub} before a rest argument that is empty discards the preceding sequence of non-whitespace characters from the macro definition. (If another macro argument precedes, none of it is discarded.) It might be better to discard the last preprocessor token instead of the last preceding sequence of non-whitespace characters; in fact, we may someday change this feature to do so. We advise you to write the macro definition so that the preceding sequence of non-whitespace characters is just a single token, so that the meaning will not change if we change the definition of this feature. @EndNode @Node "Subscripting" "gcc.guide/Subscripting" @Next "Pointer Arith" @Prev "Macro Varargs" @Toc "C Extensions" Non-Lvalue Arrays May Have Subscripts ===================================== Subscripting is allowed on arrays that are not lvalues, even though the unary @{b}&@{ub} operator is not. For example, this is valid in GNU C though not valid in other C dialects: struct foo {int a[4];}; struct foo f(); bar (int index) { return f().a[index]; } @EndNode @Node "Pointer Arith" "gcc.guide/Pointer Arith" @Next "Initializers" @Prev "Subscripting" @Toc "C Extensions" Arithmetic on @{b}void@{ub}- and Function-Pointers ========================================= In GNU C, addition and subtraction operations are supported on pointers to @{b}void@{ub} and on pointers to functions. This is done by treating the size of a @{b}void@{ub} or of a function as 1. A consequence of this is that @{b}sizeof@{ub} is also allowed on @{b}void@{ub} and on function types, and returns 1. The option @{b}-Wpointer-arith@{ub} requests a warning if these extensions are used. @EndNode @Node "Initializers" "gcc.guide/Initializers" @Next "Constructors" @Prev "Pointer Arith" @Toc "C Extensions" Non-Constant Initializers ========================= As in standard C++, the elements of an aggregate initializer for an automatic variable are not required to be constant expressions in GNU C. Here is an example of an initializer with run-time varying elements: foo (float f, float g) { float beat_freqs[2] = { f-g, f+g }; ... } @EndNode @Node "Constructors" "gcc.guide/Constructors" @Next "Labeled Elements" @Prev "Initializers" @Toc "C Extensions" Constructor Expressions ======================= GNU C supports constructor expressions. A constructor looks like a cast containing an initializer. Its value is an object of the type specified in the cast, containing the elements specified in the initializer. Usually, the specified type is a structure. Assume that @{b}struct foo@{ub} and @{b}structure@{ub} are declared as shown: struct foo {int a; char b[2];} structure; Here is an example of constructing a @{b}struct foo@{ub} with a constructor: structure = ((struct foo) {x + y, 'a', 0}); This is equivalent to writing the following: { struct foo temp = {x + y, 'a', 0}; structure = temp; } You can also construct an array. If all the elements of the constructor are (made up of) simple constant expressions, suitable for use in initializers, then the constructor is an lvalue and can be coerced to a pointer to its first element, as shown here: char **foo = (char *[]) { "x", "y", "z" }; Array constructors whose elements are not simple constants are not very useful, because the constructor is not an lvalue. There are only two valid ways to use it: to subscript it, or initialize an array variable with it. The former is probably slower than a @{b}switch@{ub} statement, while the latter does the same thing an ordinary C initializer would do. Here is an example of subscripting an array constructor: output = ((int[]) { 2, x, 28 }) [input]; Constructor expressions for scalar types and union types are is also allowed, but then the constructor expression is equivalent to a cast. @EndNode @Node "Labeled Elements" "gcc.guide/Labeled Elements" @Next "Cast to Union" @Prev "Constructors" @Toc "C Extensions" Labeled Elements in Initializers ================================ Standard C requires the elements of an initializer to appear in a fixed order, the same as the order of the elements in the array or structure being initialized. In GNU C you can give the elements in any order, specifying the array indices or structure field names they apply to. This extension is not implemented in GNU C++. To specify an array index, write @{b} [@{I}INDEX@{ui}]@{ub} or @{b} [@{I}INDEX@{ui}] =@{ub} before the element value. For example, int a[6] = { [4] 29, [2] = 15 }; is equivalent to int a[6] = { 0, 0, 15, 0, 29, 0 }; The index values must be constant expressions, even if the array being initialized is automatic. To initialize a range of elements to the same value, write @{b} [@{I}FIRST@{ui} ... @{I}LAST@{ui}] = @{I}VALUE@{ui}@{ub}. For example, int widths[] = { [0 ... 9] = 1, [10 ... 99] = 2, [100] = 3 }; Note that the length of the array is the highest value specified plus one. In a structure initializer, specify the name of a field to initialize with @{b}@{I}FIELDNAME@{ui}:@{ub} before the element value. For example, given the following structure, struct point { int x, y; }; the following initialization struct point p = { y: yvalue, x: xvalue }; is equivalent to struct point p = { xvalue, yvalue }; Another syntax which has the same meaning is @{b} .@{I}FIELDNAME@{ui} =@{ub}., as shown here: struct point p = { .y = yvalue, .x = xvalue }; You can also use an element label (with either the colon syntax or the period-equal syntax) when initializing a union, to specify which element of the union should be used. For example, union foo { int i; double d; }; union foo f = { d: 4 }; will convert 4 to a @{b}double@{ub} to store it in the union using the second element. By contrast, casting 4 to type @{b}union foo@{ub} would store it into the union as the integer @{b}i@{ub}, since it is an integer. (See @{"Cast to Union" Link "Cast to Union"}.) You can combine this technique of naming elements with ordinary C initialization of successive elements. Each initializer element that does not have a label applies to the next consecutive element of the array or structure. For example, int a[6] = { [1] = v1, v2, [4] = v4 }; is equivalent to int a[6] = { 0, v1, v2, 0, v4, 0 }; Labeling the elements of an array initializer is especially useful when the indices are characters or belong to an @{b}enum@{ub} type. For example: int whitespace[256] = { [' '] = 1, ['\\t'] = 1, ['\\h'] = 1, ['\\f'] = 1, ['\\n'] = 1, ['\\r'] = 1 }; @EndNode @Node "Case Ranges" "gcc.guide/Case Ranges" @Next "Function Attributes" @Prev "Cast to Union" @Toc "C Extensions" Case Ranges =========== You can specify a range of consecutive values in a single @{b}case@{ub} label, like this: case @{I}LOW@{ui} ... @{I}HIGH@{ui}: This has the same effect as the proper number of individual @{b}case@{ub} labels, one for each integer value from @{I}LOW@{ui} to @{I}HIGH@{ui}, inclusive. This feature is especially useful for ranges of ASCII character codes: case 'A' ... 'Z': @{i}Be careful:@{ui} Write spaces around the @{b}...@{ub}, for otherwise it may be parsed wrong when you use it with integer values. For example, write this: case 1 ... 5: rather than this: case 1...5: @EndNode @Node "Cast to Union" "gcc.guide/Cast to Union" @Next "Case Ranges" @Prev "Labeled Elements" @Toc "C Extensions" Cast to a Union Type ==================== A cast to union type is similar to other casts, except that the type specified is a union type. You can specify the type either with @{b}union @{I}TAG@{ui}@{ub} or with a typedef name. A cast to union is actually a constructor though, not a cast, and hence does not yield an lvalue like normal casts. (See @{"Constructors" Link "Constructors"}.) The types that may be cast to the union type are those of the members of the union. Thus, given the following union and variables: union foo { int i; double d; }; int x; double y; both @{b}x@{ub} and @{b}y@{ub} can be cast to type @{b}union@{ub} foo. Using the cast as the right-hand side of an assignment to a variable of union type is equivalent to storing in a member of the union: union foo u; ... u = (union foo) x == u.i = x u = (union foo) y == u.d = y You can also use the union cast as a function argument: void hack (union foo); ... hack ((union foo) x); @EndNode @Node "Function Attributes" "gcc.guide/Function Attributes" @Next "Function Prototypes" @Prev "Case Ranges" @Toc "C Extensions" Declaring Attributes of Functions ================================= In GNU C, you declare certain things about functions called in your program which help the compiler optimize function calls and check your code more carefully. The keyword @{b}__attribute__@{ub} allows you to specify special attributes when making a declaration. This keyword is followed by an attribute specification inside double parentheses. Eight attributes, @{b}noreturn@{ub}, @{b}const@{ub}, @{b}format@{ub}, @{b}section@{ub}, @{b}constructor@{ub}, @{b}destructor@{ub}, @{b}unused@{ub} and @{b}weak@{ub} are currently defined for functions. Other attributes, including @{b}section@{ub} are supported for variables declarations (see @{"Variable Attributes" Link "Variable Attributes"}) and for types (see @{"Type Attributes" Link "Type Attributes"}). You may also specify attributes with @{b}__@{ub} preceding and following each keyword. This allows you to use them in header files without being concerned about a possible macro of the same name. For example, you may use @{b}__noreturn__@{ub} instead of @{b}noreturn@{ub}. @{b}noreturn@{ub} A few standard library functions, such as @{b}abort@{ub} and @{b}exit@{ub}, cannot return. GNU CC knows this automatically. Some programs define their own functions that never return. You can declare them @{b}noreturn@{ub} to tell the compiler this fact. For example, void fatal () __attribute__ ((noreturn)); void fatal (...) { ... /* @{i}Print error message.@{ui} */ ... exit (1); } The @{b}noreturn@{ub} keyword tells the compiler to assume that @{b}fatal@{ub} cannot return. It can then optimize without regard to what would happen if @{b}fatal@{ub} ever did return. This makes slightly better code. More importantly, it helps avoid spurious warnings of uninitialized variables. Do not assume that registers saved by the calling function are restored before calling the @{b}noreturn@{ub} function. It does not make sense for a @{b}noreturn@{ub} function to have a return type other than @{b}void@{ub}. The attribute @{b}noreturn@{ub} is not implemented in GNU C versions earlier than 2.5. An alternative way to declare that a function does not return, which works in the current version and in some older versions, is as follows: typedef void voidfn (); volatile voidfn fatal; @{b}const@{ub} Many functions do not examine any values except their arguments, and have no effects except the return value. Such a function can be subject to common subexpression elimination and loop optimization just as an arithmetic operator would be. These functions should be declared with the attribute @{b}const@{ub}. For example, int square (int) __attribute__ ((const)); says that the hypothetical function @{b}square@{ub} is safe to call fewer times than the program says. The attribute @{b}const@{ub} is not implemented in GNU C versions earlier than 2.5. An alternative way to declare that a function has no side effects, which works in the current version and in some older versions, is as follows: typedef int intfn (); extern const intfn square; This approach does not work in GNU C++ from 2.6.0 on, since the language specifies that the @{b}const@{ub} must be attached to the return value. Note that a function that has pointer arguments and examines the data pointed to must @{i}not@{ui} be declared @{b}const@{ub}. Likewise, a function that calls a non-@{b}const@{ub} function usually must not be @{b}const@{ub}. It does not make sense for a @{b}const@{ub} function to return @{b}void@{ub}. @{b}format (@{I}ARCHETYPE@{ui}, @{I}STRING-INDEX@{ui}, @{I}FIRST-TO-CHECK@{ui})@{ub} The @{b}format@{ub} attribute specifies that a function takes @{b}printf@{ub} or @{b}scanf@{ub} style arguments which should be type-checked against a format string. For example, the declaration: extern int my_printf (void *my_object, const char *my_format, ...) __attribute__ ((format (printf, 2, 3))); causes the compiler to check the arguments in calls to @{b}my_printf@{ub} for consistency with the @{b}printf@{ub} style format string argument @{b}my_format@{ub}. The parameter @{I}ARCHETYPE@{ui} determines how the format string is interpreted, and should be either @{b}printf@{ub} or @{b}scanf@{ub}. The parameter @{I}STRING-INDEX@{ui} specifies which argument is the format string argument (starting from 1), while @{I}FIRST-TO-CHECK@{ui} is the number of the first argument to check against the format string. For functions where the arguments are not available to be checked (such as @{b}vprintf@{ub}), specify the third parameter as zero. In this case the compiler only checks the format string for consistency. In the example above, the format string (@{b}my_format@{ub}) is the second argument of the function @{b}my_print@{ub}, and the arguments to check start with the third argument, so the correct parameters for the format attribute are 2 and 3. The @{b}format@{ub} attribute allows you to identify your own functions which take format strings as arguments, so that GNU CC can check the calls to these functions for errors. The compiler always checks formats for the ANSI library functions @{b}printf@{ub}, @{b}fprintf@{ub}, @{b}sprintf@{ub}, @{b}scanf@{ub}, @{b}fscanf@{ub}, @{b}sscanf@{ub}, @{b}vprintf@{ub}, @{b}vfprintf@{ub} and @{b}vsprintf@{ub} whenever such warnings are requested (using @{b}-Wformat@{ub}), so there is no need to modify the header file @{b}stdio.h@{ub}. @{b}section ("section-name")@{ub} Normally, the compiler places the code it generates in the @{b}text@{ub} section. Sometimes, however, you need additional sections, or you need certain particular functions to appear in special sections. The @{b}section@{ub} attribute specifies that a function lives in a particular section. For example, the declaration: extern void foobar (void) __attribute__ ((section ("bar"))); puts the function @{b}foobar@{ub} in the @{b}bar@{ub} section. Some file formats do not support arbitrary sections so the @{b}section@{ub} attribute is not available on all platforms. If you need to map the entire contents of a module to a particular section, consider using the facilities of the linker instead. @{b}constructor@{ub} @{b}destructor@{ub} The @{b}constructor@{ub} attribute causes the function to be called automatically before execution enters @{b}main ()@{ub}. Similarly, the @{b}destructor@{ub} attribute causes the function to be called automatically after @{b}main ()@{ub} has completed or @{b}exit ()@{ub} has been called. Functions with these attributes are useful for initializing data that will be used implicitly during the execution of the program. These attributes are not currently implemented for Objective C. @{b}unused@{ub} This attribute, attached to a function, means that the function is meant to be possibly unused. GNU CC will not produce a warning for this function. @{b}weak@{ub} The @{b}weak@{ub} attribute causes the declaration to be emitted as a weak symbol rather than a global. This is primarily useful in defining library functions which can be overridden in user code, though it can also be used with non-function declarations. Weak symbols are supported for ELF targets, and also for a.out targets when using the GNU assembler and linker. @{b}alias ("target")@{ub} The @{b}alias@{ub} attribute causes the declaration to be emitted as an alias for another symbol, which must be specified. For instance, void __f () { /* do something */; } void f () __attribute__ ((weak, alias ("__f"))); declares @{b}f@{ub} to be a weak alias for @{b}__f@{ub}. In C++, the mangled name for the target must be used. @{b}regparm (@{I}NUMBER@{ui})@{ub} On the Intel 386, the @{b}regparm@{ub} attribute causes the compiler to pass up to @{I}NUMBER@{ui} integer arguments in registers @{I}EAX@{ui}, @{I}EDX@{ui}, and @{I}ECX@{ui} instead of on the stack. Functions that take a variable number of arguments will continue to be passed all of their arguments on the stack. @{b}stdcall@{ub} On the Intel 386, the @{b}stdcall@{ub} attribute causes the compiler to assume that the called function will pop off the stack space used to pass arguments, unless it takes a variable number of arguments. @{b}cdecl@{ub} On the Intel 386, the @{b}cdecl@{ub} attribute causes the compiler to assume that the called function will pop off the stack space used to pass arguments, unless it takes a variable number of arguments. This is useful to override the effects of the @{b}-mrtd@{ub} switch. You can specify multiple attributes in a declaration by separating them by commas within the double parentheses or by immediately following an attribute declaration with another attribute declaration. Some people object to the @{b}__attribute__@{ub} feature, suggesting that ANSI C's @{b}#pragma@{ub} should be used instead. There are two reasons for not doing this. 1. It is impossible to generate @{b}#pragma@{ub} commands from a macro. 2. There is no telling what the same @{b}#pragma@{ub} might mean in another compiler. These two reasons apply to almost any application that might be proposed for @{b}#pragma@{ub}. It is basically a mistake to use @{b}#pragma@{ub} for @{i}anything@{ui}. @EndNode @Node "Function Prototypes" "gcc.guide/Function Prototypes" @Next "C++ Comments" @Prev "Function Attributes" @Toc "C Extensions" Prototypes and Old-Style Function Definitions ============================================= GNU C extends ANSI C to allow a function prototype to override a later old-style non-prototype definition. Consider the following example: /* @{i}Use prototypes unless the compiler is old-fashioned.@{ui} */ #if __STDC__ #define P(x) x #else #define P(x) () #endif /* @{i}Prototype function declaration.@{ui} */ int isroot P((uid_t)); /* @{i}Old-style function definition.@{ui} */ int isroot (x) /* ??? lossage here ??? */ uid_t x; { return x == 0; } Suppose the type @{b}uid_t@{ub} happens to be @{b}short@{ub}. ANSI C does not allow this example, because subword arguments in old-style non-prototype definitions are promoted. Therefore in this example the function definition's argument is really an @{b}int@{ub}, which does not match the prototype argument type of @{b}short@{ub}. This restriction of ANSI C makes it hard to write code that is portable to traditional C compilers, because the programmer does not know whether the @{b}uid_t@{ub} type is @{b}short@{ub}, @{b}int@{ub}, or @{b}long@{ub}. Therefore, in cases like these GNU C allows a prototype to override a later old-style definition. More precisely, in GNU C, a function prototype argument type overrides the argument type specified by a later old-style definition if the former type is the same as the latter type before promotion. Thus in GNU C the above example is equivalent to the following: int isroot (uid_t); int isroot (uid_t x) { return x == 0; } GNU C++ does not support old-style function definitions, so this extension is irrelevant. @EndNode @Node "C++ Comments" "gcc.guide/C++ Comments" @Next "Dollar Signs" @Prev "Function Prototypes" @Toc "C Extensions" C++ Style Comments ================== In GNU C, you may use C++ style comments, which start with @{b}//@{ub} and continue until the end of the line. Many other C implementations allow such comments, and they are likely to be in a future C standard. However, C++ style comments are not recognized if you specify @{b}-ansi@{ub} or @{b}-traditional@{ub}, since they are incompatible with traditional constructs like @{b}dividend//*comment*/divisor@{ub}. @EndNode @Node "Dollar Signs" "gcc.guide/Dollar Signs" @Next "Character Escapes" @Prev "C++ Comments" @Toc "C Extensions" Dollar Signs in Identifier Names ================================ In GNU C, you may use dollar signs in identifier names. This is because many traditional C implementations allow such identifiers. On some machines, dollar signs are allowed in identifiers if you specify @{b}-traditional@{ub}. On a few systems they are allowed by default, even if you do not use @{b}-traditional@{ub}. But they are never allowed if you specify @{b}-ansi@{ub}. There are certain ANSI C programs (obscure, to be sure) that would compile incorrectly if dollar signs were permitted in identifiers. For example: #define foo(a) #a #define lose(b) foo (b) #define test$ lose (test) @EndNode @Node "Character Escapes" "gcc.guide/Character Escapes" @Next "Variable Attributes" @Prev "Dollar Signs" @Toc "C Extensions" The Character in Constants ================================ You can use the sequence @{b}\\e@{ub} in a string or character constant to stand for the ASCII character . @EndNode @Node "Alignment" "gcc.guide/Alignment" @Next "Inline" @Prev "Type Attributes" @Toc "C Extensions" Inquiring on Alignment of Types or Variables ============================================ The keyword @{b}__alignof__@{ub} allows you to inquire about how an object is aligned, or the minimum alignment usually required by a type. Its syntax is just like @{b}sizeof@{ub}. For example, if the target machine requires a @{b}double@{ub} value to be aligned on an 8-byte boundary, then @{b}__alignof__ (double)@{ub} is 8. This is true on many RISC machines. On more traditional machine designs, @{b}__alignof__ (double)@{ub} is 4 or even 2. Some machines never actually require alignment; they allow reference to any data type even at an odd addresses. For these machines, @{b}__alignof__@{ub} reports the @{i}recommended@{ui} alignment of a type. When the operand of @{b}__alignof__@{ub} is an lvalue rather than a type, the value is the largest alignment that the lvalue is known to have. It may have this alignment as a result of its data type, or because it is part of a structure and inherits alignment from that structure. For example, after this declaration: struct foo { int x; char y; } foo1; the value of @{b}__alignof__ (foo1.y)@{ub} is probably 2 or 4, the same as @{b}__alignof__ (int)@{ub}, even though the data type of @{b}foo1.y@{ub} does not itself demand any alignment. A related feature which lets you specify the alignment of an object is @{b}__attribute__ ((aligned (@{I}ALIGNMENT@{ui})))@{ub}; see the following section. @EndNode @Node "Variable Attributes" "gcc.guide/Variable Attributes" @Next "Type Attributes" @Prev "Character Escapes" @Toc "C Extensions" Specifying Attributes of Variables ================================== The keyword @{b}__attribute__@{ub} allows you to specify special attributes of variables or structure fields. This keyword is followed by an attribute specification inside double parentheses. Eight attributes are currently defined for variables: @{b}aligned@{ub}, @{b}mode@{ub}, @{b}nocommon@{ub}, @{b}packed@{ub}, @{b}section@{ub}, @{b}transparent_union@{ub}, @{b}unused@{ub}, and @{b}weak@{ub}. Other attributes are available for functions (see @{"Function Attributes" Link "Function Attributes"}) and for types (see @{"Type Attributes" Link "Type Attributes"}). You may also specify attributes with @{b}__@{ub} preceding and following each keyword. This allows you to use them in header files without being concerned about a possible macro of the same name. For example, you may use @{b}__aligned__@{ub} instead of @{b}aligned@{ub}. @{b}aligned (@{I}ALIGNMENT@{ui})@{ub} This attribute specifies a minimum alignment for the variable or structure field, measured in bytes. For example, the declaration: int x __attribute__ ((aligned (16))) = 0; causes the compiler to allocate the global variable @{b}x@{ub} on a 16-byte boundary. On a 68040, this could be used in conjunction with an @{b}asm@{ub} expression to access the @{b}move16@{ub} instruction which requires 16-byte aligned operands. You can also specify the alignment of structure fields. For example, to create a double-word aligned @{b}int@{ub} pair, you could write: struct foo { int x[2] __attribute__ ((aligned (8))); }; This is an alternative to creating a union with a @{b}double@{ub} member that forces the union to be double-word aligned. It is not possible to specify the alignment of functions; the alignment of functions is determined by the machine's requirements and cannot be changed. You cannot specify alignment for a typedef name because such a name is just an alias, not a distinct type. As in the preceding examples, you can explicitly specify the alignment (in bytes) that you wish the compiler to use for a given variable or structure field. Alternatively, you can leave out the alignment factor and just ask the compiler to align a variable or field to the maximum useful alignment for the target machine you are compiling for. For example, you could write: short array[3] __attribute__ ((aligned)); Whenever you leave out the alignment factor in an @{b}aligned@{ub} attribute specification, the compiler automatically sets the alignment for the declared variable or field to the largest alignment which is ever used for any data type on the target machine you are compiling for. Doing this can often make copy operations more efficient, because the compiler can use whatever instructions copy the biggest chunks of memory when performing copies to or from the variables or fields that you have aligned this way. The @{b}aligned@{ub} attribute can only increase the alignment; but you can decrease it by specifying @{b}packed@{ub} as well. See below. Note that the effectiveness of @{b}aligned@{ub} attributes may be limited by inherent limitations in your linker. On many systems, the linker is only able to arrange for variables to be aligned up to a certain maximum alignment. (For some linkers, the maximum supported alignment may be very very small.) If your linker is only able to align variables up to a maximum of 8 byte alignment, then specifying @{b}aligned(16)@{ub} in an @{b}__attribute__@{ub} will still only provide you with 8 byte alignment. See your linker documentation for further information. @{b}mode (@{I}MODE@{ui})@{ub} This attribute specifies the data type for the declaration--whichever type corresponds to the mode @{I}MODE@{ui}. This in effect lets you request an integer or floating point type according to its width. You may also specify a mode of @{b}byte@{ub} or @{b}__byte__@{ub} to indicate the mode corresponding to a one-byte integer, @{b}word@{ub} or @{b}__word__@{ub} for the mode of a one-word integer, and @{b}pointer@{ub} or @{b}__pointer__@{ub} for the mode used to represent pointers. @{b}nocommon@{ub} This attribute specifies requests GNU CC not to place a variable "common" but instead to allocate space for it directly. If you specify the @{b}-fno-common@{ub} flag, GNU CC will do this for all variables. Specifying the @{b}nocommon@{ub} attribute for a variable provides an initialization of zeros. A variable may only be initialized in one source file. @{b}packed@{ub} The @{b}packed@{ub} attribute specifies that a variable or structure field should have the smallest possible alignment--one byte for a variable, and one bit for a field, unless you specify a larger value with the @{b}aligned@{ub} attribute. Here is a structure in which the field @{b}x@{ub} is packed, so that it immediately follows @{b}a@{ub}: struct foo { char a; int x[2] __attribute__ ((packed)); }; @{b}section ("section-name")@{ub} Normally, the compiler places the objects it generates in sections like @{b}data@{ub} and @{b}bss@{ub}. Sometimes, however, you need additional sections, or you need certain particular variables to appear in special sections, for example to map to special hardware. The @{b}section@{ub} attribute specifies that a variable (or function) lives in a particular section. For example, this small program uses several specific section names: struct duart a __attribute__ ((section ("DUART_A"))) = { 0 }; struct duart b __attribute__ ((section ("DUART_B"))) = { 0 }; char stack[10000] __attribute__ ((section ("STACK"))) = { 0 }; int init_data_copy __attribute__ ((section ("INITDATACOPY"))) = 0; main() { /* Initialize stack pointer */ init_sp (stack + sizeof (stack)); /* Initialize initialized data */ memcpy (&init_data_copy, &data, &edata - &data); /* Turn on the serial ports */ init_duart (&a); init_duart (&b); } Use the @{b}section@{ub} attribute with an @{i}initialized@{ui} definition of a @{i}global@{ui} variable, as shown in the example. GNU CC issues a warning and otherwise ignores the @{b}section@{ub} attribute in uninitialized variable declarations. You may only use the @{b}section@{ub} attribute with a fully initialized global definition because of the way linkers work. The linker requires each object be defined once, with the exception that uninitialized variables tentatively go in the @{b}common@{ub} (or @{b}bss@{ub}) section and can be multiply "defined". You can force a variable to be initialized with the @{b}-fno-common@{ub} flag or the @{b}nocommon@{ub} attribute. Some file formats do not support arbitrary sections so the @{b}section@{ub} attribute is not available on all platforms. If you need to map the entire contents of a module to a particular section, consider using the facilities of the linker instead. @{b}transparent_union@{ub} This attribute, attached to a function argument variable which is a union, means to pass the argument in the same way that the first union member would be passed. You can also use this attribute on a @{b}typedef@{ub} for a union data type; then it applies to all function arguments with that type. @{b}unused@{ub} This attribute, attached to a variable, means that the variable is meant to be possibly unused. GNU CC will not produce a warning for this variable. @{b}weak@{ub} The @{b}weak@{ub} attribute is described in See @{"Function Attributes" Link "Function Attributes"}. To specify multiple attributes, separate them by commas within the double parentheses: for example, @{b}__attribute__ ((aligned (16), packed))@{ub}. @EndNode @Node "Type Attributes" "gcc.guide/Type Attributes" @Next "Alignment" @Prev "Variable Attributes" @Toc "C Extensions" Specifying Attributes of Types ============================== The keyword @{b}__attribute__@{ub} allows you to specify special attributes of @{b}struct@{ub} and @{b}union@{ub} types when you define such types. This keyword is followed by an attribute specification inside double parentheses. Three attributes are currently defined for types: @{b}aligned@{ub}, @{b}packed@{ub}, and @{b}transparent_union@{ub}. Other attributes are defined for functions (see @{"Function Attributes" Link "Function Attributes"}) and for variables (see @{"Variable Attributes" Link "Variable Attributes"}). You may also specify any one of these attributes with @{b}__@{ub} preceding and following its keyword. This allows you to use these attributes in header files without being concerned about a possible macro of the same name. For example, you may use @{b}__aligned__@{ub} instead of @{b}aligned@{ub}. You may specify the @{b}aligned@{ub} and @{b}transparent_union@{ub} attributes either in a @{b}typedef@{ub} declaration or just past the closing curly brace of a complete enum, struct or union type @{i}definition@{ui} and the @{b}packed@{ub} attribute only past the closing brace of a definition. @{b}aligned (@{I}ALIGNMENT@{ui})@{ub} This attribute specifies a minimum alignment (in bytes) for variables of the specified type. For example, the declarations: struct S { short f[3]; } __attribute__ ((aligned (8)); typedef int more_aligned_int __attribute__ ((aligned (8)); force the compiler to insure (as fas as it can) that each variable whose type is @{b}struct S@{ub} or @{b}more_aligned_int@{ub} will be allocated and aligned @{i}at least@{ui} on a 8-byte boundary. On a Sparc, having all variables of type @{b}struct S@{ub} aligned to 8-byte boundaries allows the compiler to use the @{b}ldd@{ub} and @{b}std@{ub} (doubleword load and store) instructions when copying one variable of type @{b}struct S@{ub} to another, thus improving run-time efficiency. Note that the alignment of any given @{b}struct@{ub} or @{b}union@{ub} type is required by the ANSI C standard to be at least a perfect multiple of the lowest common multiple of the alignments of all of the members of the @{b}struct@{ub} or @{b}union@{ub} in question. This means that you @{i}can@{ui} effectively adjust the alignment of a @{b}struct@{ub} or @{b}union@{ub} type by attaching an @{b}aligned@{ub} attribute to any one of the members of such a type, but the notation illustrated in the example above is a more obvious, intuitive, and readable way to request the compiler to adjust the alignment of an entire @{b}struct@{ub} or @{b}union@{ub} type. As in the preceding example, you can explicitly specify the alignment (in bytes) that you wish the compiler to use for a given @{b}struct@{ub} or @{b}union@{ub} type. Alternatively, you can leave out the alignment factor and just ask the compiler to align a type to the maximum useful alignment for the target machine you are compiling for. For example, you could write: struct S { short f[3]; } __attribute__ ((aligned)); Whenever you leave out the alignment factor in an @{b}aligned@{ub} attribute specification, the compiler automatically sets the alignment for the type to the largest alignment which is ever used for any data type on the target machine you are compiling for. Doing this can often make copy operations more efficient, because the compiler can use whatever instructions copy the biggest chunks of memory when performing copies to or from the variables which have types that you have aligned this way. In the example above, if the size of each @{b}short@{ub} is 2 bytes, then the size of the entire @{b}struct S@{ub} type is 6 bytes. The smallest power of two which is greater than or equal to that is 8, so the compiler sets the alignment for the entire @{b}struct S@{ub} type to 8 bytes. Note that although you can ask the compiler to select a time-efficient alignment for a given type and then declare only individual stand-alone objects of that type, the compiler's ability to select a time-efficient alignment is primarily useful only when you plan to create arrays of variables having the relevant (efficiently aligned) type. If you declare or use arrays of variables of an efficiently-aligned type, then it is likely that your program will also be doing pointer arithmetic (or subscripting, which amounts to the same thing) on pointers to the relevant type, and the code that the compiler generates for these pointer arithmetic operations will often be more efficient for efficiently-aligned types than for other types. The @{b}aligned@{ub} attribute can only increase the alignment; but you can decrease it by specifying @{b}packed@{ub} as well. See below. Note that the effectiveness of @{b}aligned@{ub} attributes may be limited by inherent limitations in your linker. On many systems, the linker is only able to arrange for variables to be aligned up to a certain maximum alignment. (For some linkers, the maximum supported alignment may be very very small.) If your linker is only able to align variables up to a maximum of 8 byte alignment, then specifying @{b}aligned(16)@{ub} in an @{b}__attribute__@{ub} will still only provide you with 8 byte alignment. See your linker documentation for further information. @{b}packed@{ub} This attribute, attached to an @{b}enum@{ub}, @{b}struct@{ub}, or @{b}union@{ub} type definition, specified that the minimum required memory be used to represent the type. Specifying this attribute for @{b}struct@{ub} and @{b}union@{ub} types is equivalent to specifying the @{b}packed@{ub} attribute on each of the structure or union members. Specifying the @{b}-fshort-enums@{ub} flag on the line is equivalent to specifying the @{b}packed@{ub} attribute on all @{b}enum@{ub} definitions. You may only specify this attribute after a closing curly brace on an @{b}enum@{ub} definition, not in a @{b}typedef@{ub} declaration. @{b}transparent_union@{ub} This attribute, attached to a @{b}union@{ub} type definition, indicates that any variable having that union type should, if passed to a function, be passed in the same way that the first union member would be passed. For example: union foo { char a; int x[2]; } __attribute__ ((transparent_union)); To specify multiple attributes, separate them by commas within the double parentheses: for example, @{b}__attribute__ ((aligned (16), packed))@{ub}. @EndNode @Node "Inline" "gcc.guide/Inline" @Next "Extended Asm" @Prev "Alignment" @Toc "C Extensions" An Inline Function is As Fast As a Macro ======================================== By declaring a function @{b}inline@{ub}, you can direct GNU CC to integrate that function's code into the code for its callers. This makes execution faster by eliminating the function-call overhead; in addition, if any of the actual argument values are constant, their known values may permit simplifications at compile time so that not all of the inline function's code needs to be included. The effect on code size is less predictable; object code may be larger or smaller with function inlining, depending on the particular case. Inlining of functions is an optimization and it really "works" only in optimizing compilation. If you don't use @{b}-O@{ub}, no function is really inline. To declare a function inline, use the @{b}inline@{ub} keyword in its declaration, like this: inline int inc (int *a) { (*a)++; } (If you are writing a header file to be included in ANSI C programs, write @{b}__inline__@{ub} instead of @{b}inline@{ub}. See @{"Alternate Keywords" Link "Alternate Keywords"}.) You can also make all "simple enough" functions inline with the option @{b}-finline-functions@{ub}. Note that certain usages in a function definition can make it unsuitable for inline substitution. Note that in C and Objective C, unlike C++, the @{b}inline@{ub} keyword does not affect the linkage of the function. GNU CC automatically inlines member functions defined within the class body of C++ programs even if they are not explicitly declared @{b}inline@{ub}. (You can override this with @{b}-fno-default-inline@{ub}; see @{"Options Controlling C++ Dialect" Link "C++ Dialect Options"}.) When a function is both inline and @{b}static@{ub}, if all calls to the function are integrated into the caller, and the function's address is never used, then the function's own assembler code is never referenced. In this case, GNU CC does not actually output assembler code for the function, unless you specify the option @{b}-fkeep-inline-functions@{ub}. Some calls cannot be integrated for various reasons (in particular, calls that precede the function's definition cannot be integrated, and neither can recursive calls within the definition). If there is a nonintegrated call, then the function is compiled to assembler code as usual. The function must also be compiled as usual if the program refers to its address, because that can't be inlined. When an inline function is not @{b}static@{ub}, then the compiler must assume that there may be calls from other source files; since a global symbol can be defined only once in any program, the function must not be defined in the other source files, so the calls therein cannot be integrated. Therefore, a non-@{b}static@{ub} inline function is always compiled on its own in the usual fashion. If you specify both @{b}inline@{ub} and @{b}extern@{ub} in the function definition, then the definition is used only for inlining. In no case is the function compiled on its own, not even if you refer to its address explicitly. Such an address becomes an external reference, as if you had only declared the function, and had not defined it. This combination of @{b}inline@{ub} and @{b}extern@{ub} has almost the effect of a macro. The way to use it is to put a function definition in a header file with these keywords, and put another copy of the definition (lacking @{b}inline@{ub} and @{b}extern@{ub}) in a library file. The definition in the header file will cause most calls to the function to be inlined. If any uses of the function remain, they will refer to the single copy in the library. GNU C does not inline any functions when not optimizing. It is not clear whether it is better to inline or not, in this case, but we found that a correct implementation when not optimizing was difficult. So we did the easy thing, and turned it off. @EndNode @Node "Extended Asm" "gcc.guide/Extended Asm" @Next "Asm Labels" @Prev "Inline" @Toc "C Extensions" Assembler Instructions with C Expression Operands ================================================= In an assembler instruction using @{b}asm@{ub}, you can now specify the operands of the instruction using C expressions. This means no more guessing which registers or memory locations will contain the data you want to use. You must specify an assembler instruction template much like what appears in a machine description, plus an operand constraint string for each operand. For example, here is how to use the 68881's @{b}fsinx@{ub} instruction: asm ("fsinx %1,%0" : "=f" (result) : "f" (angle)); Here @{b}angle@{ub} is the C expression for the input operand while @{b}result@{ub} is that of the output operand. Each has @{b}"f"@{ub} as its operand constraint, saying that a floating point register is required. The @{b}=@{ub} in @{b}=f@{ub} indicates that the operand is an output; all output operands' constraints must use @{b}=@{ub}. The constraints use the same language used in the machine description (see @{"Constraints" Link "Constraints"}). Each operand is described by an operand-constraint string followed by the C expression in parentheses. A colon separates the assembler template from the first output operand, and another separates the last output operand from the first input, if any. Commas separate output operands and separate inputs. The total number of operands is limited to ten or to the maximum number of operands in any instruction pattern in the machine description, whichever is greater. If there are no output operands, and there are input operands, then there must be two consecutive colons surrounding the place where the output operands would go. Output operand expressions must be lvalues; the compiler can check this. The input operands need not be lvalues. The compiler cannot check whether the operands have data types that are reasonable for the instruction being executed. It does not parse the assembler instruction template and does not know what it means, or whether it is valid assembler input. The extended @{b}asm@{ub} feature is most often used for machine instructions that the compiler itself does not know exist. If the output expression cannot be directly addressed (for example, it is a bit field), your constraint must allow a register. In that case, GNU CC will use the register as the output of the @{b}asm@{ub}, and then store that register into the output. The output operands must be write-only; GNU CC will assume that the values in these operands before the instruction are dead and need not be generated. Extended asm does not support input-output or read-write operands. For this reason, the constraint character @{b}+@{ub}, which indicates such an operand, may not be used. When the assembler instruction has a read-write operand, or an operand in which only some of the bits are to be changed, you must logically split its function into two separate operands, one input operand and one write-only output operand. The connection between them is expressed by constraints which say they need to be in the same location when the instruction executes. You can use the same C expression for both operands, or different expressions. For example, here we write the (fictitious) @{b}combine@{ub} instruction with @{b}bar@{ub} as its read-only source operand and @{b}foo@{ub} as its read-write destination: asm ("combine %2,%0" : "=r" (foo) : "0" (foo), "g" (bar)); The constraint @{b}"0"@{ub} for operand 1 says that it must occupy the same location as operand 0. A digit in constraint is allowed only in an input operand, and it must refer to an output operand. Only a digit in the constraint can guarantee that one operand will be in the same place as another. The mere fact that @{b}foo@{ub} is the value of both operands is not enough to guarantee that they will be in the same place in the generated assembler code. The following would not work: asm ("combine %2,%0" : "=r" (foo) : "r" (foo), "g" (bar)); Various optimizations or reloading could cause operands 0 and 1 to be in different registers; GNU CC knows no reason not to do so. For example, the compiler might find a copy of the value of @{b}foo@{ub} in one register and use it for operand 1, but generate the output operand 0 in a different register (copying it afterward to @{b}foo@{ub}'s own address). Of course, since the register for operand 1 is not even mentioned in the assembler code, the result will not work, but GNU CC can't tell that. Some instructions clobber specific hard registers. To describe this, write a third colon after the input operands, followed by the names of the clobbered hard registers (given as strings). Here is a realistic example for the Vax: asm volatile ("movc3 %0,%1,%2" : /* no outputs */ : "g" (from), "g" (to), "g" (count) : "r0", "r1", "r2", "r3", "r4", "r5"); If you refer to a particular hardware register from the assembler code, then you will probably have to list the register after the third colon to tell the compiler that the register's value is modified. In many assemblers, the register names begin with @{b}%@{ub}; to produce one @{b}%@{ub} in the assembler code, you must write @{b}%%@{ub} in the input. If your assembler instruction can alter the condition code register, add @{b}cc@{ub} to the list of clobbered registers. GNU CC on some machines represents the condition codes as a specific hardware register; @{b}cc@{ub} serves to name this register. On other machines, the condition code is handled differently, and specifying @{b}cc@{ub} has no effect. But it is valid no matter what the machine. If your assembler instruction modifies memory in an unpredictable fashion, add @{b}memory@{ub} to the list of clobbered registers. This will cause GNU CC to not keep memory values cached in registers across the assembler instruction. You can put multiple assembler instructions together in a single @{b}asm@{ub} template, separated either with newlines (written as @{b}\\n@{ub}) or with semicolons if the assembler allows such semicolons. The GNU assembler allows semicolons and all Unix assemblers seem to do so. The input operands are guaranteed not to use any of the clobbered registers, and neither will the output operands' addresses, so you can read and write the clobbered registers as many times as you like. Here is an example of multiple instructions in a template; it assumes that the subroutine @{b}_foo@{ub} accepts arguments in registers 9 and 10: asm ("movl %0,r9;movl %1,r10;call _foo" : /* no outputs */ : "g" (from), "g" (to) : "r9", "r10"); Unless an output operand has the @{b}&@{ub} constraint modifier, GNU CC may allocate it in the same register as an unrelated input operand, on the assumption that the inputs are consumed before the outputs are produced. This assumption may be false if the assembler code actually consists of more than one instruction. In such a case, use @{b}&@{ub} for each output operand that may not overlap an input. See @{"Modifiers" Link "Modifiers"}. If you want to test the condition code produced by an assembler instruction, you must include a branch and a label in the @{b}asm@{ub} construct, as follows: asm ("clr %0;frob %1;beq 0f;mov #1,%0;0:" : "g" (result) : "g" (input)); This assumes your assembler supports local labels, as the GNU assembler and most Unix assemblers do. Speaking of labels, jumps from one @{b}asm@{ub} to another are not supported. The compiler's optimizers do not know about these jumps, and therefore they cannot take account of them when deciding how to optimize. Usually the most convenient way to use these @{b}asm@{ub} instructions is to encapsulate them in macros that look like functions. For example, #define sin(x) \\ ({ double __value, __arg = (x); \\ asm ("fsinx %1,%0": "=f" (__value): "f" (__arg)); \\ __value; }) Here the variable @{b}__arg@{ub} is used to make sure that the instruction operates on a proper @{b}double@{ub} value, and to accept only those arguments @{b}x@{ub} which can convert automatically to a @{b}double@{ub}. Another way to make sure the instruction operates on the correct data type is to use a cast in the @{b}asm@{ub}. This is different from using a variable @{b}__arg@{ub} in that it converts more different types. For example, if the desired type were @{b}int@{ub}, casting the argument to @{b}int@{ub} would accept a pointer with no complaint, while assigning the argument to an @{b}int@{ub} variable named @{b}__arg@{ub} would warn about using a pointer unless the caller explicitly casts it. If an @{b}asm@{ub} has output operands, GNU CC assumes for optimization purposes that the instruction has no side effects except to change the output operands. This does not mean that instructions with a side effect cannot be used, but you must be careful, because the compiler may eliminate them if the output operands aren't used, or move them out of loops, or replace two with one if they constitute a common subexpression. Also, if your instruction does have a side effect on a variable that otherwise appears not to change, the old value of the variable may be reused later if it happens to be found in a register. You can prevent an @{b}asm@{ub} instruction from being deleted, moved significantly, or combined, by writing the keyword @{b}volatile@{ub} after the @{b}asm@{ub}. For example: #define set_priority(x) \\ asm volatile ("set_priority %0": /* no outputs */ : "g" (x)) An instruction without output operands will not be deleted or moved significantly, regardless, unless it is unreachable. Note that even a volatile @{b}asm@{ub} instruction can be moved in ways that appear insignificant to the compiler, such as across jump instructions. You can't expect a sequence of volatile @{b}asm@{ub} instructions to remain perfectly consecutive. If you want consecutive output, use a single @{b}asm@{ub}. It is a natural idea to look for a way to give access to the condition code left by the assembler instruction. However, when we attempted to implement this, we found no way to make it work reliably. The problem is that output operands might need reloading, which would result in additional following "store" instructions. On most machines, these instructions would alter the condition code before there was time to test it. This problem doesn't arise for ordinary "test" and "compare" instructions because they don't have any output operands. If you are writing a header file that should be includable in ANSI C programs, write @{b}__asm__@{ub} instead of @{b}asm@{ub}. See @{"Alternate Keywords" Link "Alternate Keywords"}. @EndNode @Node "Asm Labels" "gcc.guide/Asm Labels" @Next "Explicit Reg Vars" @Prev "Extended Asm" @Toc "C Extensions" Controlling Names Used in Assembler Code ======================================== You can specify the name to be used in the assembler code for a C function or variable by writing the @{b}asm@{ub} (or @{b}__asm__@{ub}) keyword after the declarator as follows: int foo asm ("myfoo") = 2; This specifies that the name to be used for the variable @{b}foo@{ub} in the assembler code should be @{b}myfoo@{ub} rather than the usual @{b}_foo@{ub}. On systems where an underscore is normally prepended to the name of a C function or variable, this feature allows you to define names for the linker that do not start with an underscore. You cannot use @{b}asm@{ub} in this way in a function @{i}definition@{ui}; but you can get the same effect by writing a declaration for the function before its definition and putting @{b}asm@{ub} there, like this: extern func () asm ("FUNC"); func (x, y) int x, y; ... It is up to you to make sure that the assembler names you choose do not conflict with any other assembler symbols. Also, you must not use a register name; that would produce completely invalid assembler code. GNU CC does not as yet have the ability to store static variables in registers. Perhaps that will be added. @EndNode @Node "Explicit Reg Vars" "gcc.guide/Explicit Reg Vars" @Next "Alternate Keywords" @Prev "Asm Labels" @Toc "C Extensions" Variables in Specified Registers ================================ GNU C allows you to put a few global variables into specified hardware registers. You can also specify the register in which an ordinary register variable should be allocated. @{b}*@{ub} Global register variables reserve registers throughout the program. This may be useful in programs such as programming language interpreters which have a couple of global variables that are accessed very often. @{b}*@{ub} Local register variables in specific registers do not reserve the registers. The compiler's data flow analysis is capable of determining where the specified registers contain live values, and where they are available for other uses. These local variables are sometimes convenient for use with the extended @{b}asm@{ub} feature (see @{"Extended Asm" Link "Extended Asm"}), if you want to write one output of the assembler instruction directly into a particular register. (This will work provided the register you specify fits the constraints specified for that operand in the @{b}asm@{ub}.) @{" Global Reg Vars " Link "Global Reg Vars"} @{" Local Reg Vars " Link "Local Reg Vars"} @EndNode @Node "Global Reg Vars" "gcc.guide/Global Reg Vars" @Next "Local Reg Vars" @Toc "Explicit Reg Vars" Defining Global Register Variables ---------------------------------- You can define a global register variable in GNU C like this: register int *foo asm ("a5"); Here @{b}a5@{ub} is the name of the register which should be used. Choose a register which is normally saved and restored by function calls on your machine, so that library routines will not clobber it. Naturally the register name is cpu-dependent, so you would need to conditionalize your program according to cpu type. The register @{b}a5@{ub} would be a good choice on a 68000 for a variable of pointer type. On machines with register windows, be sure to choose a "global" register that is not affected magically by the function call mechanism. In addition, operating systems on one type of cpu may differ in how they name the registers; then you would need additional conditionals. For example, some 68000 operating systems call this register @{b}%a5@{ub}. Eventually there may be a way of asking the compiler to choose a register automatically, but first we need to figure out how it should choose and how to enable you to guide the choice. No solution is evident. Defining a global register variable in a certain register reserves that register entirely for this use, at least within the current compilation. The register will not be allocated for any other purpose in the functions in the current compilation. The register will not be saved and restored by these functions. Stores into this register are never deleted even if they would appear to be dead, but references may be deleted or moved or simplified. It is not safe to access the global register variables from signal handlers, or from more than one thread of control, because the system library routines may temporarily use the register for other things (unless you recompile them specially for the task at hand). It is not safe for one function that uses a global register variable to call another such function @{b}foo@{ub} by way of a third function @{b}lose@{ub} that was compiled without knowledge of this variable (i.e. in a different source file in which the variable wasn't declared). This is because @{b}lose@{ub} might save the register and put some other value there. For example, you can't expect a global register variable to be available in the comparison-function that you pass to @{b}qsort@{ub}, since @{b}qsort@{ub} might have put something else in that register. (If you are prepared to recompile @{b}qsort@{ub} with the same global register variable, you can solve this problem.) If you want to recompile @{b}qsort@{ub} or other source files which do not actually use your global register variable, so that they will not use that register for any other purpose, then it suffices to specify the compiler option @{b}-ffixed-@{I}REG@{ui}@{ub}. You need not actually add a global register declaration to their source code. A function which can alter the value of a global register variable cannot safely be called from a function compiled without this variable, because it could clobber the value the caller expects to find there on return. Therefore, the function which is the entry point into the part of the program that uses the global register variable must explicitly save and restore the value which belongs to its caller. On most machines, @{b}longjmp@{ub} will restore to each global register variable the value it had at the time of the @{b}setjmp@{ub}. On some machines, however, @{b}longjmp@{ub} will not change the value of global register variables. To be portable, the function that called @{b}setjmp@{ub} should make other arrangements to save the values of the global register variables, and to restore them in a @{b}longjmp@{ub}. This way, the same thing will happen regardless of what @{b}longjmp@{ub} does. All global register variable declarations must precede all function definitions. If such a declaration could appear after function definitions, the declaration would be too late to prevent the register from being used for other purposes in the preceding functions. Global register variables may not have initial values, because an executable file has no means to supply initial contents for a register. On the Sparc, there are reports that g3 ... g7 are suitable registers, but certain library functions, such as @{b}getwd@{ub}, as well as the subroutines for division and remainder, modify g3 and g4. g1 and g2 are local temporaries. On the 68000, a2 ... a5 should be suitable, as should d2 ... d7. Of course, it will not do to use more than a few of those. @EndNode @Node "Local Reg Vars" "gcc.guide/Local Reg Vars" @Prev "Global Reg Vars" @Toc "Explicit Reg Vars" Specifying Registers for Local Variables ---------------------------------------- You can define a local register variable with a specified register like this: register int *foo asm ("a5"); Here @{b}a5@{ub} is the name of the register which should be used. Note that this is the same syntax used for defining global register variables, but for a local variable it would appear within a function. Naturally the register name is cpu-dependent, but this is not a problem, since specific registers are most often useful with explicit assembler instructions (see @{"Extended Asm" Link "Extended Asm"}). Both of these things generally require that you conditionalize your program according to cpu type. In addition, operating systems on one type of cpu may differ in how they name the registers; then you would need additional conditionals. For example, some 68000 operating systems call this register @{b}%a5@{ub}. Eventually there may be a way of asking the compiler to choose a register automatically, but first we need to figure out how it should choose and how to enable you to guide the choice. No solution is evident. Defining such a register variable does not reserve the register; it remains available for other uses in places where flow control determines the variable's value is not live. However, these registers are made unavailable for use in the reload pass. I would not be surprised if excessive use of this feature leaves the compiler too few available registers to compile certain functions. @EndNode @Node "Alternate Keywords" "gcc.guide/Alternate Keywords" @Next "Incomplete Enums" @Prev "Explicit Reg Vars" @Toc "C Extensions" Alternate Keywords ================== The option @{b}-traditional@{ub} disables certain keywords; @{b}-ansi@{ub} disables certain others. This causes trouble when you want to use GNU C extensions, or ANSI C features, in a general-purpose header file that should be usable by all programs, including ANSI C programs and traditional ones. The keywords @{b}asm@{ub}, @{b}typeof@{ub} and @{b}inline@{ub} cannot be used since they won't work in a program compiled with @{b}-ansi@{ub}, while the keywords @{b}const@{ub}, @{b}volatile@{ub}, @{b}signed@{ub}, @{b}typeof@{ub} and @{b}inline@{ub} won't work in a program compiled with @{b}-traditional@{ub}. The way to solve these problems is to put @{b}__@{ub} at the beginning and end of each problematical keyword. For example, use @{b}__asm__@{ub} instead of @{b}asm@{ub}, @{b}__const__@{ub} instead of @{b}const@{ub}, and @{b}__inline__@{ub} instead of @{b}inline@{ub}. Other C compilers won't accept these alternative keywords; if you want to compile with another compiler, you can define the alternate keywords as macros to replace them with the customary keywords. It looks like this: #ifndef __GNUC__ #define __asm__ asm #endif @{b}-pedantic@{ub} causes warnings for many GNU C extensions. You can prevent such warnings within one expression by writing @{b}__extension__@{ub} before the expression. @{b}__extension__@{ub} has no effect aside from this. @EndNode @Node "Incomplete Enums" "gcc.guide/Incomplete Enums" @Next "Function Names" @Prev "Alternate Keywords" @Toc "C Extensions" Incomplete @{b}enum@{ub} Types ===================== You can define an @{b}enum@{ub} tag without specifying its possible values. This results in an incomplete type, much like what you get if you write @{b}struct foo@{ub} without describing the elements. A later declaration which does specify the possible values completes the type. You can't allocate variables or storage using the type while it is incomplete. However, you can work with pointers to that type. This extension may not be very useful, but it makes the handling of @{b}enum@{ub} more consistent with the way @{b}struct@{ub} and @{b}union@{ub} are handled. This extension is not supported by GNU C++. @EndNode @Node "Function Names" "gcc.guide/Function Names" @Prev "Incomplete Enums" @Toc "C Extensions" Function Names as Strings ========================= GNU CC predefines two string variables to be the name of the current function. The variable @{b}__FUNCTION__@{ub} is the name of the function as it appears in the source. The variable @{b}__PRETTY_FUNCTION__@{ub} is the name of the function pretty printed in a language specific fashion. These names are always the same in a C function, but in a C++ function they may be different. For example, this program: extern "C" { extern int printf (char *, ...); } class a { public: sub (int i) { printf ("__FUNCTION__ = %s\\n", __FUNCTION__); printf ("__PRETTY_FUNCTION__ = %s\\n", __PRETTY_FUNCTION__); } }; int main (void) { a ax; ax.sub (0); return 0; } gives this output: __FUNCTION__ = sub __PRETTY_FUNCTION__ = int a::sub (int) These names are not macros: they are predefined string variables. For example, @{b}#ifdef __FUNCTION__@{ub} does not have any special meaning inside a function, since the preprocessor does not do anything special with the identifier @{b}__FUNCTION__@{ub}. @EndNode @Node "C++ Extensions" "gcc.guide/C++ Extensions" @Next "Trouble" @Prev "C Extensions" @Toc "Main" Extensions to the C++ Language ****************************** The GNU compiler provides these extensions to the C++ language (and you can also use most of the C language extensions in your C++ programs). If you want to write code that checks whether these features are available, you can test for the GNU compiler the same way as for C programs: check for a predefined macro @{b}__GNUC__@{ub}. You can also use @{b}__GNUG__@{ub} to test specifically for GNU C++ (see @{"Standard Predefined Macros" Link "cpp.guide/Standard Predefined"}). @{" Naming Results " Link "Naming Results"} Giving a name to C++ function return values. @{" Min and Max " Link "Min and Max"} C++ Minimum and maximum operators. @{" Destructors and Goto " Link "Destructors and Goto"} Goto is safe to use in C++ even when destructors are needed. @{" C++ Interface " Link "C++ Interface"} You can use a single C++ header file for both declarations and definitions. @{" Template Instantiation " Link "Template Instantiation"} Methods for ensuring that exactly one copy of each needed template instantiation is emitted. @{" C++ Signatures " Link "C++ Signatures"} You can specify abstract types to get subtype polymorphism independent from inheritance. @EndNode @Node "Naming Results" "gcc.guide/Naming Results" @Next "Min and Max" @Toc "C++ Extensions" Named Return Values in C++ ========================== GNU C++ extends the function-definition syntax to allow you to specify a name for the result of a function outside the body of the definition, in C++ programs: @{I}TYPE@{ui} @{I}FUNCTIONNAME@{ui} (@{I}ARGS@{ui}) return @{I}RESULTNAME@{ui}; { ... @{I}BODY@{ui} ... } You can use this feature to avoid an extra constructor call when a function result has a class type. For example, consider a function @{b}m@{ub}, declared as @{b}X v = m ();@{ub}, whose result is of class @{b}X@{ub}: X m () { X b; b.a = 23; return b; } Although @{b}m@{ub} appears to have no arguments, in fact it has one implicit argument: the address of the return value. At invocation, the address of enough space to hold @{b}v@{ub} is sent in as the implicit argument. Then @{b}b@{ub} is constructed and its @{b}a@{ub} field is set to the value 23. Finally, a copy constructor (a constructor of the form @{b}X(X&)@{ub}) is applied to @{b}b@{ub}, with the (implicit) return value location as the target, so that @{b}v@{ub} is now bound to the return value. But this is wasteful. The local @{b}b@{ub} is declared just to hold something that will be copied right out. While a compiler that combined an "elision" algorithm with interprocedural data flow analysis could conceivably eliminate all of this, it is much more practical to allow you to assist the compiler in generating efficient code by manipulating the return value explicitly, thus avoiding the local variable and copy constructor altogether. Using the extended GNU C++ function-definition syntax, you can avoid the temporary allocation and copying by naming @{b}r@{ub} as your return value at the outset, and assigning to its @{b}a@{ub} field directly: X m () return r; { r.a = 23; } The declaration of @{b}r@{ub} is a standard, proper declaration, whose effects are executed @{i}before@{ui} any of the body of @{b}m@{ub}. Functions of this type impose no additional restrictions; in particular, you can execute @{b}return@{ub} statements, or return implicitly by reaching the end of the function body ("falling off the edge"). Cases like X m () return r (23); { return; } (or even @{b}X m () return r (23); { }@{ub}) are unambiguous, since the return value @{b}r@{ub} has been initialized in either case. The following code may be hard to read, but also works predictably: X m () return r; { X b; return b; } The return value slot denoted by @{b}r@{ub} is initialized at the outset, but the statement @{b}return b;@{ub} overrides this value. The compiler deals with this by destroying @{b}r@{ub} (calling the destructor if there is one, or doing nothing if there is not), and then reinitializing @{b}r@{ub} with @{b}b@{ub}. This extension is provided primarily to help people who use overloaded operators, where there is a great need to control not just the arguments, but the return values of functions. For classes where the copy constructor incurs a heavy performance penalty (especially in the common case where there is a quick default constructor), this is a major savings. The disadvantage of this extension is that you do not control when the default constructor for the return value is called: it is always called at the beginning. @EndNode @Node "Min and Max" "gcc.guide/Min and Max" @Next "Destructors and Goto" @Prev "Naming Results" @Toc "C++ Extensions" Minimum and Maximum Operators in C++ ==================================== It is very convenient to have operators which return the "minimum" or the "maximum" of two arguments. In GNU C++ (but not in GNU C), @{b}@{I}A@{ui} ? @{I}B@{ui}@{ub} is the @{i}maximum@{ui}, returning the larger of the numeric values @{I}A@{ui} and @{I}B@{ui}. These operations are not primitive in ordinary C++, since you can use a macro to return the minimum of two things in C++, as in the following example. #define MIN(X,Y) ((X) < (Y) ? : (X) : (Y)) You might then use @{b}int min = MIN (i, j);@{ub} to set @{I}MIN@{ui} to the minimum value of variables @{I}I@{ui} and @{I}J@{ui}. However, side effects in @{b}X@{ub} or @{b}Y@{ub} may cause unintended behavior. For example, @{b}MIN (i++, j++)@{ub} will fail, incrementing the smaller counter twice. A GNU C extension allows you to write safe macros that avoid this kind of problem (see @{"Naming an Expression's Type" Link "Naming Types"}). However, writing @{b}MIN@{ub} and @{b}MAX@{ub} as macros also forces you to use function-call notation notation for a fundamental arithmetic operation. Using GNU C++ extensions, you can write @{b}int min = i ?@{ub} are built into the compiler, they properly handle expressions with side-effects; @{b}int min = i++ ; template ostream& operator << (ostream&, const A&); This strategy will work with code written for either model. If you are using code written for the Cfront model, the file containing a class template and the file containing its member templates should be implemented in the same translation unit. A slight variation on this approach is to use the flag -falt-external-templates instead; this flag causes template instances to be emitted in the translation unit that implements the header where they are first instantiated, rather than the one which implements the file where the templates are defined. This header must be the same in all translation units, or things are likely to break. See @{"Declarations and Definitions in One Header" Link "C++ Interface"}, for more discussion of these pragmas. 3. Explicitly instantiate all the template instances you use, and compile with -fno-implicit-templates. This is probably your best bet; it may require more knowledge of exactly which templates you are using, but it's less mysterious than the previous approach, and it doesn't require any @{b}#pragma@{ub}s or other g++-specific code. You can scatter the instantiations throughout your program, you can create one big file to do all the instantiations, or you can create tiny files like #include "Foo.h" #include "Foo.cc" template class Foo; for each instance you need, and create a template instantiation library from those. I'm partial to the last, but your mileage may vary. If you are using Cfront-model code, you can probably get away with not using -fno-implicit-templates when compiling files that don't @{b}#include@{ub} the member template definitions. @EndNode @Node "C++ Signatures" "gcc.guide/C++ Signatures" @Prev "Template Instantiation" @Toc "C++ Extensions" Type Abstraction using Signatures ================================= In GNU C++, you can use the keyword @{b}signature@{ub} to define a completely abstract class interface as a datatype. You can connect this abstraction with actual classes using signature pointers. If you want to use signatures, run the GNU compiler with the @{b}-fhandle-signatures@{ub} command-line option. (With this option, the compiler reserves a second keyword @{b}sigof@{ub} as well, for a future extension.) Roughly, signatures are type abstractions or interfaces of classes. Some other languages have similar facilities. C++ signatures are related to ML's signatures, Haskell's type classes, definition modules in Modula-2, interface modules in Modula-3, abstract types in Emerald, type modules in Trellis/Owl, categories in Scratchpad II, and types in POOL-I. For a more detailed discussion of signatures, see `Signatures: A Language Extension for Improving Type Abstraction and Subtype Polymorphism in C++' by Gerald Baumgartner and Vincent F. Russo (Tech report CSD-TR-95-051, Dept. of Computer Sciences, Purdue University, August 1995, a slightly improved version appeared in @{i}Software--Practice & Experience@{ui}, @{i}25@{ui}(8), pp. 863-889, August 1995). You can get the tech report by anonymous FTP from @{b}ftp.cs.purdue.edu@{ub} in @{b}pub/gb/Signature-design.ps.gz@{ub}. Syntactically, a signature declaration is a collection of member function declarations and nested type declarations. For example, this signature declaration defines a new abstract type @{b}S@{ub} with member functions @{b}int foo ()@{ub} and @{b}int bar (int)@{ub}: signature S { int foo (); int bar (int); }; Since signature types do not include implementation definitions, you cannot write an instance of a signature directly. Instead, you can define a pointer to any class that contains the required interfaces as a @{i}signature pointer@{ui}. Such a class @{i}implements@{ui} the signature type. To use a class as an implementation of @{b}S@{ub}, you must ensure that the class has public member functions @{b}int foo ()@{ub} and @{b}int bar (int)@{ub}. The class can have other member functions as well, public or not; as long as it offers what's declared in the signature, it is suitable as an implementation of that signature type. For example, suppose that @{b}C@{ub} is a class that meets the requirements of signature @{b}S@{ub} (@{b}C@{ub} @{i}conforms to@{ui} @{b}S@{ub}). Then C obj; S * p = &obj; defines a signature pointer @{b}p@{ub} and initializes it to point to an object of type @{b}C@{ub}. The member function call @{b}int i = p->foo ();@{ub} executes @{b}obj.foo ()@{ub}. Abstract virtual classes provide somewhat similar facilities in standard C++. There are two main advantages to using signatures instead: 1. Subtyping becomes independent from inheritance. A class or signature type @{b}T@{ub} is a subtype of a signature type @{b}S@{ub} independent of any inheritance hierarchy as long as all the member functions declared in @{b}S@{ub} are also found in @{b}T@{ub}. So you can define a subtype hierarchy that is completely independent from any inheritance (implementation) hierarchy, instead of being forced to use types that mirror the class inheritance hierarchy. 2. Signatures allow you to work with existing class hierarchies as implementations of a signature type. If those class hierarchies are only available in compiled form, you're out of luck with abstract virtual classes, since an abstract virtual class cannot be retrofitted on top of existing class hierarchies. So you would be required to write interface classes as subtypes of the abstract virtual class. There is one more detail about signatures. A signature declaration can contain member function @{i}definitions@{ui} as well as member function declarations. A signature member function with a full definition is called a @{i}default implementation@{ui}; classes need not contain that particular interface in order to conform. For example, a class @{b}C@{ub} can conform to the signature signature T { int f (int); int f0 () { return f (0); }; }; whether or not @{b}C@{ub} implements the member function @{b}int f0 ()@{ub}. If you define @{b}C::f0@{ub}, that definition takes precedence; otherwise, the default implementation @{b}S::f0@{ub} applies. @EndNode @Node "Trouble" "gcc.guide/Trouble" @Next "Bugs" @Prev "C++ Extensions" @Toc "Main" Known Causes of Trouble with GNU CC *********************************** This section describes known problems that affect users of GNU CC. Most of these are not GNU CC bugs per se--if they were, we would fix them. But the result for a user may be like the result of a bug. Some of these problems are due to bugs in other software, some are missing features that are too much work to add, and some are places where people's opinions differ as to what is best. @{" Actual Bugs " Link "Actual Bugs"} Bugs we will fix later. @{" Installation Problems " Link "Installation Problems"} Problems that manifest when you install GNU CC. @{" Cross-Compiler Problems " Link "Cross-Compiler Problems"} Common problems of cross compiling with GNU CC. @{" Interoperation " Link "Interoperation"} Problems using GNU CC with other compilers, and with certain linkers, assemblers and debuggers. @{" External Bugs " Link "External Bugs"} Problems compiling certain programs. @{" Incompatibilities " Link "Incompatibilities"} GNU CC is incompatible with traditional C. @{" Fixed Headers " Link "Fixed Headers"} GNU C uses corrected versions of system header files. This is necessary, but doesn't always work smoothly. @{" Standard Libraries " Link "Standard Libraries"} GNU C uses the system C library, which might not be compliant with the ISO/ANSI C standard. @{" Disappointments " Link "Disappointments"} Regrettable things we can't change, but not quite bugs. @{" C++ Misunderstandings " Link "C++ Misunderstandings"} Common misunderstandings with GNU C++. @{" Protoize Caveats " Link "Protoize Caveats"} Things to watch out for when using @{b}protoize@{ub}. @{" Non-bugs " Link "Non-bugs"} Things we think are right, but some others disagree. @{" Warnings and Errors " Link "Warnings and Errors"} Which problems in your code get warnings, and which get errors. @EndNode @Node "Actual Bugs" "gcc.guide/Actual Bugs" @Next "Installation Problems" @Toc "Trouble" Actual Bugs We Haven't Fixed Yet ================================ @{b}*@{ub} The @{b}fixincludes@{ub} script interacts badly with automounters; if the directory of system header files is automounted, it tends to be unmounted while @{b}fixincludes@{ub} is running. This would seem to be a bug in the automounter. We don't know any good way to work around it. @{b}*@{ub} The @{b}fixproto@{ub} script will sometimes add prototypes for the @{b}sigsetjmp@{ub} and @{b}siglongjmp@{ub} functions that reference the @{b}jmp_buf@{ub} type before that type is defined. To work around this, edit the offending file and place the typedef in front of the prototypes. @{b}*@{ub} There are several obscure case of mis-using struct, union, and enum tags that are not detected as errors by the compiler. @{b}*@{ub} When @{b}-pedantic-errors@{ub} is specified, GNU C will incorrectly give an error message when a function name is specified in an expression involving the comma operator. @{b}*@{ub} Loop unrolling doesn't work properly for certain C++ programs. This is a bug in the C++ front end. It sometimes emits incorrect debug info, and the loop unrolling code is unable to recover from this error. @EndNode @Node "Installation Problems" "gcc.guide/Installation Problems" @Next "Cross-Compiler Problems" @Prev "Actual Bugs" @Toc "Trouble" Installation Problems ===================== This is a list of problems (and some apparent problems which don't really mean anything is wrong) that show up during installation of GNU CC. @{b}*@{ub} On certain systems, defining certain environment variables such as @{b}CC@{ub} can interfere with the functioning of @{b}make@{ub}. @{b}*@{ub} If you encounter seemingly strange errors when trying to build the compiler in a directory other than the source directory, it could be because you have previously configured the compiler in the source directory. Make sure you have done all the necessary preparations. See @{"Other Dir" Link "Other Dir"}. @{b}*@{ub} If you build GNU CC on a BSD system using a directory stored in a System V file system, problems may occur in running @{b}fixincludes@{ub} if the System V file system doesn't support symbolic links. These problems result in a failure to fix the declaration of @{b}size_t@{ub} in @{b}sys/types.h@{ub}. If you find that @{b}size_t@{ub} is a signed type and that type mismatches occur, this could be the cause. The solution is not to use such a directory for building GNU CC. @{b}*@{ub} In previous versions of GNU CC, the @{b}gcc@{ub} driver program looked for @{b}as@{ub} and @{b}ld@{ub} in various places; for example, in files beginning with @{b}/usr/local/lib/gcc-@{ub}. GNU CC version 2 looks for them in the directory @{b}/usr/local/lib/gcc-lib/@{I}TARGET@{ui}/@{I}VERSION@{ui}@{ub}. Thus, to use a version of @{b}as@{ub} or @{b}ld@{ub} that is not the system default, for example @{b}gas@{ub} or GNU @{b}ld@{ub}, you must put them in that directory (or make links to them from that directory). @{b}*@{ub} Some commands executed when making the compiler may fail (return a non-zero status) and be ignored by @{b}make@{ub}. These failures, which are often due to files that were not found, are expected, and can safely be ignored. @{b}*@{ub} It is normal to have warnings in compiling certain files about unreachable code and about enumeration type clashes. These files' names begin with @{b}insn-@{ub}. Also, @{b}real.c@{ub} may get some warnings that you can ignore. @{b}*@{ub} Sometimes @{b}make@{ub} recompiles parts of the compiler when installing the compiler. In one case, this was traced down to a bug in @{b}make@{ub}. Either ignore the problem or switch to GNU Make. @{b}*@{ub} If you have installed a program known as purify, you may find that it causes errors while linking @{b}enquire@{ub}, which is part of building GNU CC. The fix is to get rid of the file @{b}real-ld@{ub} which purify installs--so that GNU CC won't try to use it. @{b}*@{ub} On SLS 1.01, a Linux-based GNU system, there is a problem with @{b}libc.a@{ub}: it does not contain the obstack functions. However, GNU CC assumes that the obstack functions are in @{b}libc.a@{ub} when it is the GNU C library. To work around this problem, change the @{b}__GNU_LIBRARY__@{ub} conditional around line 31 to @{b}#if 1@{ub}. @{b}*@{ub} On some 386 systems, building the compiler never finishes because @{b}enquire@{ub} hangs due to a hardware problem in the motherboard--it reports floating point exceptions to the kernel incorrectly. You can install GNU CC except for @{b}float.h@{ub} by patching out the command to run @{b}enquire@{ub}. You may also be able to fix the problem for real by getting a replacement motherboard. This problem was observed in Revision E of the Micronics motherboard, and is fixed in Revision F. It has also been observed in the MYLEX MXA-33 motherboard. If you encounter this problem, you may also want to consider removing the FPU from the socket during the compilation. Alternatively, if you are running SCO Unix, you can reboot and force the FPU to be ignored. To do this, type @{b}hd(40)unix auto ignorefpu@{ub}. @{b}*@{ub} On some 386 systems, GNU CC crashes trying to compile @{b}enquire.c@{ub}. This happens on machines that don't have a 387 FPU chip. On 386 machines, the system kernel is supposed to emulate the 387 when you don't have one. The crash is due to a bug in the emulator. One of these systems is the Unix from Interactive Systems: 386/ix. On this system, an alternate emulator is provided, and it does work. To use it, execute this command as super-user: ln /etc/emulator.rel1 /etc/emulator and then reboot the system. (The default emulator file remains present under the name @{b}emulator.dflt@{ub}.) Try using @{b}/etc/emulator.att@{ub}, if you have such a problem on the SCO system. Another system which has this problem is Esix. We don't know whether it has an alternate emulator that works. On NetBSD 0.8, a similar problem manifests itself as these error messages: enquire.c: In function `fprop': enquire.c:2328: floating overflow @{b}*@{ub} On SCO systems, when compiling GNU CC with the system's compiler, do not use @{b}-O@{ub}. Some versions of the system's compiler miscompile GNU CC with @{b}-O@{ub}. @{b}*@{ub} Sometimes on a Sun 4 you may observe a crash in the program @{b}genflags@{ub} or @{b}genoutput@{ub} while building GNU CC. This is said to be due to a bug in @{b}sh@{ub}. You can probably get around it by running @{b}genflags@{ub} or @{b}genoutput@{ub} manually and then retrying the @{b}make@{ub}. @{b}*@{ub} On Solaris 2, executables of GNU CC version 2.0.2 are commonly available, but they have a bug that shows up when compiling current versions of GNU CC: undefined symbol errors occur during assembly if you use @{b}-g@{ub}. The solution is to compile the current version of GNU CC without @{b}-g@{ub}. That makes a working compiler which you can use to recompile with @{b}-g@{ub}. @{b}*@{ub} Solaris 2 comes with a number of optional OS packages. Some of these packages are needed to use GNU CC fully. If you did not install all optional packages when installing Solaris, you will need to verify that the packages that GNU CC needs are installed. To check whether an optional package is installed, use the @{b}pkginfo@{ub} command. To add an optional package, use the @{b}pkgadd@{ub} command. For further details, see the Solaris documentation. For Solaris 2.0 and 2.1, GNU CC needs six packages: @{b}SUNWarc@{ub}, @{b}SUNWbtool@{ub}, @{b}SUNWesu@{ub}, @{b}SUNWhea@{ub}, @{b}SUNWlibm@{ub}, and @{b}SUNWtoo@{ub}. For Solaris 2.2, GNU CC needs an additional seventh package: @{b}SUNWsprot@{ub}. @{b}*@{ub} On Solaris 2, trying to use the linker and other tools in @{b}/usr/ucb@{ub} to install GNU CC has been observed to cause trouble. For example, the linker may hang indefinitely. The fix is to remove @{b}/usr/ucb@{ub} from your @{b}PATH@{ub}. @{b}*@{ub} If you use the 1.31 version of the MIPS assembler (such as was shipped with Ultrix 3.1), you will need to use the -fno-delayed-branch switch when optimizing floating point code. Otherwise, the assembler will complain when the GCC compiler fills a branch delay slot with a floating point instruction, such as @{b}add.d@{ub}. @{b}*@{ub} If on a MIPS system you get an error message saying "does not have gp sections for all it's [sic] sectons [sic]", don't worry about it. This happens whenever you use GAS with the MIPS linker, but there is not really anything wrong, and it is okay to use the output file. You can stop such warnings by installing the GNU linker. It would be nice to extend GAS to produce the gp tables, but they are optional, and there should not be a warning about their absence. @{b}*@{ub} In Ultrix 4.0 on the MIPS machine, @{b}stdio.h@{ub} does not work with GNU CC at all unless it has been fixed with @{b}fixincludes@{ub}. This causes problems in building GNU CC. Once GNU CC is installed, the problems go away. To work around this problem, when making the stage 1 compiler, specify this option to Make: GCC_FOR_TARGET="./xgcc -B./ -I./include" When making stage 2 and stage 3, specify this option: CFLAGS="-g -I./include" @{b}*@{ub} Users have reported some problems with version 2.0 of the MIPS compiler tools that were shipped with Ultrix 4.1. Version 2.10 which came with Ultrix 4.2 seems to work fine. Users have also reported some problems with version 2.20 of the MIPS compiler tools that were shipped with RISC/os 4.x. The earlier version 2.11 seems to work fine. @{b}*@{ub} Some versions of the MIPS linker will issue an assertion failure when linking code that uses @{b}alloca@{ub} against shared libraries on RISC-OS 5.0, and DEC's OSF/1 systems. This is a bug in the linker, that is supposed to be fixed in future revisions. To protect against this, GNU CC passes @{b}-non_shared@{ub} to the linker unless you pass an explicit @{b}-shared@{ub} or @{b}-call_shared@{ub} switch. @{b}*@{ub} On System V release 3, you may get this error message while linking: ld fatal: failed to write symbol name @{I}SOMETHING@{ui} in strings table for file @{I}WHATEVER@{ui} This probably indicates that the disk is full or your ULIMIT won't allow the file to be as large as it needs to be. This problem can also result because the kernel parameter @{b}MAXUMEM@{ub} is too small. If so, you must regenerate the kernel and make the value much larger. The default value is reported to be 1024; a value of 32768 is said to work. Smaller values may also work. @{b}*@{ub} On System V, if you get an error like this, /usr/local/lib/bison.simple: In function `yyparse': /usr/local/lib/bison.simple:625: virtual memory exhausted that too indicates a problem with disk space, ULIMIT, or @{b}MAXUMEM@{ub}. @{b}*@{ub} Current GNU CC versions probably do not work on version 2 of the NeXT operating system. @{b}*@{ub} On NeXTStep 3.0, the Objective C compiler does not work, due, apparently, to a kernel bug that it happens to trigger. This problem does not happen on 3.1. @{b}*@{ub} On the Tower models 4@{I}N@{ui}0 and 6@{I}N@{ui}0, by default a process is not allowed to have more than one megabyte of memory. GNU CC cannot compile itself (or many other programs) with @{b}-O@{ub} in that much memory. To solve this problem, reconfigure the kernel adding the following line to the configuration file: MAXUMEM = 4096 @{b}*@{ub} On HP 9000 series 300 or 400 running HP-UX release 8.0, there is a bug in the assembler that must be fixed before GNU CC can be built. This bug manifests itself during the first stage of compilation, while building @{b}libgcc2.a@{ub}: _floatdisf cc1: warning: `-g' option not supported on this version of GCC cc1: warning: `-g1' option not supported on this version of GCC ./xgcc: Internal compiler error: program as got fatal signal 11 A patched version of the assembler is available by anonymous ftp from @{b}altdorf.ai.mit.edu@{ub} as the file @{b}archive/cph/hpux-8.0-assembler@{ub}. If you have HP software support, the patch can also be obtained directly from HP, as described in the following note: This is the patched assembler, to patch SR#1653-010439, where the assembler aborts on floating point constants. The bug is not really in the assembler, but in the shared library version of the function "cvtnum(3c)". The bug on "cvtnum(3c)" is SR#4701-078451. Anyway, the attached assembler uses the archive library version of "cvtnum(3c)" and thus does not exhibit the bug. This patch is also known as PHCO_4484. @{b}*@{ub} On HP-UX version 8.05, but not on 8.07 or more recent versions, the @{b}fixproto@{ub} shell script triggers a bug in the system shell. If you encounter this problem, upgrade your operating system or use BASH (the GNU shell) to run @{b}fixproto@{ub}. @{b}*@{ub} Some versions of the Pyramid C compiler are reported to be unable to compile GNU CC. You must use an older version of GNU CC for bootstrapping. One indication of this problem is if you get a crash when GNU CC compiles the function @{b}muldi3@{ub} in file @{b}libgcc2.c@{ub}. You may be able to succeed by getting GNU CC version 1, installing it, and using it to compile GNU CC version 2. The bug in the Pyramid C compiler does not seem to affect GNU CC version 1. @{b}*@{ub} There may be similar problems on System V Release 3.1 on 386 systems. @{b}*@{ub} On the Intel Paragon (an i860 machine), if you are using operating system version 1.0, you will get warnings or errors about redefinition of @{b}va_arg@{ub} when you build GNU CC. If this happens, then you need to link most programs with the library @{b}iclib.a@{ub}. You must also modify @{b}stdio.h@{ub} as follows: before the lines #if defined(__i860__) && !defined(_VA_LIST) #include insert the line #if __PGC__ and after the lines extern int vprintf(const char *, va_list ); extern int vsprintf(char *, const char *, va_list ); #endif insert the line #endif /* __PGC__ */ These problems don't exist in operating system version 1.1. @{b}*@{ub} On the Altos 3068, programs compiled with GNU CC won't work unless you fix a kernel bug. This happens using system versions V.2.2 1.0gT1 and V.2.2 1.0e and perhaps later versions as well. See the file @{b}README.ALTOS@{ub}. @{b}*@{ub} You will get several sorts of compilation and linking errors on the we32k if you don't follow the special instructions. See @{"Configurations" Link "Configurations"}. @{b}*@{ub} A bug in the HP-UX 8.05 (and earlier) shell will cause the fixproto program to report an error of the form: ./fixproto: sh internal 1K buffer overflow To fix this, change the first line of the fixproto script to look like: #!/bin/ksh @EndNode @Node "Cross-Compiler Problems" "gcc.guide/Cross-Compiler Problems" @Next "Interoperation" @Prev "Installation Problems" @Toc "Trouble" Cross-Compiler Problems ======================= You may run into problems with cross compilation on certain machines, for several reasons. @{b}*@{ub} Cross compilation can run into trouble for certain machines because some target machines' assemblers require floating point numbers to be written as @{i}integer@{ui} constants in certain contexts. The compiler writes these integer constants by examining the floating point value as an integer and printing that integer, because this is simple to write and independent of the details of the floating point representation. But this does not work if the compiler is running on a different machine with an incompatible floating point format, or even a different byte-ordering. In addition, correct constant folding of floating point values requires representing them in the target machine's format. (The C standard does not quite require this, but in practice it is the only way to win.) It is now possible to overcome these problems by defining macros such as @{b}REAL_VALUE_TYPE@{ub}. But doing so is a substantial amount of work for each target machine. See @{"Cross-compilation" Link "Cross-compilation"}. @{b}*@{ub} At present, the program @{b}mips-tfile@{ub} which adds debug support to object files on MIPS systems does not work in a cross compile environment. @EndNode @Node "Interoperation" "gcc.guide/Interoperation" @Next "External Bugs" @Prev "Cross-Compiler Problems" @Toc "Trouble" Interoperation ============== This section lists various difficulties encountered in using GNU C or GNU C++ together with other compilers or with the assemblers, linkers, libraries and debuggers on certain systems. @{b}*@{ub} Objective C does not work on the RS/6000. @{b}*@{ub} GNU C++ does not do name mangling in the same way as other C++ compilers. This means that object files compiled with one compiler cannot be used with another. This effect is intentional, to protect you from more subtle problems. Compilers differ as to many internal details of C++ implementation, including: how class instances are laid out, how multiple inheritance is implemented, and how virtual function calls are handled. If the name encoding were made the same, your programs would link against libraries provided from other compilers--but the programs would then crash when run. Incompatible libraries are then detected at link time, rather than at run time. @{b}*@{ub} Older GDB versions sometimes fail to read the output of GNU CC version 2. If you have trouble, get GDB version 4.4 or later. @{b}*@{ub} DBX rejects some files produced by GNU CC, though it accepts similar constructs in output from PCC. Until someone can supply a coherent description of what is valid DBX input and what is not, there is nothing I can do about these problems. You are on your own. @{b}*@{ub} The GNU assembler (GAS) does not support PIC. To generate PIC code, you must use some other assembler, such as @{b}/bin/as@{ub}. @{b}*@{ub} On some BSD systems, including some versions of Ultrix, use of profiling causes static variable destructors (currently used only in C++) not to be run. @{b}*@{ub} Use of @{b}-I/usr/include@{ub} may cause trouble. Many systems come with header files that won't work with GNU CC unless corrected by @{b}fixincludes@{ub}. The corrected header files go in a new directory; GNU CC searches this directory before @{b}/usr/include@{ub}. If you use @{b}-I/usr/include@{ub}, this tells GNU CC to search @{b}/usr/include@{ub} earlier on, before the corrected headers. The result is that you get the uncorrected header files. Instead, you should use these options (when compiling C programs): -I/usr/local/lib/gcc-lib/@{I}TARGET@{ui}/@{I}VERSION@{ui}/include -I/usr/include For C++ programs, GNU CC also uses a special directory that defines C++ interfaces to standard C subroutines. This directory is meant to be searched @{i}before@{ui} other standard include directories, so that it takes precedence. If you are compiling C++ programs and specifying include directories explicitly, use this option first, then the two options above: -I/usr/local/lib/g++-include @{b}*@{ub} On some SGI systems, when you use @{b}-lgl_s@{ub} as an option, it gets translated magically to @{b}-lgl_s -lX11_s -lc_s@{ub}. Naturally, this does not happen when you use GNU CC. You must specify all three options explicitly. @{b}*@{ub} On a Sparc, GNU CC aligns all values of type @{b}double@{ub} on an 8-byte boundary, and it expects every @{b}double@{ub} to be so aligned. The Sun compiler usually gives @{b}double@{ub} values 8-byte alignment, with one exception: function arguments of type @{b}double@{ub} may not be aligned. As a result, if a function compiled with Sun CC takes the address of an argument of type @{b}double@{ub} and passes this pointer of type @{b}double *@{ub} to a function compiled with GNU CC, dereferencing the pointer may cause a fatal signal. One way to solve this problem is to compile your entire program with GNU CC. Another solution is to modify the function that is compiled with Sun CC to copy the argument into a local variable; local variables are always properly aligned. A third solution is to modify the function that uses the pointer to dereference it via the following function @{b}access_double@{ub} instead of directly with @{b}*@{ub}: inline double access_double (double *unaligned_ptr) { union d2i { double d; int i[2]; }; union d2i *p = (union d2i *) unaligned_ptr; union d2i u; u.i[0] = p->i[0]; u.i[1] = p->i[1]; return u.d; } Storing into the pointer can be done likewise with the same union. @{b}*@{ub} On Solaris, the @{b}malloc@{ub} function in the @{b}libmalloc.a@{ub} library may allocate memory that is only 4 byte aligned. Since GNU CC on the Sparc assumes that doubles are 8 byte aligned, this may result in a fatal signal if doubles are stored in memory allocated by the @{b}libmalloc.a@{ub} library. The solution is to not use the @{b}libmalloc.a@{ub} library. Use instead @{b}malloc@{ub} and related functions from @{b}libc.a@{ub}; they do not have this problem. @{b}*@{ub} Sun forgot to include a static version of @{b}libdl.a@{ub} with some versions of SunOS (mainly 4.1). This results in undefined symbols when linking static binaries (that is, if you use @{b}-static@{ub}). If you see undefined symbols @{b}_dlclose@{ub}, @{b}_dlsym@{ub} or @{b}_dlopen@{ub} when linking, compile and link against the file @{b}mit/util/misc/dlsym.c@{ub} from the MIT version of X windows. @{b}*@{ub} The 128-bit long double format that the Sparc port supports currently works by using the architecturally defined quad-word floating point instructions. Since there is no hardware that supports these instructions they must be emulated by the operating system. Long doubles do not work in Sun OS versions 4.0.3 and earlier, because the kernel emulator uses an obsolete and incompatible format. Long doubles do not work in Sun OS version 4.1.1 due to a problem in a Sun library. Long doubles do work on Sun OS versions 4.1.2 and higher, but GNU CC does not enable them by default. Long doubles appear to work in Sun OS 5.x (Solaris 2.x). @{b}*@{ub} On HP-UX version 9.01 on the HP PA, the HP compiler @{b}cc@{ub} does not compile GNU CC correctly. We do not yet know why. However, GNU CC compiled on earlier HP-UX versions works properly on HP-UX 9.01 and can compile itself properly on 9.01. @{b}*@{ub} On the HP PA machine, ADB sometimes fails to work on functions compiled with GNU CC. Specifically, it fails to work on functions that use @{b}alloca@{ub} or variable-size arrays. This is because GNU CC doesn't generate HP-UX unwind descriptors for such functions. It may even be impossible to generate them. @{b}*@{ub} Debugging (@{b}-g@{ub}) is not supported on the HP PA machine, unless you use the preliminary GNU tools (see @{"Installation" Link "Installation"}). @{b}*@{ub} Taking the address of a label may generate errors from the HP-UX PA assembler. GAS for the PA does not have this problem. @{b}*@{ub} Using floating point parameters for indirect calls to static functions will not work when using the HP assembler. There simply is no way for GCC to specify what registers hold arguments for static functions when using the HP assembler. GAS for the PA does not have this problem. @{b}*@{ub} In extremely rare cases involving some very large functions you may receive errors from the HP linker complaining about an out of bounds unconditional branch offset. This used to occur more often in previous versions of GNU CC, but is now exceptionally rare. If you should run into it, you can work around by making your function smaller. @{b}*@{ub} GNU CC compiled code sometimes emits warnings from the HP-UX assembler of the form: (warning) Use of GR3 when frame >= 8192 may cause conflict. These warnings are harmless and can be safely ignored. @{b}*@{ub} The current version of the assembler (@{b}/bin/as@{ub}) for the RS/6000 has certain problems that prevent the @{b}-g@{ub} option in GCC from working. Note that @{b}Makefile.in@{ub} uses @{b}-g@{ub} by default when compiling @{b}libgcc2.c@{ub}. IBM has produced a fixed version of the assembler. The upgraded assembler unfortunately was not included in any of the AIX 3.2 update PTF releases (3.2.2, 3.2.3, or 3.2.3e). Users of AIX 3.1 should request PTF U403044 from IBM and users of AIX 3.2 should request PTF U416277. See the file @{b}README.RS6000@{ub} for more details on these updates. You can test for the presense of a fixed assembler by using the command as -u < /dev/null If the command exits normally, the assembler fix already is installed. If the assembler complains that "-u" is an unknown flag, you need to order the fix. @{b}*@{ub} On the IBM RS/6000, compiling code of the form extern int foo; ... foo ... static int foo; will cause the linker to report an undefined symbol @{b}foo@{ub}. Although this behavior differs from most other systems, it is not a bug because redefining an @{b}extern@{ub} variable as @{b}static@{ub} is undefined in ANSI C. @{b}*@{ub} AIX on the RS/6000 provides support (NLS) for environments outside of the United States. Compilers and assemblers use NLS to support locale-specific representations of various objects including floating-point numbers ("." vs "," for separating decimal fractions). There have been problems reported where the library linked with GCC does not produce the same floating-point formats that the assembler accepts. If you have this problem, set the LANG environment variable to "C" or "En_US". @{b}*@{ub} Even if you specify @{b}-fdollars-in-identifiers@{ub}, you cannot successfully use @{b}$@{ub} in identifiers on the RS/6000 due to a restriction in the IBM assembler. GAS supports these identifiers. @{b}*@{ub} On the RS/6000, XLC version 1.3.0.0 will miscompile @{b}jump.c@{ub}. XLC version 1.3.0.1 or later fixes this problem. You can obtain XLC-1.3.0.2 by requesting PTF 421749 from IBM. @{b}*@{ub} There is an assembler bug in versions of DG/UX prior to 5.4.2.01 that occurs when the @{b}fldcr@{ub} instruction is used. GNU CC uses @{b}fldcr@{ub} on the 88100 to serialize volatile memory references. Use the option @{b}-mno-serialize-volatile@{ub} if your version of the assembler has this bug. @{b}*@{ub} On VMS, GAS versions 1.38.1 and earlier may cause spurious warning messages from the linker. These warning messages complain of mismatched psect attributes. You can ignore them. See @{"VMS Install" Link "VMS Install"}. @{b}*@{ub} On NewsOS version 3, if you include both of the files @{b}stddef.h@{ub} and @{b}sys/types.h@{ub}, you get an error because there are two typedefs of @{b}size_t@{ub}. You should change @{b}sys/types.h@{ub} by adding these lines around the definition of @{b}size_t@{ub}: #ifndef _SIZE_T #define _SIZE_T @{I}ACTUAL TYPEDEF HERE@{ui} #endif @{b}*@{ub} On the Alliant, the system's own convention for returning structures and unions is unusual, and is not compatible with GNU CC no matter what options are used. @{b}*@{ub} On the IBM RT PC, the MetaWare HighC compiler (hc) uses a different convention for structure and union returning. Use the option @{b}-mhc-struct-return@{ub} to tell GNU CC to use a convention compatible with it. @{b}*@{ub} On Ultrix, the Fortran compiler expects registers 2 through 5 to be saved by function calls. However, the C compiler uses conventions compatible with BSD Unix: registers 2 through 5 may be clobbered by function calls. GNU CC uses the same convention as the Ultrix C compiler. You can use these options to produce code compatible with the Fortran compiler: -fcall-saved-r2 -fcall-saved-r3 -fcall-saved-r4 -fcall-saved-r5 @{b}*@{ub} On the WE32k, you may find that programs compiled with GNU CC do not work with the standard shared C library. You may need to link with the ordinary C compiler. If you do so, you must specify the following options: -L/usr/local/lib/gcc-lib/we32k-att-sysv/2.7.1 -lgcc -lc_s The first specifies where to find the library @{b}libgcc.a@{ub} specified with the @{b}-lgcc@{ub} option. GNU CC does linking by invoking @{b}ld@{ub}, just as @{b}cc@{ub} does, and there is no reason why it @{i}should@{ui} matter which compilation program you use to invoke @{b}ld@{ub}. If someone tracks this problem down, it can probably be fixed easily. @{b}*@{ub} On the Alpha, you may get assembler errors about invalid syntax as a result of floating point constants. This is due to a bug in the C library functions @{b}ecvt@{ub}, @{b}fcvt@{ub} and @{b}gcvt@{ub}. Given valid floating point numbers, they sometimes print @{b}NaN@{ub}. @{b}*@{ub} On Irix 4.0.5F (and perhaps in some other versions), an assembler bug sometimes reorders instructions incorrectly when optimization is turned on. If you think this may be happening to you, try using the GNU assembler; GAS version 2.1 supports ECOFF on Irix. Or use the @{b}-noasmopt@{ub} option when you compile GNU CC with itself, and then again when you compile your program. (This is a temporary kludge to turn off assembler optimization on Irix.) If this proves to be what you need, edit the assembler spec in the file @{b}specs@{ub} so that it unconditionally passes @{b}-O0@{ub} to the assembler, and never passes @{b}-O2@{ub} or @{b}-O3@{ub}. @EndNode @Node "External Bugs" "gcc.guide/External Bugs" @Next "Incompatibilities" @Prev "Interoperation" @Toc "Trouble" Problems Compiling Certain Programs =================================== Certain programs have problems compiling. @{b}*@{ub} Parse errors may occur compiling X11 on a Decstation running Ultrix 4.2 because of problems in DEC's versions of the X11 header files @{b}X11/Xlib.h@{ub} and @{b}X11/Xutil.h@{ub}. People recommend adding @{b}-I/usr/include/mit@{ub} to use the MIT versions of the header files, using the @{b}-traditional@{ub} switch to turn off ANSI C, or fixing the header files by adding this: #ifdef __STDC__ #define NeedFunctionPrototypes 0 #endif @{b}*@{ub} If you have trouble compiling Perl on a SunOS 4 system, it may be because Perl specifies @{b}-I/usr/ucbinclude@{ub}. This accesses the unfixed header files. Perl specifies the options -traditional -Dvolatile=__volatile__ -I/usr/include/sun -I/usr/ucbinclude -fpcc-struct-return most of which are unnecessary with GCC 2.4.5 and newer versions. You can make a properly working Perl by setting @{b}ccflags@{ub} to @{b}-fwritable-strings@{ub} (implied by the @{b}-traditional@{ub} in the original options) and @{b}cppflags@{ub} to empty in @{b}config.sh@{ub}, then typing @{b}./doSH; make depend; make@{ub}. @{b}*@{ub} On various 386 Unix systems derived from System V, including SCO, ISC, and ESIX, you may get error messages about running out of virtual memory while compiling certain programs. You can prevent this problem by linking GNU CC with the GNU malloc (which thus replaces the malloc that comes with the system). GNU malloc is available as a separate package, and also in the file @{b}src/gmalloc.c@{ub} in the GNU Emacs 19 distribution. If you have installed GNU malloc as a separate library package, use this option when you relink GNU CC: MALLOC=/usr/local/lib/libgmalloc.a Alternatively, if you have compiled @{b}gmalloc.c@{ub} from Emacs 19, copy the object file to @{b}gmalloc.o@{ub} and use this option when you relink GNU CC: MALLOC=gmalloc.o @EndNode @Node "Incompatibilities" "gcc.guide/Incompatibilities" @Next "Fixed Headers" @Prev "External Bugs" @Toc "Trouble" Incompatibilities of GNU CC =========================== There are several noteworthy incompatibilities between GNU C and most existing (non-ANSI) versions of C. The @{b}-traditional@{ub} option eliminates many of these incompatibilities, @{i}but not all@{ui}, by telling GNU C to behave like the other C compilers. @{b}*@{ub} GNU CC normally makes string constants read-only. If several identical-looking string constants are used, GNU CC stores only one copy of the string. One consequence is that you cannot call @{b}mktemp@{ub} with a string constant argument. The function @{b}mktemp@{ub} always alters the string its argument points to. Another consequence is that @{b}sscanf@{ub} does not work on some systems when passed a string constant as its format control string or input. This is because @{b}sscanf@{ub} incorrectly tries to write into the string constant. Likewise @{b}fscanf@{ub} and @{b}scanf@{ub}. The best solution to these problems is to change the program to use @{b}char@{ub}-array variables with initialization strings for these purposes instead of string constants. But if this is not possible, you can use the @{b}-fwritable-strings@{ub} flag, which directs GNU CC to handle string constants the same way most C compilers do. @{b}-traditional@{ub} also has this effect, among others. @{b}*@{ub} @{b}-2147483648@{ub} is positive. This is because 2147483648 cannot fit in the type @{b}int@{ub}, so (following the ANSI C rules) its data type is @{b}unsigned long int@{ub}. Negating this value yields 2147483648 again. @{b}*@{ub} GNU CC does not substitute macro arguments when they appear inside of string constants. For example, the following macro in GNU CC #define foo(a) "a" will produce output @{b}"a"@{ub} regardless of what the argument @{I}A@{ui} is. The @{b}-traditional@{ub} option directs GNU CC to handle such cases (among others) in the old-fashioned (non-ANSI) fashion. @{b}*@{ub} When you use @{b}setjmp@{ub} and @{b}longjmp@{ub}, the only automatic variables guaranteed to remain valid are those declared @{b}volatile@{ub}. This is a consequence of automatic register allocation. Consider this function: jmp_buf j; foo () { int a, b; a = fun1 (); if (setjmp (j)) return a; a = fun2 (); /* @{i}@{b}longjmp (j)@{ub} may occur in @{b}fun3@{ub}.@{ui} */ return a + fun3 (); } Here @{b}a@{ub} may or may not be restored to its first value when the @{b}longjmp@{ub} occurs. If @{b}a@{ub} is allocated in a register, then its first value is restored; otherwise, it keeps the last value stored in it. If you use the @{b}-W@{ub} option with the @{b}-O@{ub} option, you will get a warning when GNU CC thinks such a problem might be possible. The @{b}-traditional@{ub} option directs GNU C to put variables in the stack by default, rather than in registers, in functions that call @{b}setjmp@{ub}. This results in the behavior found in traditional C compilers. @{b}*@{ub} Programs that use preprocessing directives in the middle of macro arguments do not work with GNU CC. For example, a program like this will not work: foobar ( #define luser hack) ANSI C does not permit such a construct. It would make sense to support it when @{b}-traditional@{ub} is used, but it is too much work to implement. @{b}*@{ub} Declarations of external variables and functions within a block apply only to the block containing the declaration. In other words, they have the same scope as any other declaration in the same place. In some other C compilers, a @{b}extern@{ub} declaration affects all the rest of the file even if it happens within a block. The @{b}-traditional@{ub} option directs GNU C to treat all @{b}extern@{ub} declarations as global, like traditional compilers. @{b}*@{ub} In traditional C, you can combine @{b}long@{ub}, etc., with a typedef name, as shown here: typedef int foo; typedef long foo bar; In ANSI C, this is not allowed: @{b}long@{ub} and other type modifiers require an explicit @{b}int@{ub}. Because this criterion is expressed by Bison grammar rules rather than C code, the @{b}-traditional@{ub} flag cannot alter it. @{b}*@{ub} PCC allows typedef names to be used as function parameters. The difficulty described immediately above applies here too. @{b}*@{ub} PCC allows whitespace in the middle of compound assignment operators such as @{b}+=@{ub}. GNU CC, following the ANSI standard, does not allow this. The difficulty described immediately above applies here too. @{b}*@{ub} GNU CC complains about unterminated character constants inside of preprocessing conditionals that fail. Some programs have English comments enclosed in conditionals that are guaranteed to fail; if these comments contain apostrophes, GNU CC will probably report an error. For example, this code would produce an error: #if 0 You can't expect this to work. #endif The best solution to such a problem is to put the text into an actual C comment delimited by @{b}/*...*/@{ub}. However, @{b}-traditional@{ub} suppresses these error messages. @{b}*@{ub} Many user programs contain the declaration @{b}long time ();@{ub}. In the past, the system header files on many systems did not actually declare @{b}time@{ub}, so it did not matter what type your program declared it to return. But in systems with ANSI C headers, @{b}time@{ub} is declared to return @{b}time_t@{ub}, and if that is not the same as @{b}long@{ub}, then @{b}long time ();@{ub} is erroneous. The solution is to change your program to use @{b}time_t@{ub} as the return type of @{b}time@{ub}. @{b}*@{ub} When compiling functions that return @{b}float@{ub}, PCC converts it to a double. GNU CC actually returns a @{b}float@{ub}. If you are concerned with PCC compatibility, you should declare your functions to return @{b}double@{ub}; you might as well say what you mean. @{b}*@{ub} When compiling functions that return structures or unions, GNU CC output code normally uses a method different from that used on most versions of Unix. As a result, code compiled with GNU CC cannot call a structure-returning function compiled with PCC, and vice versa. The method used by GNU CC is as follows: a structure or union which is 1, 2, 4 or 8 bytes long is returned like a scalar. A structure or union with any other size is stored into an address supplied by the caller (usually in a special, fixed register, but on some machines it is passed on the stack). The machine-description macros @{b}STRUCT_VALUE@{ub} and @{b}STRUCT_INCOMING_VALUE@{ub} tell GNU CC where to pass this address. By contrast, PCC on most target machines returns structures and unions of any size by copying the data into an area of static storage, and then returning the address of that storage as if it were a pointer value. The caller must copy the data from that memory area to the place where the value is wanted. GNU CC does not use this method because it is slower and nonreentrant. On some newer machines, PCC uses a reentrant convention for all structure and union returning. GNU CC on most of these machines uses a compatible convention when returning structures and unions in memory, but still returns small structures and unions in registers. You can tell GNU CC to use a compatible convention for all structure and union returning with the option @{b}-fpcc-struct-return@{ub}. @{b}*@{ub} GNU C complains about program fragments such as @{b}0x74ae-0x4000@{ub} which appear to be two hexadecimal constants separated by the minus operator. Actually, this string is a single @{i}preprocessing token@{ui}. Each such token must correspond to one token in C. Since this does not, GNU C prints an error message. Although it may appear obvious that what is meant is an operator and two values, the ANSI C standard specifically requires that this be treated as erroneous. A @{i}preprocessing token@{ui} is a @{i}preprocessing number@{ui} if it begins with a digit and is followed by letters, underscores, digits, periods and @{b}e+@{ub}, @{b}e-@{ub}, @{b}E+@{ub}, or @{b}E-@{ub} character sequences. To make the above program fragment valid, place whitespace in front of the minus sign. This whitespace will end the preprocessing number. @EndNode @Node "Fixed Headers" "gcc.guide/Fixed Headers" @Next "Standard Libraries" @Prev "Incompatibilities" @Toc "Trouble" Fixed Header Files ================== GNU CC needs to install corrected versions of some system header files. This is because most target systems have some header files that won't work with GNU CC unless they are changed. Some have bugs, some are incompatible with ANSI C, and some depend on special features of other compilers. Installing GNU CC automatically creates and installs the fixed header files, by running a program called @{b}fixincludes@{ub} (or for certain targets an alternative such as @{b}fixinc.svr4@{ub}). Normally, you don't need to pay attention to this. But there are cases where it doesn't do the right thing automatically. @{b}*@{ub} If you update the system's header files, such as by installing a new system version, the fixed header files of GNU CC are not automatically updated. The easiest way to update them is to reinstall GNU CC. (If you want to be clever, look in the makefile and you can find a shortcut.) @{b}*@{ub} On some systems, in particular SunOS 4, header file directories contain machine-specific symbolic links in certain places. This makes it possible to share most of the header files among hosts running the same version of SunOS 4 on different machine models. The programs that fix the header files do not understand this special way of using symbolic links; therefore, the directory of fixed header files is good only for the machine model used to build it. In SunOS 4, only programs that look inside the kernel will notice the difference between machine models. Therefore, for most purposes, you need not be concerned about this. It is possible to make separate sets of fixed header files for the different machine models, and arrange a structure of symbolic links so as to use the proper set, but you'll have to do this by hand. @{b}*@{ub} On Lynxos, GNU CC by default does not fix the header files. This is because bugs in the shell cause the @{b}fixincludes@{ub} script to fail. This means you will encounter problems due to bugs in the system header files. It may be no comfort that they aren't GNU CC's fault, but it does mean that there's nothing for us to do about them. @EndNode @Node "Standard Libraries" "gcc.guide/Standard Libraries" @Next "Disappointments" @Prev "Fixed Headers" @Toc "Trouble" Standard Libraries ================== GNU CC by itself attempts to be what the ISO/ANSI C standard calls a @{i}conforming freestanding implementation@{ui}. This means all ANSI C language features are available, as well as the contents of @{b}float.h@{ub}, @{b}limits.h@{ub}, @{b}stdarg.h@{ub}, and @{b}stddef.h@{ub}. The rest of the C library is supplied by the vendor of the operating system. If that C library doesn't conform to the C standards, then your programs might get warnings (especially when using @{b}-Wall@{ub}) that you don't expect. For example, the @{b}sprintf@{ub} function on SunOS 4.1.3 returns @{b}char *@{ub} while the C standard says that @{b}sprintf@{ub} returns an @{b}int@{ub}. The @{b}fixincludes@{ub} program could make the prototype for this function match the Standard, but that would be wrong, since the function will still return @{b}char *@{ub}. If you need a Standard compliant library, then you need to find one, as GNU CC does not provide one. The GNU C library (called @{b}glibc@{ub}) has been ported to a number of operating systems, and provides ANSI/ISO, POSIX, BSD and SystemV compatibility. You could also ask your operating system vendor if newer libraries are available. @EndNode @Node "Disappointments" "gcc.guide/Disappointments" @Next "C++ Misunderstandings" @Prev "Standard Libraries" @Toc "Trouble" Disappointments and Misunderstandings ===================================== These problems are perhaps regrettable, but we don't know any practical way around them. @{b}*@{ub} Certain local variables aren't recognized by debuggers when you compile with optimization. This occurs because sometimes GNU CC optimizes the variable out of existence. There is no way to tell the debugger how to compute the value such a variable "would have had", and it is not clear that would be desirable anyway. So GNU CC simply does not mention the eliminated variable when it writes debugging information. You have to expect a certain amount of disagreement between the executable and your source code, when you use optimization. @{b}*@{ub} Users often think it is a bug when GNU CC reports an error for code like this: int foo (struct mumble *); struct mumble { ... }; int foo (struct mumble *x) { ... } This code really is erroneous, because the scope of @{b}struct mumble@{ub} in the prototype is limited to the argument list containing it. It does not refer to the @{b}struct mumble@{ub} defined with file scope immediately below--they are two unrelated types with similar names in different scopes. But in the definition of @{b}foo@{ub}, the file-scope type is used because that is available to be inherited. Thus, the definition and the prototype do not match, and you get an error. This behavior may seem silly, but it's what the ANSI standard specifies. It is easy enough for you to make your code work by moving the definition of @{b}struct mumble@{ub} above the prototype. It's not worth being incompatible with ANSI C just to avoid an error for the example shown above. @{b}*@{ub} Accesses to bitfields even in volatile objects works by accessing larger objects, such as a byte or a word. You cannot rely on what size of object is accessed in order to read or write the bitfield; it may even vary for a given bitfield according to the precise usage. If you care about controlling the amount of memory that is accessed, use volatile but do not use bitfields. @{b}*@{ub} GNU CC comes with shell scripts to fix certain known problems in system header files. They install corrected copies of various header files in a special directory where only GNU CC will normally look for them. The scripts adapt to various systems by searching all the system header files for the problem cases that we know about. If new system header files are installed, nothing automatically arranges to update the corrected header files. You will have to reinstall GNU CC to fix the new header files. More specifically, go to the build directory and delete the files @{b}stmp-fixinc@{ub} and @{b}stmp-headers@{ub}, and the subdirectory @{b}include@{ub}; then do @{b}make install@{ub} again. @{b}*@{ub} On 68000 systems, you can get paradoxical results if you test the precise values of floating point numbers. For example, you can find that a floating point value which is not a NaN is not equal to itself. This results from the fact that the the floating point registers hold a few more bits of precision than fit in a @{b}double@{ub} in memory. Compiled code moves values between memory and floating point registers at its convenience, and moving them into memory truncates them. You can partially avoid this problem by using the @{b}-ffloat-store@{ub} option (see @{"Optimize Options" Link "Optimize Options"}). @{b}*@{ub} On the MIPS, variable argument functions using @{b}varargs.h@{ub} cannot have a floating point value for the first argument. The reason for this is that in the absence of a prototype in scope, if the first argument is a floating point, it is passed in a floating point register, rather than an integer register. If the code is rewritten to use the ANSI standard @{b}stdarg.h@{ub} method of variable arguments, and the prototype is in scope at the time of the call, everything will work fine. @EndNode @Node "C++ Misunderstandings" "gcc.guide/C++ Misunderstandings" @Next "Protoize Caveats" @Prev "Disappointments" @Toc "Trouble" Common Misunderstandings with GNU C++ ===================================== C++ is a complex language and an evolving one, and its standard definition (the ANSI C++ draft standard) is also evolving. As a result, your C++ compiler may occasionally surprise you, even when its behavior is correct. This section discusses some areas that frequently give rise to questions of this sort. @{" Static Definitions " Link "Static Definitions"} Static member declarations are not definitions @{" Temporaries " Link "Temporaries"} Temporaries may vanish before you expect @EndNode @Node "Static Definitions" "gcc.guide/Static Definitions" @Next "Temporaries" @Toc "C++ Misunderstandings" Declare @{i}and@{ui} Define Static Members --------------------------------- When a class has static data members, it is not enough to @{i}declare@{ui} the static member; you must also @{i}define@{ui} it. For example: class Foo { ... void method(); static int bar; }; This declaration only establishes that the class @{b}Foo@{ub} has an @{b}int@{ub} named @{b}Foo::bar@{ub}, and a member function named @{b}Foo::method@{ub}. But you still need to define @{i}both@{ui} @{b}method@{ub} and @{b}bar@{ub} elsewhere. According to the draft ANSI standard, you must supply an initializer in one (and only one) source file, such as: int Foo::bar = 0; Other C++ compilers may not correctly implement the standard behavior. As a result, when you switch to @{b}g++@{ub} from one of these compilers, you may discover that a program that appeared to work correctly in fact does not conform to the standard: @{b}g++@{ub} reports as undefined symbols any static data members that lack definitions. @EndNode @Node "Temporaries" "gcc.guide/Temporaries" @Prev "Static Definitions" @Toc "C++ Misunderstandings" Temporaries May Vanish Before You Expect ---------------------------------------- It is dangerous to use pointers or references to @{i}portions@{ui} of a temporary object. The compiler may very well delete the object before you expect it to, leaving a pointer to garbage. The most common place where this problem crops up is in classes like the libg++ @{b}String@{ub} class, that define a conversion function to type @{b}char *@{ub} or @{b}const char *@{ub}. However, any class that returns a pointer to some internal structure is potentially subject to this problem. For example, a program may use a function @{b}strfunc@{ub} that returns @{b}String@{ub} objects, and another function @{b}charfunc@{ub} that operates on pointers to @{b}char@{ub}: String strfunc (); void charfunc (const char *); In this situation, it may seem natural to write @{b}charfunc (strfunc ());@{ub} based on the knowledge that class @{b}String@{ub} has an explicit conversion to @{b}char@{ub} pointers. However, what really happens is akin to @{b}charfunc (strfunc ().convert ());@{ub}, where the @{b}convert@{ub} method is a function to do the same data conversion normally performed by a cast. Since the last use of the temporary @{b}String@{ub} object is the call to the conversion function, the compiler may delete that object before actually calling @{b}charfunc@{ub}. The compiler has no way of knowing that deleting the @{b}String@{ub} object will invalidate the pointer. The pointer then points to garbage, so that by the time @{b}charfunc@{ub} is called, it gets an invalid argument. Code like this may run successfully under some other compilers, especially those that delete temporaries relatively late. However, the GNU C++ behavior is also standard-conforming, so if your program depends on late destruction of temporaries it is not portable. If you think this is surprising, you should be aware that the ANSI C++ committee continues to debate the lifetime-of-temporaries problem. For now, at least, the safe way to write such code is to give the temporary a name, which forces it to remain until the end of the scope of the name. For example: String& tmp = strfunc (); charfunc (tmp); @EndNode @Node "Protoize Caveats" "gcc.guide/Protoize Caveats" @Next "Non-bugs" @Prev "C++ Misunderstandings" @Toc "Trouble" Caveats of using @{b}protoize@{ub} ========================= The conversion programs @{b}protoize@{ub} and @{b}unprotoize@{ub} can sometimes change a source file in a way that won't work unless you rearrange it. @{b}*@{ub} @{b}protoize@{ub} can insert references to a type name or type tag before the definition, or in a file where they are not defined. If this happens, compiler error messages should show you where the new references are, so fixing the file by hand is straightforward. @{b}*@{ub} There are some C constructs which @{b}protoize@{ub} cannot figure out. For example, it can't determine argument types for declaring a pointer-to-function variable; this you must do by hand. @{b}protoize@{ub} inserts a comment containing @{b}???@{ub} each time it finds such a variable; so you can find all such variables by searching for this string. ANSI C does not require declaring the argument types of pointer-to-function types. @{b}*@{ub} Using @{b}unprotoize@{ub} can easily introduce bugs. If the program relied on prototypes to bring about conversion of arguments, these conversions will not take place in the program without prototypes. One case in which you can be sure @{b}unprotoize@{ub} is safe is when you are removing prototypes that were made with @{b}protoize@{ub}; if the program worked before without any prototypes, it will work again without them. You can find all the places where this problem might occur by compiling the program with the @{b}-Wconversion@{ub} option. It prints a warning whenever an argument is converted. @{b}*@{ub} Both conversion programs can be confused if there are macro calls in and around the text to be converted. In other words, the standard syntax for a declaration or definition must not result from expanding a macro. This problem is inherent in the design of C and cannot be fixed. If only a few functions have confusing macro calls, you can easily convert them manually. @{b}*@{ub} @{b}protoize@{ub} cannot get the argument types for a function whose definition was not actually compiled due to preprocessing conditionals. When this happens, @{b}protoize@{ub} changes nothing in regard to such a function. @{b}protoize@{ub} tries to detect such instances and warn about them. You can generally work around this problem by using @{b}protoize@{ub} step by step, each time specifying a different set of @{b}-D@{ub} options for compilation, until all of the functions have been converted. There is no automatic way to verify that you have got them all, however. @{b}*@{ub} Confusion may result if there is an occasion to convert a function declaration or definition in a region of source code where there is more than one formal parameter list present. Thus, attempts to convert code containing multiple (conditionally compiled) versions of a single function header (in the same vicinity) may not produce the desired (or expected) results. If you plan on converting source files which contain such code, it is recommended that you first make sure that each conditionally compiled region of source code which contains an alternative function header also contains at least one additional follower token (past the final right parenthesis of the function header). This should circumvent the problem. @{b}*@{ub} @{b}unprotoize@{ub} can become confused when trying to convert a function definition or declaration which contains a declaration for a pointer-to-function formal argument which has the same name as the function being defined or declared. We recommand you avoid such choices of formal parameter names. @{b}*@{ub} You might also want to correct some of the indentation by hand and break long lines. (The conversion programs don't write lines longer than eighty characters in any case.) @EndNode @Node "Non-bugs" "gcc.guide/Non-bugs" @Next "Warnings and Errors" @Prev "Protoize Caveats" @Toc "Trouble" Certain Changes We Don't Want to Make ===================================== This section lists changes that people frequently request, but which we do not make because we think GNU CC is better without them. @{b}*@{ub} Checking the number and type of arguments to a function which has an old-fashioned definition and no prototype. Such a feature would work only occasionally--only for calls that appear in the same file as the called function, following the definition. The only way to check all calls reliably is to add a prototype for the function. But adding a prototype eliminates the motivation for this feature. So the feature is not worthwhile. @{b}*@{ub} Warning about using an expression whose type is signed as a shift count. Shift count operands are probably signed more often than unsigned. Warning about this would cause far more annoyance than good. @{b}*@{ub} Warning about assigning a signed value to an unsigned variable. Such assignments must be very common; warning about them would cause more annoyance than good. @{b}*@{ub} Warning about unreachable code. It's very common to have unreachable code in machine-generated programs. For example, this happens normally in some files of GNU C itself. @{b}*@{ub} Warning when a non-void function value is ignored. Coming as I do from a Lisp background, I balk at the idea that there is something dangerous about discarding a value. There are functions that return values which some callers may find useful; it makes no sense to clutter the program with a cast to @{b}void@{ub} whenever the value isn't useful. @{b}*@{ub} Assuming (for optimization) that the address of an external symbol is never zero. This assumption is false on certain systems when @{b}#pragma weak@{ub} is used. @{b}*@{ub} Making @{b}-fshort-enums@{ub} the default. This would cause storage layout to be incompatible with most other C compilers. And it doesn't seem very important, given that you can get the same result in other ways. The case where it matters most is when the enumeration-valued object is inside a structure, and in that case you can specify a field width explicitly. @{b}*@{ub} Making bitfields unsigned by default on particular machines where "the ABI standard" says to do so. The ANSI C standard leaves it up to the implementation whether a bitfield declared plain @{b}int@{ub} is signed or not. This in effect creates two alternative dialects of C. The GNU C compiler supports both dialects; you can specify the signed dialect with @{b}-fsigned-bitfields@{ub} and the unsigned dialect with @{b}-funsigned-bitfields@{ub}. However, this leaves open the question of which dialect to use by default. Currently, the preferred dialect makes plain bitfields signed, because this is simplest. Since @{b}int@{ub} is the same as @{b}signed int@{ub} in every other context, it is cleanest for them to be the same in bitfields as well. Some computer manufacturers have published Application Binary Interface standards which specify that plain bitfields should be unsigned. It is a mistake, however, to say anything about this issue in an ABI. This is because the handling of plain bitfields distinguishes two dialects of C. Both dialects are meaningful on every type of machine. Whether a particular object file was compiled using signed bitfields or unsigned is of no concern to other object files, even if they access the same bitfields in the same data structures. A given program is written in one or the other of these two dialects. The program stands a chance to work on most any machine if it is compiled with the proper dialect. It is unlikely to work at all if compiled with the wrong dialect. Many users appreciate the GNU C compiler because it provides an environment that is uniform across machines. These users would be inconvenienced if the compiler treated plain bitfields differently on certain machines. Occasionally users write programs intended only for a particular machine type. On these occasions, the users would benefit if the GNU C compiler were to support by default the same dialect as the other compilers on that machine. But such applications are rare. And users writing a program to run on more than one type of machine cannot possibly benefit from this kind of compatibility. This is why GNU CC does and will treat plain bitfields in the same fashion on all types of machines (by default). There are some arguments for making bitfields unsigned by default on all machines. If, for example, this becomes a universal de facto standard, it would make sense for GNU CC to go along with it. This is something to be considered in the future. (Of course, users strongly concerned about portability should indicate explicitly in each bitfield whether it is signed or not. In this way, they write programs which have the same meaning in both C dialects.) @{b}*@{ub} Undefining @{b}__STDC__@{ub} when @{b}-ansi@{ub} is not used. Currently, GNU CC defines @{b}__STDC__@{ub} as long as you don't use @{b}-traditional@{ub}. This provides good results in practice. Programmers normally use conditionals on @{b}__STDC__@{ub} to ask whether it is safe to use certain features of ANSI C, such as function prototypes or ANSI token concatenation. Since plain @{b}gcc@{ub} supports all the features of ANSI C, the correct answer to these questions is "yes". Some users try to use @{b}__STDC__@{ub} to check for the availability of certain library facilities. This is actually incorrect usage in an ANSI C program, because the ANSI C standard says that a conforming freestanding implementation should define @{b}__STDC__@{ub} even though it does not have the library facilities. @{b}gcc -ansi -pedantic@{ub} is a conforming freestanding implementation, and it is therefore required to define @{b}__STDC__@{ub}, even though it does not come with an ANSI C library. Sometimes people say that defining @{b}__STDC__@{ub} in a compiler that does not completely conform to the ANSI C standard somehow violates the standard. This is illogical. The standard is a standard for compilers that claim to support ANSI C, such as @{b}gcc -ansi@{ub}--not for other compilers such as plain @{b}gcc@{ub}. Whatever the ANSI C standard says is relevant to the design of plain @{b}gcc@{ub} without @{b}-ansi@{ub} only for pragmatic reasons, not as a requirement. @{b}*@{ub} Undefining @{b}__STDC__@{ub} in C++. Programs written to compile with C++-to-C translators get the value of @{b}__STDC__@{ub} that goes with the C compiler that is subsequently used. These programs must test @{b}__STDC__@{ub} to determine what kind of C preprocessor that compiler uses: whether they should concatenate tokens in the ANSI C fashion or in the traditional fashion. These programs work properly with GNU C++ if @{b}__STDC__@{ub} is defined. They would not work otherwise. In addition, many header files are written to provide prototypes in ANSI C but not in traditional C. Many of these header files can work without change in C++ provided @{b}__STDC__@{ub} is defined. If @{b}__STDC__@{ub} is not defined, they will all fail, and will all need to be changed to test explicitly for C++ as well. @{b}*@{ub} Deleting "empty" loops. GNU CC does not delete "empty" loops because the most likely reason you would put one in a program is to have a delay. Deleting them will not make real programs run any faster, so it would be pointless. It would be different if optimization of a nonempty loop could produce an empty one. But this generally can't happen. @{b}*@{ub} Making side effects happen in the same order as in some other compiler. It is never safe to depend on the order of evaluation of side effects. For example, a function call like this may very well behave differently from one compiler to another: void func (int, int); int i = 2; func (i++, i++); There is no guarantee (in either the C or the C++ standard language definitions) that the increments will be evaluated in any particular order. Either increment might happen first. @{b}func@{ub} might get the arguments @{b}2, 3@{ub}, or it might get @{b}3, 2@{ub}, or even @{b}2, 2@{ub}. @{b}*@{ub} Not allowing structures with volatile fields in registers. Strictly speaking, there is no prohibition in the ANSI C standard against allowing structures with volatile fields in registers, but it does not seem to make any sense and is probably not what you wanted to do. So the compiler will give an error message in this case. @EndNode @Node "Warnings and Errors" "gcc.guide/Warnings and Errors" @Prev "Non-bugs" @Toc "Trouble" Warning Messages and Error Messages =================================== The GNU compiler can produce two kinds of diagnostics: errors and warnings. Each kind has a different purpose: @{i}Errors@{ui} report problems that make it impossible to compile your program. GNU CC reports errors with the source file name and line number where the problem is apparent. @{i}Warnings@{ui} report other unusual conditions in your code that @{i}may@{ui} indicate a problem, although compilation can (and does) proceed. Warning messages also report the source file name and line number, but include the text @{b}warning:@{ub} to distinguish them from error messages. Warnings may indicate danger points where you should check to make sure that your program really does what you intend; or the use of obsolete features; or the use of nonstandard features of GNU C or C++. Many warnings are issued only if you ask for them, with one of the @{b}-W@{ub} options (for instance, @{b}-Wall@{ub} requests a variety of useful warnings). GNU CC always tries to compile your program if possible; it never gratuitously rejects a program whose meaning is clear merely because (for instance) it fails to conform to a standard. In some cases, however, the C and C++ standards specify that certain extensions are forbidden, and a diagnostic @{i}must@{ui} be issued by a conforming compiler. The @{b}-pedantic@{ub} option tells GNU CC to issue warnings in such cases; @{b}-pedantic-errors@{ub} says to make them errors instead. This does not mean that @{i}all@{ui} non-ANSI constructs get warnings or errors. See @{"Options to Request or Suppress Warnings" Link "Warning Options"}, for more detail on these and related command-line options. @EndNode @Node "Bugs" "gcc.guide/Bugs" @Next "Service" @Prev "Trouble" @Toc "Main" Reporting Bugs ************** Your bug reports play an essential role in making GNU CC reliable. When you encounter a problem, the first thing to do is to see if it is already known. See @{"Trouble" Link "Trouble"}. If it isn't known, then you should report the problem. Reporting a bug may help you by bringing a solution to your problem, or it may not. (If it does not, look in the service directory; see @{"Service" Link "Service"}.) In any case, the principal function of a bug report is to help the entire community by making the next version of GNU CC work better. Bug reports are your contribution to the maintenance of GNU CC. Since the maintainers are very overloaded, we cannot respond to every bug report. However, if the bug has not been fixed, we are likely to send you a patch and ask you to tell us whether it works. In order for a bug report to serve its purpose, you must include the information that makes for fixing the bug. @{" Criteria: Bug Criteria " Link "Bug Criteria"} Have you really found a bug? @{" Where: Bug Lists " Link "Bug Lists"} Where to send your bug report. @{" Reporting: Bug Reporting " Link "Bug Reporting"} How to report a bug effectively. @{" Patches: Sending Patches " Link "Sending Patches"} How to send a patch for GNU CC. @{" Known: Trouble " Link "Trouble"} Known problems. @{" Help: Service " Link "Service"} Where to ask for help. @EndNode @Node "Bug Criteria" "gcc.guide/Bug Criteria" @Next "Bug Lists" @Toc "Bugs" Have You Found a Bug? ===================== If you are not sure whether you have found a bug, here are some guidelines: @{b}*@{ub} If the compiler gets a fatal signal, for any input whatever, that is a compiler bug. Reliable compilers never crash. @{b}*@{ub} If the compiler produces invalid assembly code, for any input whatever (except an @{b}asm@{ub} statement), that is a compiler bug, unless the compiler reports errors (not just warnings) which would ordinarily prevent the assembler from being run. @{b}*@{ub} If the compiler produces valid assembly code that does not correctly execute the input source code, that is a compiler bug. However, you must double-check to make sure, because you may have run into an incompatibility between GNU C and traditional C (see @{"Incompatibilities" Link "Incompatibilities"}). These incompatibilities might be considered bugs, but they are inescapable consequences of valuable features. Or you may have a program whose behavior is undefined, which happened by chance to give the desired results with another C or C++ compiler. For example, in many nonoptimizing compilers, you can write @{b}x;@{ub} at the end of a function instead of @{b}return x;@{ub}, with the same results. But the value of the function is undefined if @{b}return@{ub} is omitted; it is not a bug when GNU CC produces different results. Problems often result from expressions with two increment operators, as in @{b}f (*p++, *p++)@{ub}. Your previous compiler might have interpreted that expression the way you intended; GNU CC might interpret it another way. Neither compiler is wrong. The bug is in your code. After you have localized the error to a single source line, it should be easy to check for these things. If your program is correct and well defined, you have found a compiler bug. @{b}*@{ub} If the compiler produces an error message for valid input, that is a compiler bug. @{b}*@{ub} If the compiler does not produce an error message for invalid input, that is a compiler bug. However, you should note that your idea of "invalid input" might be my idea of "an extension" or "support for traditional practice". @{b}*@{ub} If you are an experienced user of C or C++ compilers, your suggestions for improvement of GNU CC or GNU C++ are welcome in any case. @EndNode @Node "Bug Lists" "gcc.guide/Bug Lists" @Next "Bug Reporting" @Prev "Bug Criteria" @Toc "Bugs" Where to Report Bugs ==================== Send bug reports for GNU C to @{b}bug-gcc@prep.ai.mit.edu@{ub}. Send bug reports for GNU C++ to @{b}bug-g++@prep.ai.mit.edu@{ub}. If your bug involves the C++ class library libg++, send mail to @{b}bug-lib-g++@prep.ai.mit.edu@{ub}. If you're not sure, you can send the bug report to both lists. @{i}Do not send bug reports to @{b}help-gcc@prep.ai.mit.edu@{ub} or to the newsgroup @{b}gnu.gcc.help@{ub}.@{ui} Most users of GNU CC do not want to receive bug reports. Those that do, have asked to be on @{b}bug-gcc@{ub} and/or @{b}bug-g++@{ub}. The mailing lists @{b}bug-gcc@{ub} and @{b}bug-g++@{ub} both have newsgroups which serve as repeaters: @{b}gnu.gcc.bug@{ub} and @{b}gnu.g++.bug@{ub}. Each mailing list and its newsgroup carry exactly the same messages. Often people think of posting bug reports to the newsgroup instead of mailing them. This appears to work, but it has one problem which can be crucial: a newsgroup posting does not contain a mail path back to the sender. Thus, if maintainers need more information, they may be unable to reach you. For this reason, you should always send bug reports by mail to the proper mailing list. As a last resort, send bug reports on paper to: GNU Compiler Bugs Free Software Foundation 59 Temple Place - Suite 330 Boston, MA 02111-1307, USA @EndNode @Node "Bug Reporting" "gcc.guide/Bug Reporting" @Next "Sending Patches" @Prev "Bug Lists" @Toc "Bugs" How to Report Bugs ================== The fundamental principle of reporting bugs usefully is this: @{i}report all the facts@{ui}. If you are not sure whether to state a fact or leave it out, state it! Often people omit facts because they think they know what causes the problem and they conclude that some details don't matter. Thus, you might assume that the name of the variable you use in an example does not matter. Well, probably it doesn't, but one cannot be sure. Perhaps the bug is a stray memory reference which happens to fetch from the location where that name is stored in memory; perhaps, if the name were different, the contents of that location would fool the compiler into doing the right thing despite the bug. Play it safe and give a specific, complete example. That is the easiest thing for you to do, and the most helpful. Keep in mind that the purpose of a bug report is to enable someone to fix the bug if it is not known. It isn't very important what happens if the bug is already known. Therefore, always write your bug reports on the assumption that the bug is not known. Sometimes people give a few sketchy facts and ask, "Does this ring a bell?" This cannot help us fix a bug, so it is basically useless. We respond by asking for enough details to enable us to investigate. You might as well expedite matters by sending them to begin with. Try to make your bug report self-contained. If we have to ask you for more information, it is best if you include all the previous information in your response, as well as the information that was missing. Please report each bug in a separate message. This makes it easier for us to track which bugs have been fixed and to forward your bugs reports to the appropriate maintainer. Do not compress and encode any part of your bug report using programs such as @{b}uuencode@{ub}. If you do so it will slow down the processing of your bug. If you must submit multiple large files, use @{b}shar@{ub}, which allows us to read your message without having to run any decompression programs. To enable someone to investigate the bug, you should include all these things: @{b}*@{ub} The version of GNU CC. You can get this by running it with the @{b}-v@{ub} option. Without this, we won't know whether there is any point in looking for the bug in the current version of GNU CC. @{b}*@{ub} A complete input file that will reproduce the bug. If the bug is in the C preprocessor, send a source file and any header files that it requires. If the bug is in the compiler proper (@{b}cc1@{ub}), run your source file through the C preprocessor by doing @{b}gcc -E @{I}SOURCEFILE@{ui} > @{I}OUTFILE@{ui}@{ub}, then include the contents of @{I}OUTFILE@{ui} in the bug report. (When you do this, use the same @{b}-I@{ub}, @{b}-D@{ub} or @{b}-U@{ub} options that you used in actual compilation.) A single statement is not enough of an example. In order to compile it, it must be embedded in a complete file of compiler input; and the bug might depend on the details of how this is done. Without a real example one can compile, all anyone can do about your bug report is wish you luck. It would be futile to try to guess how to provoke the bug. For example, bugs in register allocation and reloading frequently depend on every little detail of the function they happen in. Even if the input file that fails comes from a GNU program, you should still send the complete test case. Don't ask the GNU CC maintainers to do the extra work of obtaining the program in question--they are all overworked as it is. Also, the problem may depend on what is in the header files on your system; it is unreliable for the GNU CC maintainers to try the problem with the header files available to them. By sending CPP output, you can eliminate this source of uncertainty and save us a certain percentage of wild goose chases. @{b}*@{ub} The command arguments you gave GNU CC or GNU C++ to compile that example and observe the bug. For example, did you use @{b}-O@{ub}? To guarantee you won't omit something important, list all the options. If we were to try to guess the arguments, we would probably guess wrong and then we would not encounter the bug. @{b}*@{ub} The type of machine you are using, and the operating system name and version number. @{b}*@{ub} The operands you gave to the @{b}configure@{ub} command when you installed the compiler. @{b}*@{ub} A complete list of any modifications you have made to the compiler source. (We don't promise to investigate the bug unless it happens in an unmodified compiler. But if you've made modifications and don't tell us, then you are sending us on a wild goose chase.) Be precise about these changes. A description in English is not enough--send a context diff for them. Adding files of your own (such as a machine description for a machine we don't support) is a modification of the compiler source. @{b}*@{ub} Details of any other deviations from the standard procedure for installing GNU CC. @{b}*@{ub} A description of what behavior you observe that you believe is incorrect. For example, "The compiler gets a fatal signal," or, "The assembler instruction at line 208 in the output is incorrect." Of course, if the bug is that the compiler gets a fatal signal, then one can't miss it. But if the bug is incorrect output, the maintainer might not notice unless it is glaringly wrong. None of us has time to study all the assembler code from a 50-line C program just on the chance that one instruction might be wrong. We need @{i}you@{ui} to do this part! Even if the problem you experience is a fatal signal, you should still say so explicitly. Suppose something strange is going on, such as, your copy of the compiler is out of synch, or you have encountered a bug in the C library on your system. (This has happened!) Your copy might crash and the copy here would not. If you @{i}said@{ui} to expect a crash, then when the compiler here fails to crash, we would know that the bug was not happening. If you don't say to expect a crash, then we would not know whether the bug was happening. We would not be able to draw any conclusion from our observations. If the problem is a diagnostic when compiling GNU CC with some other compiler, say whether it is a warning or an error. Often the observed symptom is incorrect output when your program is run. Sad to say, this is not enough information unless the program is short and simple. None of us has time to study a large program to figure out how it would work if compiled correctly, much less which line of it was compiled wrong. So you will have to do that. Tell us which source line it is, and what incorrect result happens when that line is executed. A person who understands the program can find this as easily as finding a bug in the program itself. @{b}*@{ub} If you send examples of assembler code output from GNU CC or GNU C++, please use @{b}-g@{ub} when you make them. The debugging information includes source line numbers which are essential for correlating the output with the input. @{b}*@{ub} If you wish to mention something in the GNU CC source, refer to it by context, not by line number. The line numbers in the development sources don't match those in your sources. Your line numbers would convey no useful information to the maintainers. @{b}*@{ub} Additional information from a debugger might enable someone to find a problem on a machine which he does not have available. However, you need to think when you collect this information if you want it to have any chance of being useful. For example, many people send just a backtrace, but that is never useful by itself. A simple backtrace with arguments conveys little about GNU CC because the compiler is largely data-driven; the same functions are called over and over for different RTL insns, doing different things depending on the details of the insn. Most of the arguments listed in the backtrace are useless because they are pointers to RTL list structure. The numeric values of the pointers, which the debugger prints in the backtrace, have no significance whatever; all that matters is the contents of the objects they point to (and most of the contents are other such pointers). In addition, most compiler passes consist of one or more loops that scan the RTL insn sequence. The most vital piece of information about such a loop--which insn it has reached--is usually in a local variable, not in an argument. What you need to provide in addition to a backtrace are the values of the local variables for several stack frames up. When a local variable or an argument is an RTX, first print its value and then use the GDB command @{b}pr@{ub} to print the RTL expression that it points to. (If GDB doesn't run on your machine, use your debugger to call the function @{b}debug_rtx@{ub} with the RTX as an argument.) In general, whenever a variable is a pointer, its value is no use without the data it points to. Here are some things that are not necessary: @{b}*@{ub} A description of the envelope of the bug. Often people who encounter a bug spend a lot of time investigating which changes to the input file will make the bug go away and which changes will not affect it. This is often time consuming and not very useful, because the way we will find the bug is by running a single example under the debugger with breakpoints, not by pure deduction from a series of examples. You might as well save your time for something else. Of course, if you can find a simpler example to report @{i}instead@{ui} of the original one, that is a convenience. Errors in the output will be easier to spot, running under the debugger will take less time, etc. Most GNU CC bugs involve just one function, so the most straightforward way to simplify an example is to delete all the function definitions except the one where the bug occurs. Those earlier in the file may be replaced by external declarations if the crucial function depends on them. (Exception: inline functions may affect compilation of functions defined later in the file.) However, simplification is not vital; if you don't want to do this, report the bug anyway and send the entire test case you used. @{b}*@{ub} In particular, some people insert conditionals @{b}#ifdef BUG@{ub} around a statement which, if removed, makes the bug not happen. These are just clutter; we won't pay any attention to them anyway. Besides, you should send us cpp output, and that can't have conditionals. @{b}*@{ub} A patch for the bug. A patch for the bug is useful if it is a good one. But don't omit the necessary information, such as the test case, on the assumption that a patch is all we need. We might see problems with your patch and decide to fix the problem another way, or we might not understand it at all. Sometimes with a program as complicated as GNU CC it is very hard to construct an example that will make the program follow a certain path through the code. If you don't send the example, we won't be able to construct one, so we won't be able to verify that the bug is fixed. And if we can't understand what bug you are trying to fix, or why your patch should be an improvement, we won't install it. A test case will help us to understand. See @{"Sending Patches" Link "Sending Patches"}, for guidelines on how to make it easy for us to understand and install your patches. @{b}*@{ub} A guess about what the bug is or what it depends on. Such guesses are usually wrong. Even I can't guess right about such things without first using the debugger to find the facts. @{b}*@{ub} A core dump file. We have no way of examining a core dump for your type of machine unless we have an identical system--and if we do have one, we should be able to reproduce the crash ourselves. @EndNode @Node "Sending Patches" "gcc.guide/Sending Patches" @Prev "Bug Reporting" @Toc "Bugs" Sending Patches for GNU CC ========================== If you would like to write bug fixes or improvements for the GNU C compiler, that is very helpful. Send suggested fixes to the bug report mailing list, @{b}bug-gcc@prep.ai.mit.edu@{ub}. Please follow these guidelines so we can study your patches efficiently. If you don't follow these guidelines, your information might still be useful, but using it will take extra work. Maintaining GNU C is a lot of work in the best of circumstances, and we can't keep up unless you do your best to help. @{b}*@{ub} Send an explanation with your changes of what problem they fix or what improvement they bring about. For a bug fix, just include a copy of the bug report, and explain why the change fixes the bug. (Referring to a bug report is not as good as including it, because then we will have to look it up, and we have probably already deleted it if we've already fixed the bug.) @{b}*@{ub} Always include a proper bug report for the problem you think you have fixed. We need to convince ourselves that the change is right before installing it. Even if it is right, we might have trouble judging it if we don't have a way to reproduce the problem. @{b}*@{ub} Include all the comments that are appropriate to help people reading the source in the future understand why this change was needed. @{b}*@{ub} Don't mix together changes made for different reasons. Send them @{i}individually@{ui}. If you make two changes for separate reasons, then we might not want to install them both. We might want to install just one. If you send them all jumbled together in a single set of diffs, we have to do extra work to disentangle them--to figure out which parts of the change serve which purpose. If we don't have time for this, we might have to ignore your changes entirely. If you send each change as soon as you have written it, with its own explanation, then the two changes never get tangled up, and we can consider each one properly without any extra work to disentangle them. Ideally, each change you send should be impossible to subdivide into parts that we might want to consider separately, because each of its parts gets its motivation from the other parts. @{b}*@{ub} Send each change as soon as that change is finished. Sometimes people think they are helping us by accumulating many changes to send them all together. As explained above, this is absolutely the worst thing you could do. Since you should send each change separately, you might as well send it right away. That gives us the option of installing it immediately if it is important. @{b}*@{ub} Use @{b}diff -c@{ub} to make your diffs. Diffs without context are hard for us to install reliably. More than that, they make it hard for us to study the diffs to decide whether we want to install them. Unidiff format is better than contextless diffs, but not as easy to read as @{b}-c@{ub} format. If you have GNU diff, use @{b}diff -cp@{ub}, which shows the name of the function that each change occurs in. @{b}*@{ub} Write the change log entries for your changes. We get lots of changes, and we don't have time to do all the change log writing ourselves. Read the @{b}ChangeLog@{ub} file to see what sorts of information to put in, and to learn the style that we use. The purpose of the change log is to show people where to find what was changed. So you need to be specific about what functions you changed; in large functions, it's often helpful to indicate where within the function the change was. On the other hand, once you have shown people where to find the change, you need not explain its purpose. Thus, if you add a new function, all you need to say about it is that it is new. If you feel that the purpose needs explaining, it probably does--but the explanation will be much more useful if you put it in comments in the code. If you would like your name to appear in the header line for who made the change, send us the header line. @{b}*@{ub} When you write the fix, keep in mind that we can't install a change that would break other systems. People often suggest fixing a problem by changing machine-independent files such as @{b}toplev.c@{ub} to do something special that a particular system needs. Sometimes it is totally obvious that such changes would break GNU CC for almost all users. We can't possibly make a change like that. At best it might tell us how to write another patch that would solve the problem acceptably. Sometimes people send fixes that @{i}might@{ui} be an improvement in general--but it is hard to be sure of this. It's hard to install such changes because we have to study them very carefully. Of course, a good explanation of the reasoning by which you concluded the change was correct can help convince us. The safest changes are changes to the configuration files for a particular machine. These are safe because they can't create new bugs on other machines. Please help us keep up with the workload by designing the patch in a form that is good to install. @EndNode @Node "Service" "gcc.guide/Service" @Next "VMS" @Prev "Bugs" @Toc "Main" How To Get Help with GNU CC *************************** If you need help installing, using or changing GNU CC, there are two ways to find it: @{b}*@{ub} Send a message to a suitable network mailing list. First try @{b}bug-gcc@prep.ai.mit.edu@{ub}, and if that brings no response, try @{b}help-gcc@prep.ai.mit.edu@{ub}. @{b}*@{ub} Look in the service directory for someone who might help you for a fee. The service directory is found in the file named @{b}SERVICE@{ub} in the GNU CC distribution. @EndNode @Node "VMS" "gcc.guide/VMS" @Next "Portability" @Prev "Service" @Toc "Main" Using GNU CC on VMS ******************* Here is how to use GNU CC on VMS. @{" Include Files and VMS " Link "Include Files and VMS"} Where the preprocessor looks for the include files. @{" Global Declarations " Link "Global Declarations"} How to do globaldef, globalref and globalvalue with GNU CC. @{" VMS Misc " Link "VMS Misc"} Misc information. @EndNode @Node "Include Files and VMS" "gcc.guide/Include Files and VMS" @Next "Global Declarations" @Toc "VMS" Include Files and VMS ===================== Due to the differences between the filesystems of Unix and VMS, GNU CC attempts to translate file names in @{b}#include@{ub} into names that VMS will understand. The basic strategy is to prepend a prefix to the specification of the include file, convert the whole filename to a VMS filename, and then try to open the file. GNU CC tries various prefixes one by one until one of them succeeds: 1. The first prefix is the @{b}GNU_CC_INCLUDE:@{ub} logical name: this is where GNU C header files are traditionally stored. If you wish to store header files in non-standard locations, then you can assign the logical @{b}GNU_CC_INCLUDE@{ub} to be a search list, where each element of the list is suitable for use with a rooted logical. 2. The next prefix tried is @{b}SYS$SYSROOT:[SYSLIB.]@{ub}. This is where VAX-C header files are traditionally stored. 3. If the include file specification by itself is a valid VMS filename, the preprocessor then uses this name with no prefix in an attempt to open the include file. 4. If the file specification is not a valid VMS filename (i.e. does not contain a device or a directory specifier, and contains a @{b}/@{ub} character), the preprocessor tries to convert it from Unix syntax to VMS syntax. Conversion works like this: the first directory name becomes a device, and the rest of the directories are converted into VMS-format directory names. For example, the name @{b}X11/foobar.h@{ub} is translated to @{b}X11:[000000]foobar.h@{ub} or @{b}X11:foobar.h@{ub}, whichever one can be opened. This strategy allows you to assign a logical name to point to the actual location of the header files. 5. If none of these strategies succeeds, the @{b}#include@{ub} fails. Include directives of the form: #include foobar are a common source of incompatibility between VAX-C and GNU CC. VAX-C treats this much like a standard @{b}#include @{ub} directive. That is incompatible with the ANSI C behavior implemented by GNU CC: to expand the name @{b}foobar@{ub} as a macro. Macro expansion should eventually yield one of the two standard formats for @{b}#include@{ub}: #include "@{I}FILE@{ui}" #include <@{I}FILE@{ui}> If you have this problem, the best solution is to modify the source to convert the @{b}#include@{ub} directives to one of the two standard forms. That will work with either compiler. If you want a quick and dirty fix, define the file names as macros with the proper expansion, like this: #define stdio This will work, as long as the name doesn't conflict with anything else in the program. Another source of incompatibility is that VAX-C assumes that: #include "foobar" is actually asking for the file @{b}foobar.h@{ub}. GNU CC does not make this assumption, and instead takes what you ask for literally; it tries to read the file @{b}foobar@{ub}. The best way to avoid this problem is to always specify the desired file extension in your include directives. GNU CC for VMS is distributed with a set of include files that is sufficient to compile most general purpose programs. Even though the GNU CC distribution does not contain header files to define constants and structures for some VMS system-specific functions, there is no reason why you cannot use GNU CC with any of these functions. You first may have to generate or create header files, either by using the public domain utility @{b}UNSDL@{ub} (which can be found on a DECUS tape), or by extracting the relevant modules from one of the system macro libraries, and using an editor to construct a C header file. A @{b}#include@{ub} file name cannot contain a DECNET node name. The preprocessor reports an I/O error if you attempt to use a node name, whether explicitly, or implicitly via a logical name. @EndNode @Node "Global Declarations" "gcc.guide/Global Declarations" @Next "VMS Misc" @Prev "Include Files and VMS" @Toc "VMS" Global Declarations and VMS =========================== GNU CC does not provide the @{b}globalref@{ub}, @{b}globaldef@{ub} and @{b}globalvalue@{ub} keywords of VAX-C. You can get the same effect with an obscure feature of GAS, the GNU assembler. (This requires GAS version 1.39 or later.) The following macros allow you to use this feature in a fairly natural way: #ifdef __GNUC__ #define GLOBALREF(TYPE,NAME) \\ TYPE NAME \\ asm ("_$$PsectAttributes_GLOBALSYMBOL$$" #NAME) #define GLOBALDEF(TYPE,NAME,VALUE) \\ TYPE NAME \\ asm ("_$$PsectAttributes_GLOBALSYMBOL$$" #NAME) \\ = VALUE #define GLOBALVALUEREF(TYPE,NAME) \\ const TYPE NAME[1] \\ asm ("_$$PsectAttributes_GLOBALVALUE$$" #NAME) #define GLOBALVALUEDEF(TYPE,NAME,VALUE) \\ const TYPE NAME[1] \\ asm ("_$$PsectAttributes_GLOBALVALUE$$" #NAME) \\ = {VALUE} #else #define GLOBALREF(TYPE,NAME) \\ globalref TYPE NAME #define GLOBALDEF(TYPE,NAME,VALUE) \\ globaldef TYPE NAME = VALUE #define GLOBALVALUEDEF(TYPE,NAME,VALUE) \\ globalvalue TYPE NAME = VALUE #define GLOBALVALUEREF(TYPE,NAME) \\ globalvalue TYPE NAME #endif (The @{b}_$$PsectAttributes_GLOBALSYMBOL@{ub} prefix at the start of the name is removed by the assembler, after it has modified the attributes of the symbol). These macros are provided in the VMS binaries distribution in a header file @{b}GNU_HACKS.H@{ub}. An example of the usage is: GLOBALREF (int, ijk); GLOBALDEF (int, jkl, 0); The macros @{b}GLOBALREF@{ub} and @{b}GLOBALDEF@{ub} cannot be used straightforwardly for arrays, since there is no way to insert the array dimension into the declaration at the right place. However, you can declare an array with these macros if you first define a typedef for the array type, like this: typedef int intvector[10]; GLOBALREF (intvector, foo); Array and structure initializers will also break the macros; you can define the initializer to be a macro of its own, or you can expand the @{b}GLOBALDEF@{ub} macro by hand. You may find a case where you wish to use the @{b}GLOBALDEF@{ub} macro with a large array, but you are not interested in explicitly initializing each element of the array. In such cases you can use an initializer like: @{b}{0,}@{ub}, which will initialize the entire array to @{b}0@{ub}. A shortcoming of this implementation is that a variable declared with @{b}GLOBALVALUEREF@{ub} or @{b}GLOBALVALUEDEF@{ub} is always an array. For example, the declaration: GLOBALVALUEREF(int, ijk); declares the variable @{b}ijk@{ub} as an array of type @{b}int [1]@{ub}. This is done because a globalvalue is actually a constant; its "value" is what the linker would normally consider an address. That is not how an integer value works in C, but it is how an array works. So treating the symbol as an array name gives consistent results--with the exception that the value seems to have the wrong type. @{i}Don't try to access an element of the array.@{ui} It doesn't have any elements. The array "address" may not be the address of actual storage. The fact that the symbol is an array may lead to warnings where the variable is used. Insert type casts to avoid the warnings. Here is an example; it takes advantage of the ANSI C feature allowing macros that expand to use the same name as the macro itself. GLOBALVALUEREF (int, ss$_normal); GLOBALVALUEDEF (int, xyzzy,123); #ifdef __GNUC__ #define ss$_normal ((int) ss$_normal) #define xyzzy ((int) xyzzy) #endif Don't use @{b}globaldef@{ub} or @{b}globalref@{ub} with a variable whose type is an enumeration type; this is not implemented. Instead, make the variable an integer, and use a @{b}globalvaluedef@{ub} for each of the enumeration values. An example of this would be: #ifdef __GNUC__ GLOBALDEF (int, color, 0); GLOBALVALUEDEF (int, RED, 0); GLOBALVALUEDEF (int, BLUE, 1); GLOBALVALUEDEF (int, GREEN, 3); #else enum globaldef color {RED, BLUE, GREEN = 3}; #endif @EndNode @Node "VMS Misc" "gcc.guide/VMS Misc" @Prev "Global Declarations" @Toc "VMS" Other VMS Issues ================ GNU CC automatically arranges for @{b}main@{ub} to return 1 by default if you fail to specify an explicit return value. This will be interpreted by VMS as a status code indicating a normal successful completion. Version 1 of GNU CC did not provide this default. GNU CC on VMS works only with the GNU assembler, GAS. You need version 1.37 or later of GAS in order to produce value debugging information for the VMS debugger. Use the ordinary VMS linker with the object files produced by GAS. Under previous versions of GNU CC, the generated code would occasionally give strange results when linked to the sharable @{b}VAXCRTL@{ub} library. Now this should work. A caveat for use of @{b}const@{ub} global variables: the @{b}const@{ub} modifier must be specified in every external declaration of the variable in all of the source files that use that variable. Otherwise the linker will issue warnings about conflicting attributes for the variable. Your program will still work despite the warnings, but the variable will be placed in writable storage. Although the VMS linker does distinguish between upper and lower case letters in global symbols, most VMS compilers convert all such symbols into upper case and most run-time library routines also have upper case names. To be able to reliably call such routines, GNU CC (by means of the assembler GAS) converts global symbols into upper case like other VMS compilers. However, since the usual practice in C is to distinguish case, GNU CC (via GAS) tries to preserve usual C behavior by augmenting each name that is not all lower case. This means truncating the name to at most 23 characters and then adding more characters at the end which encode the case pattern of those 23. Names which contain at least one dollar sign are an exception; they are converted directly into upper case without augmentation. Name augmentation yields bad results for programs that use precompiled libraries (such as Xlib) which were generated by another compiler. You can use the compiler option @{b}/NOCASE_HACK@{ub} to inhibit augmentation; it makes external C functions and variables case-independent as is usual on VMS. Alternatively, you could write all references to the functions and variables in such libraries using lower case; this will work on VMS, but is not portable to other systems. The compiler option @{b}/NAMES@{ub} also provides control over global name handling. Function and variable names are handled somewhat differently with GNU C++. The GNU C++ compiler performs @{i}name mangling@{ui} on function names, which means that it adds information to the function name to describe the data types of the arguments that the function takes. One result of this is that the name of a function can become very long. Since the VMS linker only recognizes the first 31 characters in a name, special action is taken to ensure that each function and variable has a unique name that can be represented in 31 characters. If the name (plus a name augmentation, if required) is less than 32 characters in length, then no special action is performed. If the name is longer than 31 characters, the assembler (GAS) will generate a hash string based upon the function name, truncate the function name to 23 characters, and append the hash string to the truncated name. If the @{b}/VERBOSE@{ub} compiler option is used, the assembler will print both the full and truncated names of each symbol that is truncated. The @{b}/NOCASE_HACK@{ub} compiler option should not be used when you are compiling programs that use libg++. libg++ has several instances of objects (i.e. @{b}Filebuf@{ub} and @{b}filebuf@{ub}) which become indistinguishable in a case-insensitive environment. This leads to cases where you need to inhibit augmentation selectively (if you were using libg++ and Xlib in the same program, for example). There is no special feature for doing this, but you can get the result by defining a macro for each mixed case symbol for which you wish to inhibit augmentation. The macro should expand into the lower case equivalent of itself. For example: #define StuDlyCapS studlycaps These macro definitions can be placed in a header file to minimize the number of changes to your source code. @EndNode @Node "Portability" "gcc.guide/Portability" @Next "Interface" @Prev "VMS" @Toc "Main" GNU CC and Portability ********************** The main goal of GNU CC was to make a good, fast compiler for machines in the class that the GNU system aims to run on: 32-bit machines that address 8-bit bytes and have several general registers. Elegance, theoretical power and simplicity are only secondary. GNU CC gets most of the information about the target machine from a machine description which gives an algebraic formula for each of the machine's instructions. This is a very clean way to describe the target. But when the compiler needs information that is difficult to express in this fashion, I have not hesitated to define an ad-hoc parameter to the machine description. The purpose of portability is to reduce the total work needed on the compiler; it was not of interest for its own sake. GNU CC does not contain machine dependent code, but it does contain code that depends on machine parameters such as endianness (whether the most significant byte has the highest or lowest address of the bytes in a word) and the availability of autoincrement addressing. In the RTL-generation pass, it is often necessary to have multiple strategies for generating code for a particular kind of syntax tree, strategies that are usable for different combinations of parameters. Often I have not tried to address all possible cases, but only the common ones or only the ones that I have encountered. As a result, a new target may require additional strategies. You will know if this happens because the compiler will call @{b}abort@{ub}. Fortunately, the new strategies can be added in a machine-independent fashion, and will affect only the target machines that need them. @EndNode @Node "Interface" "gcc.guide/Interface" @Next "Passes" @Prev "Portability" @Toc "Main" Interfacing to GNU CC Output **************************** GNU CC is normally configured to use the same function calling convention normally in use on the target system. This is done with the machine-description macros described (see @{"Target Macros" Link "Target Macros"}). However, returning of structure and union values is done differently on some target machines. As a result, functions compiled with PCC returning such types cannot be called from code compiled with GNU CC, and vice versa. This does not cause trouble often because few Unix library routines return structures or unions. GNU CC code returns structures and unions that are 1, 2, 4 or 8 bytes long in the same registers used for @{b}int@{ub} or @{b}double@{ub} return values. (GNU CC typically allocates variables of such types in registers also.) Structures and unions of other sizes are returned by storing them into an address passed by the caller (usually in a register). The machine-description macros @{b}STRUCT_VALUE@{ub} and @{b}STRUCT_INCOMING_VALUE@{ub} tell GNU CC where to pass this address. By contrast, PCC on most target machines returns structures and unions of any size by copying the data into an area of static storage, and then returning the address of that storage as if it were a pointer value. The caller must copy the data from that memory area to the place where the value is wanted. This is slower than the method used by GNU CC, and fails to be reentrant. On some target machines, such as RISC machines and the 80386, the standard system convention is to pass to the subroutine the address of where to return the value. On these machines, GNU CC has been configured to be compatible with the standard compiler, when this method is used. It may not be compatible for structures of 1, 2, 4 or 8 bytes. GNU CC uses the system's standard convention for passing arguments. On some machines, the first few arguments are passed in registers; in others, all are passed on the stack. It would be possible to use registers for argument passing on any machine, and this would probably result in a significant speedup. But the result would be complete incompatibility with code that follows the standard convention. So this change is practical only if you are switching to GNU CC as the sole C compiler for the system. We may implement register argument passing on certain machines once we have a complete GNU system so that we can compile the libraries with GNU CC. On some machines (particularly the Sparc), certain types of arguments are passed "by invisible reference". This means that the value is stored in memory, and the address of the memory location is passed to the subroutine. If you use @{b}longjmp@{ub}, beware of automatic variables. ANSI C says that automatic variables that are not declared @{b}volatile@{ub} have undefined values after a @{b}longjmp@{ub}. And this is all GNU CC promises to do, because it is very difficult to restore register variables correctly, and one of GNU CC's features is that it can put variables in registers without your asking it to. If you want a variable to be unaltered by @{b}longjmp@{ub}, and you don't want to write @{b}volatile@{ub} because old C compilers don't accept it, just take the address of the variable. If a variable's address is ever taken, even if just to compute it and ignore it, then the variable cannot go in a register: { int careful; &careful; ... } Code compiled with GNU CC may call certain library routines. Most of them handle arithmetic for which there are no instructions. This includes multiply and divide on some machines, and floating point operations on any machine for which floating point support is disabled with @{b}-msoft-float@{ub}. Some standard parts of the C library, such as @{b}bcopy@{ub} or @{b}memcpy@{ub}, are also called automatically. The usual function call interface is used for calling the library routines. These library routines should be defined in the library @{b}libgcc.a@{ub}, which GNU CC automatically searches whenever it links a program. On machines that have multiply and divide instructions, if hardware floating point is in use, normally @{b}libgcc.a@{ub} is not needed, but it is searched just in case. Each arithmetic function is defined in @{b}libgcc1.c@{ub} to use the corresponding C arithmetic operator. As long as the file is compiled with another C compiler, which supports all the C arithmetic operators, this file will work portably. However, @{b}libgcc1.c@{ub} does not work if compiled with GNU CC, because each arithmetic function would compile into a call to itself! @EndNode @Node "Passes" "gcc.guide/Passes" @Next "RTL" @Prev "Interface" @Toc "Main" Passes and Files of the Compiler ******************************** The overall control structure of the compiler is in @{b}toplev.c@{ub}. This file is responsible for initialization, decoding arguments, opening and closing files, and sequencing the passes. The parsing pass is invoked only once, to parse the entire input. The RTL intermediate code for a function is generated as the function is parsed, a statement at a time. Each statement is read in as a syntax tree and then converted to RTL; then the storage for the tree for the statement is reclaimed. Storage for types (and the expressions for their sizes), declarations, and a representation of the binding contours and how they nest, remain until the function is finished being compiled; these are all needed to output the debugging information. Each time the parsing pass reads a complete function definition or top-level declaration, it calls either the function @{b}rest_of_compilation@{ub}, or the function @{b}rest_of_decl_compilation@{ub} in @{b}toplev.c@{ub}, which are responsible for all further processing necessary, ending with output of the assembler language. All other compiler passes run, in sequence, within @{b}rest_of_compilation@{ub}. When that function returns from compiling a function definition, the storage used for that function definition's compilation is entirely freed, unless it is an inline function (see @{"An Inline Function is As Fast As a Macro" Link "Inline"}). Here is a list of all the passes of the compiler and their source files. Also included is a description of where debugging dumps can be requested with @{b}-d@{ub} options. @{b}*@{ub} Parsing. This pass reads the entire text of a function definition, constructing partial syntax trees. This and RTL generation are no longer truly separate passes (formerly they were), but it is easier to think of them as separate. The tree representation does not entirely follow C syntax, because it is intended to support other languages as well. Language-specific data type analysis is also done in this pass, and every tree node that represents an expression has a data type attached. Variables are represented as declaration nodes. Constant folding and some arithmetic simplifications are also done during this pass. The language-independent source files for parsing are @{b}stor-layout.c@{ub}, @{b}fold-const.c@{ub}, and @{b}tree.c@{ub}. There are also header files @{b}tree.h@{ub} and @{b}tree.def@{ub} which define the format of the tree representation. The source files to parse C are @{b}c-parse.in@{ub}, @{b}c-decl.c@{ub}, @{b}c-typeck.c@{ub}, @{b}c-aux-info.c@{ub}, @{b}c-convert.c@{ub}, and @{b}c-lang.c@{ub} along with header files @{b}c-lex.h@{ub}, and @{b}c-tree.h@{ub}. The source files for parsing C++ are @{b}cp-parse.y@{ub}, @{b}cp-class.c@{ub}, @{b}cp-cvt.c@{ub}, @{b}cp-decl.c@{ub}, @{b}cp-decl2.c@{ub}, @{b}cp-dem.c@{ub}, @{b}cp-except.c@{ub}, @{b}cp-expr.c@{ub}, @{b}cp-init.c@{ub}, @{b}cp-lex.c@{ub}, @{b}cp-method.c@{ub}, @{b}cp-ptree.c@{ub}, @{b}cp-search.c@{ub}, @{b}cp-tree.c@{ub}, @{b}cp-type2.c@{ub}, and @{b}cp-typeck.c@{ub}, along with header files @{b}cp-tree.def@{ub}, @{b}cp-tree.h@{ub}, and @{b}cp-decl.h@{ub}. The special source files for parsing Objective C are @{b}objc-parse.y@{ub}, @{b}objc-actions.c@{ub}, @{b}objc-tree.def@{ub}, and @{b}objc-actions.h@{ub}. Certain C-specific files are used for this as well. The file @{b}c-common.c@{ub} is also used for all of the above languages. @{b}*@{ub} RTL generation. This is the conversion of syntax tree into RTL code. It is actually done statement-by-statement during parsing, but for most purposes it can be thought of as a separate pass. This is where the bulk of target-parameter-dependent code is found, since often it is necessary for strategies to apply only when certain standard kinds of instructions are available. The purpose of named instruction patterns is to provide this information to the RTL generation pass. Optimization is done in this pass for @{b}if@{ub}-conditions that are comparisons, boolean operations or conditional expressions. Tail recursion is detected at this time also. Decisions are made about how best to arrange loops and how to output @{b}switch@{ub} statements. The source files for RTL generation include @{b}stmt.c@{ub}, @{b}calls.c@{ub}, @{b}expr.c@{ub}, @{b}explow.c@{ub}, @{b}expmed.c@{ub}, @{b}function.c@{ub}, @{b}optabs.c@{ub} and @{b}emit-rtl.c@{ub}. Also, the file @{b}insn-emit.c@{ub}, generated from the machine description by the program @{b}genemit@{ub}, is used in this pass. The header file @{b}expr.h@{ub} is used for communication within this pass. The header files @{b}insn-flags.h@{ub} and @{b}insn-codes.h@{ub}, generated from the machine description by the programs @{b}genflags@{ub} and @{b}gencodes@{ub}, tell this pass which standard names are available for use and which patterns correspond to them. Aside from debugging information output, none of the following passes refers to the tree structure representation of the function (only part of which is saved). The decision of whether the function can and should be expanded inline in its subsequent callers is made at the end of rtl generation. The function must meet certain criteria, currently related to the size of the function and the types and number of parameters it has. Note that this function may contain loops, recursive calls to itself (tail-recursive functions can be inlined!), gotos, in short, all constructs supported by GNU CC. The file @{b}integrate.c@{ub} contains the code to save a function's rtl for later inlining and to inline that rtl when the function is called. The header file @{b}integrate.h@{ub} is also used for this purpose. The option @{b}-dr@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.rtl@{ub} to the input file name. @{b}*@{ub} Jump optimization. This pass simplifies jumps to the following instruction, jumps across jumps, and jumps to jumps. It deletes unreferenced labels and unreachable code, except that unreachable code that contains a loop is not recognized as unreachable in this pass. (Such loops are deleted later in the basic block analysis.) It also converts some code originally written with jumps into sequences of instructions that directly set values from the results of comparisons, if the machine has such instructions. Jump optimization is performed two or three times. The first time is immediately following RTL generation. The second time is after CSE, but only if CSE says repeated jump optimization is needed. The last time is right before the final pass. That time, cross-jumping and deletion of no-op move instructions are done together with the optimizations described above. The source file of this pass is @{b}jump.c@{ub}. The option @{b}-dj@{ub} causes a debugging dump of the RTL code after this pass is run for the first time. This dump file's name is made by appending @{b}.jump@{ub} to the input file name. @{b}*@{ub} Register scan. This pass finds the first and last use of each register, as a guide for common subexpression elimination. Its source is in @{b}regclass.c@{ub}. @{b}*@{ub} Jump threading. This pass detects a condition jump that branches to an identical or inverse test. Such jumps can be @{b}threaded@{ub} through the second conditional test. The source code for this pass is in @{b}jump.c@{ub}. This optimization is only performed if @{b}-fthread-jumps@{ub} is enabled. @{b}*@{ub} Common subexpression elimination. This pass also does constant propagation. Its source file is @{b}cse.c@{ub}. If constant propagation causes conditional jumps to become unconditional or to become no-ops, jump optimization is run again when CSE is finished. The option @{b}-ds@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.cse@{ub} to the input file name. @{b}*@{ub} Loop optimization. This pass moves constant expressions out of loops, and optionally does strength-reduction and loop unrolling as well. Its source files are @{b}loop.c@{ub} and @{b}unroll.c@{ub}, plus the header @{b}loop.h@{ub} used for communication between them. Loop unrolling uses some functions in @{b}integrate.c@{ub} and the header @{b}integrate.h@{ub}. The option @{b}-dL@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.loop@{ub} to the input file name. @{b}*@{ub} If @{b}-frerun-cse-after-loop@{ub} was enabled, a second common subexpression elimination pass is performed after the loop optimization pass. Jump threading is also done again at this time if it was specified. The option @{b}-dt@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.cse2@{ub} to the input file name. @{b}*@{ub} Stupid register allocation is performed at this point in a nonoptimizing compilation. It does a little data flow analysis as well. When stupid register allocation is in use, the next pass executed is the reloading pass; the others in between are skipped. The source file is @{b}stupid.c@{ub}. @{b}*@{ub} Data flow analysis (@{b}flow.c@{ub}). This pass divides the program into basic blocks (and in the process deletes unreachable loops); then it computes which pseudo-registers are live at each point in the program, and makes the first instruction that uses a value point at the instruction that computed the value. This pass also deletes computations whose results are never used, and combines memory references with add or subtract instructions to make autoincrement or autodecrement addressing. The option @{b}-df@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.flow@{ub} to the input file name. If stupid register allocation is in use, this dump file reflects the full results of such allocation. @{b}*@{ub} Instruction combination (@{b}combine.c@{ub}). This pass attempts to combine groups of two or three instructions that are related by data flow into single instructions. It combines the RTL expressions for the instructions by substitution, simplifies the result using algebra, and then attempts to match the result against the machine description. The option @{b}-dc@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.combine@{ub} to the input file name. @{b}*@{ub} Instruction scheduling (@{b}sched.c@{ub}). This pass looks for instructions whose output will not be available by the time that it is used in subsequent instructions. (Memory loads and floating point instructions often have this behavior on RISC machines). It re-orders instructions within a basic block to try to separate the definition and use of items that otherwise would cause pipeline stalls. Instruction scheduling is performed twice. The first time is immediately after instruction combination and the second is immediately after reload. The option @{b}-dS@{ub} causes a debugging dump of the RTL code after this pass is run for the first time. The dump file's name is made by appending @{b}.sched@{ub} to the input file name. @{b}*@{ub} Register class preferencing. The RTL code is scanned to find out which register class is best for each pseudo register. The source file is @{b}regclass.c@{ub}. @{b}*@{ub} Local register allocation (@{b}local-alloc.c@{ub}). This pass allocates hard registers to pseudo registers that are used only within one basic block. Because the basic block is linear, it can use fast and powerful techniques to do a very good job. The option @{b}-dl@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.lreg@{ub} to the input file name. @{b}*@{ub} Global register allocation (@{b}global.c@{ub}). This pass allocates hard registers for the remaining pseudo registers (those whose life spans are not contained in one basic block). @{b}*@{ub} Reloading. This pass renumbers pseudo registers with the hardware registers numbers they were allocated. Pseudo registers that did not get hard registers are replaced with stack slots. Then it finds instructions that are invalid because a value has failed to end up in a register, or has ended up in a register of the wrong kind. It fixes up these instructions by reloading the problematical values temporarily into registers. Additional instructions are generated to do the copying. The reload pass also optionally eliminates the frame pointer and inserts instructions to save and restore call-clobbered registers around calls. Source files are @{b}reload.c@{ub} and @{b}reload1.c@{ub}, plus the header @{b}reload.h@{ub} used for communication between them. The option @{b}-dg@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.greg@{ub} to the input file name. @{b}*@{ub} Instruction scheduling is repeated here to try to avoid pipeline stalls due to memory loads generated for spilled pseudo registers. The option @{b}-dR@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.sched2@{ub} to the input file name. @{b}*@{ub} Jump optimization is repeated, this time including cross-jumping and deletion of no-op move instructions. The option @{b}-dJ@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.jump2@{ub} to the input file name. @{b}*@{ub} Delayed branch scheduling. This optional pass attempts to find instructions that can go into the delay slots of other instructions, usually jumps and calls. The source file name is @{b}reorg.c@{ub}. The option @{b}-dd@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.dbr@{ub} to the input file name. @{b}*@{ub} Conversion from usage of some hard registers to usage of a register stack may be done at this point. Currently, this is supported only for the floating-point registers of the Intel 80387 coprocessor. The source file name is @{b}reg-stack.c@{ub}. The options @{b}-dk@{ub} causes a debugging dump of the RTL code after this pass. This dump file's name is made by appending @{b}.stack@{ub} to the input file name. @{b}*@{ub} Final. This pass outputs the assembler code for the function. It is also responsible for identifying spurious test and compare instructions. Machine-specific peephole optimizations are performed at the same time. The function entry and exit sequences are generated directly as assembler code in this pass; they never exist as RTL. The source files are @{b}final.c@{ub} plus @{b}insn-output.c@{ub}; the latter is generated automatically from the machine description by the tool @{b}genoutput@{ub}. The header file @{b}conditions.h@{ub} is used for communication between these files. @{b}*@{ub} Debugging information output. This is run after final because it must output the stack slot offsets for pseudo registers that did not get hard registers. Source files are @{b}dbxout.c@{ub} for DBX symbol table format, @{b}sdbout.c@{ub} for SDB symbol table format, and @{b}dwarfout.c@{ub} for DWARF symbol table format. Some additional files are used by all or many passes: @{b}*@{ub} Every pass uses @{b}machmode.def@{ub} and @{b}machmode.h@{ub} which define the machine modes. @{b}*@{ub} Several passes use @{b}real.h@{ub}, which defines the default representation of floating point constants and how to operate on them. @{b}*@{ub} All the passes that work with RTL use the header files @{b}rtl.h@{ub} and @{b}rtl.def@{ub}, and subroutines in file @{b}rtl.c@{ub}. The tools @{b}gen*@{ub} also use these files to read and work with the machine description RTL. @{b}*@{ub} Several passes refer to the header file @{b}insn-config.h@{ub} which contains a few parameters (C macro definitions) generated automatically from the machine description RTL by the tool @{b}genconfig@{ub}. @{b}*@{ub} Several passes use the instruction recognizer, which consists of @{b}recog.c@{ub} and @{b}recog.h@{ub}, plus the files @{b}insn-recog.c@{ub} and @{b}insn-extract.c@{ub} that are generated automatically from the machine description by the tools @{b}genrecog@{ub} and @{b}genextract@{ub}. @{b}*@{ub} Several passes use the header files @{b}regs.h@{ub} which defines the information recorded about pseudo register usage, and @{b}basic-block.h@{ub} which defines the information recorded about basic blocks. @{b}*@{ub} @{b}hard-reg-set.h@{ub} defines the type @{b}HARD_REG_SET@{ub}, a bit-vector with a bit for each hard register, and some macros to manipulate it. This type is just @{b}int@{ub} if the machine has few enough hard registers; otherwise it is an array of @{b}int@{ub} and some of the macros expand into loops. @{b}*@{ub} Several passes use instruction attributes. A definition of the attributes defined for a particular machine is in file @{b}insn-attr.h@{ub}, which is generated from the machine description by the program @{b}genattr@{ub}. The file @{b}insn-attrtab.c@{ub} contains subroutines to obtain the attribute values for insns. It is generated from the machine description by the program @{b}genattrtab@{ub}. @EndNode @Node "RTL" "gcc.guide/RTL" @Next "Machine Desc" @Prev "Passes" @Toc "Main" RTL Representation ****************** Most of the work of the compiler is done on an intermediate representation called register transfer language. In this language, the instructions to be output are described, pretty much one by one, in an algebraic form that describes what the instruction does. RTL is inspired by Lisp lists. It has both an internal form, made up of structures that point at other structures, and a textual form that is used in the machine description and in printed debugging dumps. The textual form uses nested parentheses to indicate the pointers in the internal form. @{" RTL Objects " Link "RTL Objects"} Expressions vs vectors vs strings vs integers. @{" Accessors " Link "Accessors"} Macros to access expression operands or vector elts. @{" Flags " Link "Flags"} Other flags in an RTL expression. @{" Machine Modes " Link "Machine Modes"} Describing the size and format of a datum. @{" Constants " Link "Constants"} Expressions with constant values. @{" Regs and Memory " Link "Regs and Memory"} Expressions representing register contents or memory. @{" Arithmetic " Link "Arithmetic"} Expressions representing arithmetic on other expressions. @{" Comparisons " Link "Comparisons"} Expressions representing comparison of expressions. @{" Bit Fields " Link "Bit Fields"} Expressions representing bitfields in memory or reg. @{" Conversions " Link "Conversions"} Extending, truncating, floating or fixing. @{" RTL Declarations " Link "RTL Declarations"} Declaring volatility, constancy, etc. @{" Side Effects " Link "Side Effects"} Expressions for storing in registers, etc. @{" Incdec " Link "Incdec"} Embedded side-effects for autoincrement addressing. @{" Assembler " Link "Assembler"} Representing @{b}asm@{ub} with operands. @{" Insns " Link "Insns"} Expression types for entire insns. @{" Calls " Link "Calls"} RTL representation of function call insns. @{" Sharing " Link "Sharing"} Some expressions are unique; others *must* be copied. @{" Reading RTL " Link "Reading RTL"} Reading textual RTL from a file. @EndNode @Node "RTL Objects" "gcc.guide/RTL Objects" @Next "Accessors" @Prev "RTL" @Toc "RTL" RTL Object Types ================ RTL uses five kinds of objects: expressions, integers, wide integers, strings and vectors. Expressions are the most important ones. An RTL expression ("RTX", for short) is a C structure, but it is usually referred to with a pointer; a type that is given the typedef name @{b}rtx@{ub}. An integer is simply an @{b}int@{ub}; their written form uses decimal digits. A wide integer is an integral object whose type is @{b}HOST_WIDE_INT@{ub} (see @{"Config" Link "Config"}); their written form uses decimal digits. A string is a sequence of characters. In core it is represented as a @{b}char *@{ub} in usual C fashion, and it is written in C syntax as well. However, strings in RTL may never be null. If you write an empty string in a machine description, it is represented in core as a null pointer rather than as a pointer to a null character. In certain contexts, these null pointers instead of strings are valid. Within RTL code, strings are most commonly found inside @{b}symbol_ref@{ub} expressions, but they appear in other contexts in the RTL expressions that make up machine descriptions. A vector contains an arbitrary number of pointers to expressions. The number of elements in the vector is explicitly present in the vector. The written form of a vector consists of square brackets (@{b} [...]@{ub}) surrounding the elements, in sequence and with whitespace separating them. Vectors of length zero are not created; null pointers are used instead. Expressions are classified by @{i}expression codes@{ui} (also called RTX codes). The expression code is a name defined in @{b}rtl.def@{ub}, which is also (in upper case) a C enumeration constant. The possible expression codes and their meanings are machine-independent. The code of an RTX can be extracted with the macro @{b}GET_CODE (@{I}X@{ui})@{ub} and altered with @{b}PUT_CODE (@{I}X@{ui}, @{I}NEWCODE@{ui})@{ub}. The expression code determines how many operands the expression contains, and what kinds of objects they are. In RTL, unlike Lisp, you cannot tell by looking at an operand what kind of object it is. Instead, you must know from its context--from the expression code of the containing expression. For example, in an expression of code @{b}subreg@{ub}, the first operand is to be regarded as an expression and the second operand as an integer. In an expression of code @{b}plus@{ub}, there are two operands, both of which are to be regarded as expressions. In a @{b}symbol_ref@{ub} expression, there is one operand, which is to be regarded as a string. Expressions are written as parentheses containing the name of the expression type, its flags and machine mode if any, and then the operands of the expression (separated by spaces). Expression code names in the @{b}md@{ub} file are written in lower case, but when they appear in C code they are written in upper case. In this manual, they are shown as follows: @{b}const_int@{ub}. In a few contexts a null pointer is valid where an expression is normally wanted. The written form of this is @{b}(nil)@{ub}. @EndNode @Node "Accessors" "gcc.guide/Accessors" @Next "Flags" @Prev "RTL Objects" @Toc "RTL" Access to Operands ================== For each expression type @{b}rtl.def@{ub} specifies the number of contained objects and their kinds, with four possibilities: @{b}e@{ub} for expression (actually a pointer to an expression), @{b}i@{ub} for integer, @{b}w@{ub} for wide integer, @{b}s@{ub} for string, and @{b}E@{ub} for vector of expressions. The sequence of letters for an expression code is called its @{i}format@{ui}. Thus, the format of @{b}subreg@{ub} is @{b}ei@{ub}. A few other format characters are used occasionally: @{b}u@{ub} @{b}u@{ub} is equivalent to @{b}e@{ub} except that it is printed differently in debugging dumps. It is used for pointers to insns. @{b}n@{ub} @{b}n@{ub} is equivalent to @{b}i@{ub} except that it is printed differently in debugging dumps. It is used for the line number or code number of a @{b}note@{ub} insn. @{b}S@{ub} @{b}S@{ub} indicates a string which is optional. In the RTL objects in core, @{b}S@{ub} is equivalent to @{b}s@{ub}, but when the object is read, from an @{b}md@{ub} file, the string value of this operand may be omitted. An omitted string is taken to be the null string. @{b}V@{ub} @{b}V@{ub} indicates a vector which is optional. In the RTL objects in core, @{b}V@{ub} is equivalent to @{b}E@{ub}, but when the object is read from an @{b}md@{ub} file, the vector value of this operand may be omitted. An omitted vector is effectively the same as a vector of no elements. @{b}0@{ub} @{b}0@{ub} means a slot whose contents do not fit any normal category. @{b}0@{ub} slots are not printed at all in dumps, and are often used in special ways by small parts of the compiler. There are macros to get the number of operands, the format, and the class of an expression code: @{b}GET_RTX_LENGTH (@{I}CODE@{ui})@{ub} Number of operands of an RTX of code @{I}CODE@{ui}. @{b}GET_RTX_FORMAT (@{I}CODE@{ui})@{ub} The format of an RTX of code @{I}CODE@{ui}, as a C string. @{b}GET_RTX_CLASS (@{I}CODE@{ui})@{ub} A single character representing the type of RTX operation that code @{I}CODE@{ui} performs. The following classes are defined: @{b}o@{ub} An RTX code that represents an actual object, such as @{b}reg@{ub} or @{b}mem@{ub}. @{b}subreg@{ub} is not in this class. @{b}<@{ub} An RTX code for a comparison. The codes in this class are @{b}NE@{ub}, @{b}EQ@{ub}, @{b}LE@{ub}, @{b}LT@{ub}, @{b}GE@{ub}, @{b}GT@{ub}, @{b}LEU@{ub}, @{b}LTU@{ub}, @{b}GEU@{ub}, @{b}GTU@{ub}. @{b}1@{ub} An RTX code for a unary arithmetic operation, such as @{b}neg@{ub}. @{b}c@{ub} An RTX code for a commutative binary operation, other than @{b}NE@{ub} and @{b}EQ@{ub} (which have class @{b}<@{ub}). @{b}2@{ub} An RTX code for a noncommutative binary operation, such as @{b}MINUS@{ub}. @{b}b@{ub} An RTX code for a bitfield operation, either @{b}ZERO_EXTRACT@{ub} or @{b}SIGN_EXTRACT@{ub}. @{b}3@{ub} An RTX code for other three input operations, such as @{b}IF_THEN_ELSE@{ub}. @{b}i@{ub} An RTX code for a machine insn (@{b}INSN@{ub}, @{b}JUMP_INSN@{ub}, and @{b}CALL_INSN@{ub}). @{b}m@{ub} An RTX code for something that matches in insns, such as @{b}MATCH_DUP@{ub}. @{b}x@{ub} All other RTX codes. Operands of expressions are accessed using the macros @{b}XEXP@{ub}, @{b}XINT@{ub}, @{b}XWINT@{ub} and @{b}XSTR@{ub}. Each of these macros takes two arguments: an expression-pointer (RTX) and an operand number (counting from zero). Thus, XEXP (@{I}X@{ui}, 2) accesses operand 2 of expression @{I}X@{ui}, as an expression. XINT (@{I}X@{ui}, 2) accesses the same operand as an integer. @{b}XSTR@{ub}, used in the same fashion, would access it as a string. Any operand can be accessed as an integer, as an expression or as a string. You must choose the correct method of access for the kind of value actually stored in the operand. You would do this based on the expression code of the containing expression. That is also how you would know how many operands there are. For example, if @{I}X@{ui} is a @{b}subreg@{ub} expression, you know that it has two operands which can be correctly accessed as @{b}XEXP (@{I}X@{ui}, 0)@{ub} and @{b}XINT (@{I}X@{ui}, 1)@{ub}. If you did @{b}XINT (@{I}X@{ui}, 0)@{ub}, you would get the address of the expression operand but cast as an integer; that might occasionally be useful, but it would be cleaner to write @{b}(int) XEXP (@{I}X@{ui}, 0)@{ub}. @{b}XEXP (@{I}X@{ui}, 1)@{ub} would also compile without error, and would return the second, integer operand cast as an expression pointer, which would probably result in a crash when accessed. Nothing stops you from writing @{b}XEXP (@{I}X@{ui}, 28)@{ub} either, but this will access memory past the end of the expression with unpredictable results. Access to operands which are vectors is more complicated. You can use the macro @{b}XVEC@{ub} to get the vector-pointer itself, or the macros @{b}XVECEXP@{ub} and @{b}XVECLEN@{ub} to access the elements and length of a vector. @{b}XVEC (@{I}EXP@{ui}, @{I}IDX@{ui})@{ub} Access the vector-pointer which is operand number @{I}IDX@{ui} in @{I}EXP@{ui}. @{b}XVECLEN (@{I}EXP@{ui}, @{I}IDX@{ui})@{ub} Access the length (number of elements) in the vector which is in operand number @{I}IDX@{ui} in @{I}EXP@{ui}. This value is an @{b}int@{ub}. @{b}XVECEXP (@{I}EXP@{ui}, @{I}IDX@{ui}, @{I}ELTNUM@{ui})@{ub} Access element number @{I}ELTNUM@{ui} in the vector which is in operand number @{I}IDX@{ui} in @{I}EXP@{ui}. This value is an RTX. It is up to you to make sure that @{I}ELTNUM@{ui} is not negative and is less than @{b}XVECLEN (@{I}EXP@{ui}, @{I}IDX@{ui})@{ub}. All the macros defined in this section expand into lvalues and therefore can be used to assign the operands, lengths and vector elements as well as to access them. @EndNode @Node "Flags" "gcc.guide/Flags" @Next "Machine Modes" @Prev "Accessors" @Toc "RTL" Flags in an RTL Expression ========================== RTL expressions contain several flags (one-bit bitfields) that are used in certain types of expression. Most often they are accessed with the following macros: @{b}MEM_VOLATILE_P (@{I}X@{ui})@{ub} In @{b}mem@{ub} expressions, nonzero for volatile memory references. Stored in the @{b}volatil@{ub} field and printed as @{b}/v@{ub}. @{b}MEM_IN_STRUCT_P (@{I}X@{ui})@{ub} In @{b}mem@{ub} expressions, nonzero for reference to an entire structure, union or array, or to a component of one. Zero for references to a scalar variable or through a pointer to a scalar. Stored in the @{b}in_struct@{ub} field and printed as @{b}/s@{ub}. @{b}REG_LOOP_TEST_P@{ub} In @{b}reg@{ub} expressions, nonzero if this register's entire life is contained in the exit test code for some loop. Stored in the @{b}in_struct@{ub} field and printed as @{b}/s@{ub}. @{b}REG_USERVAR_P (@{I}X@{ui})@{ub} In a @{b}reg@{ub}, nonzero if it corresponds to a variable present in the user's source code. Zero for temporaries generated internally by the compiler. Stored in the @{b}volatil@{ub} field and printed as @{b}/v@{ub}. @{b}REG_FUNCTION_VALUE_P (@{I}X@{ui})@{ub} Nonzero in a @{b}reg@{ub} if it is the place in which this function's value is going to be returned. (This happens only in a hard register.) Stored in the @{b}integrated@{ub} field and printed as @{b}/i@{ub}. The same hard register may be used also for collecting the values of functions called by this one, but @{b}REG_FUNCTION_VALUE_P@{ub} is zero in this kind of use. @{b}SUBREG_PROMOTED_VAR_P@{ub} Nonzero in a @{b}subreg@{ub} if it was made when accessing an object that was promoted to a wider mode in accord with the @{b}PROMOTED_MODE@{ub} machine description macro (see @{"Storage Layout" Link "Storage Layout"}). In this case, the mode of the @{b}subreg@{ub} is the declared mode of the object and the mode of @{b}SUBREG_REG@{ub} is the mode of the register that holds the object. Promoted variables are always either sign- or zero-extended to the wider mode on every assignment. Stored in the @{b}in_struct@{ub} field and printed as @{b}/s@{ub}. @{b}SUBREG_PROMOTED_UNSIGNED_P@{ub} Nonzero in a @{b}subreg@{ub} that has @{b}SUBREG_PROMOTED_VAR_P@{ub} nonzero if the object being referenced is kept zero-extended and zero if it is kept sign-extended. Stored in the @{b}unchanging@{ub} field and printed as @{b}/u@{ub}. @{b}RTX_UNCHANGING_P (@{I}X@{ui})@{ub} Nonzero in a @{b}reg@{ub} or @{b}mem@{ub} if the value is not changed. (This flag is not set for memory references via pointers to constants. Such pointers only guarantee that the object will not be changed explicitly by the current function. The object might be changed by other functions or by aliasing.) Stored in the @{b}unchanging@{ub} field and printed as @{b}/u@{ub}. @{b}RTX_INTEGRATED_P (@{I}INSN@{ui})@{ub} Nonzero in an insn if it resulted from an in-line function call. Stored in the @{b}integrated@{ub} field and printed as @{b}/i@{ub}. This may be deleted; nothing currently depends on it. @{b}SYMBOL_REF_USED (@{I}X@{ui})@{ub} In a @{b}symbol_ref@{ub}, indicates that @{I}X@{ui} has been used. This is normally only used to ensure that @{I}X@{ui} is only declared external once. Stored in the @{b}used@{ub} field. @{b}SYMBOL_REF_FLAG (@{I}X@{ui})@{ub} In a @{b}symbol_ref@{ub}, this is used as a flag for machine-specific purposes. Stored in the @{b}volatil@{ub} field and printed as @{b}/v@{ub}. @{b}LABEL_OUTSIDE_LOOP_P@{ub} In @{b}label_ref@{ub} expressions, nonzero if this is a reference to a label that is outside the innermost loop containing the reference to the label. Stored in the @{b}in_struct@{ub} field and printed as @{b}/s@{ub}. @{b}INSN_DELETED_P (@{I}INSN@{ui})@{ub} In an insn, nonzero if the insn has been deleted. Stored in the @{b}volatil@{ub} field and printed as @{b}/v@{ub}. @{b}INSN_ANNULLED_BRANCH_P (@{I}INSN@{ui})@{ub} In an @{b}insn@{ub} in the delay slot of a branch insn, indicates that an annulling branch should be used. See the discussion under @{b}sequence@{ub} below. Stored in the @{b}unchanging@{ub} field and printed as @{b}/u@{ub}. @{b}INSN_FROM_TARGET_P (@{I}INSN@{ui})@{ub} In an @{b}insn@{ub} in a delay slot of a branch, indicates that the insn is from the target of the branch. If the branch insn has @{b}INSN_ANNULLED_BRANCH_P@{ub} set, this insn should only be executed if the branch is taken. For annulled branches with this bit clear, the insn should be executed only if the branch is not taken. Stored in the @{b}in_struct@{ub} field and printed as @{b}/s@{ub}. @{b}CONSTANT_POOL_ADDRESS_P (@{I}X@{ui})@{ub} Nonzero in a @{b}symbol_ref@{ub} if it refers to part of the current function's "constants pool". These are addresses close to the beginning of the function, and GNU CC assumes they can be addressed directly (perhaps with the help of base registers). Stored in the @{b}unchanging@{ub} field and printed as @{b}/u@{ub}. @{b}CONST_CALL_P (@{I}X@{ui})@{ub} In a @{b}call_insn@{ub}, indicates that the insn represents a call to a const function. Stored in the @{b}unchanging@{ub} field and printed as @{b}/u@{ub}. @{b}LABEL_PRESERVE_P (@{I}X@{ui})@{ub} In a @{b}code_label@{ub}, indicates that the label can never be deleted. Labels referenced by a non-local goto will have this bit set. Stored in the @{b}in_struct@{ub} field and printed as @{b}/s@{ub}. @{b}SCHED_GROUP_P (@{I}INSN@{ui})@{ub} During instruction scheduling, in an insn, indicates that the previous insn must be scheduled together with this insn. This is used to ensure that certain groups of instructions will not be split up by the instruction scheduling pass, for example, @{b}use@{ub} insns before a @{b}call_insn@{ub} may not be separated from the @{b}call_insn@{ub}. Stored in the @{b}in_struct@{ub} field and printed as @{b}/s@{ub}. These are the fields which the above macros refer to: @{b}used@{ub} Normally, this flag is used only momentarily, at the end of RTL generation for a function, to count the number of times an expression appears in insns. Expressions that appear more than once are copied, according to the rules for shared structure (see @{"Sharing" Link "Sharing"}). In a @{b}symbol_ref@{ub}, it indicates that an external declaration for the symbol has already been written. In a @{b}reg@{ub}, it is used by the leaf register renumbering code to ensure that each register is only renumbered once. @{b}volatil@{ub} This flag is used in @{b}mem@{ub}, @{b}symbol_ref@{ub} and @{b}reg@{ub} expressions and in insns. In RTL dump files, it is printed as @{b}/v@{ub}. In a @{b}mem@{ub} expression, it is 1 if the memory reference is volatile. Volatile memory references may not be deleted, reordered or combined. In a @{b}symbol_ref@{ub} expression, it is used for machine-specific purposes. In a @{b}reg@{ub} expression, it is 1 if the value is a user-level variable. 0 indicates an internal compiler temporary. In an insn, 1 means the insn has been deleted. @{b}in_struct@{ub} In @{b}mem@{ub} expressions, it is 1 if the memory datum referred to is all or part of a structure or array; 0 if it is (or might be) a scalar variable. A reference through a C pointer has 0 because the pointer might point to a scalar variable. This information allows the compiler to determine something about possible cases of aliasing. In an insn in the delay slot of a branch, 1 means that this insn is from the target of the branch. During instruction scheduling, in an insn, 1 means that this insn must be scheduled as part of a group together with the previous insn. In @{b}reg@{ub} expressions, it is 1 if the register has its entire life contained within the test expression of some loop. In @{b}subreg@{ub} expressions, 1 means that the @{b}subreg@{ub} is accessing an object that has had its mode promoted from a wider mode. In @{b}label_ref@{ub} expressions, 1 means that the referenced label is outside the innermost loop containing the insn in which the @{b}label_ref@{ub} was found. In @{b}code_label@{ub} expressions, it is 1 if the label may never be deleted. This is used for labels which are the target of non-local gotos. In an RTL dump, this flag is represented as @{b}/s@{ub}. @{b}unchanging@{ub} In @{b}reg@{ub} and @{b}mem@{ub} expressions, 1 means that the value of the expression never changes. In @{b}subreg@{ub} expressions, it is 1 if the @{b}subreg@{ub} references an unsigned object whose mode has been promoted to a wider mode. In an insn, 1 means that this is an annulling branch. In a @{b}symbol_ref@{ub} expression, 1 means that this symbol addresses something in the per-function constants pool. In a @{b}call_insn@{ub}, 1 means that this instruction is a call to a const function. In an RTL dump, this flag is represented as @{b}/u@{ub}. @{b}integrated@{ub} In some kinds of expressions, including insns, this flag means the rtl was produced by procedure integration. In a @{b}reg@{ub} expression, this flag indicates the register containing the value to be returned by the current function. On machines that pass parameters in registers, the same register number may be used for parameters as well, but this flag is not set on such uses. @EndNode @Node "Machine Modes" "gcc.guide/Machine Modes" @Next "Constants" @Prev "Flags" @Toc "RTL" Machine Modes ============= A machine mode describes a size of data object and the representation used for it. In the C code, machine modes are represented by an enumeration type, @{b}enum machine_mode@{ub}, defined in @{b}machmode.def@{ub}. Each RTL expression has room for a machine mode and so do certain kinds of tree expressions (declarations and types, to be precise). In debugging dumps and machine descriptions, the machine mode of an RTL expression is written after the expression code with a colon to separate them. The letters @{b}mode@{ub} which appear at the end of each machine mode name are omitted. For example, @{b}(reg:SI 38)@{ub} is a @{b}reg@{ub} expression with machine mode @{b}SImode@{ub}. If the mode is @{b}VOIDmode@{ub}, it is not written at all. Here is a table of machine modes. The term "byte" below refers to an object of @{b}BITS_PER_UNIT@{ub} bits (see @{"Storage Layout" Link "Storage Layout"}). @{b}QImode@{ub} "Quarter-Integer" mode represents a single byte treated as an integer. @{b}HImode@{ub} "Half-Integer" mode represents a two-byte integer. @{b}PSImode@{ub} "Partial Single Integer" mode represents an integer which occupies four bytes but which doesn't really use all four. On some machines, this is the right mode to use for pointers. @{b}SImode@{ub} "Single Integer" mode represents a four-byte integer. @{b}PDImode@{ub} "Partial Double Integer" mode represents an integer which occupies eight bytes but which doesn't really use all eight. On some machines, this is the right mode to use for certain pointers. @{b}DImode@{ub} "Double Integer" mode represents an eight-byte integer. @{b}TImode@{ub} "Tetra Integer" (?) mode represents a sixteen-byte integer. @{b}SFmode@{ub} "Single Floating" mode represents a single-precision (four byte) floating point number. @{b}DFmode@{ub} "Double Floating" mode represents a double-precision (eight byte) floating point number. @{b}XFmode@{ub} "Extended Floating" mode represents a triple-precision (twelve byte) floating point number. This mode is used for IEEE extended floating point. On some systems not all bits within these bytes will actually be used. @{b}TFmode@{ub} "Tetra Floating" mode represents a quadruple-precision (sixteen byte) floating point number. @{b}CCmode@{ub} "Condition Code" mode represents the value of a condition code, which is a machine-specific set of bits used to represent the result of a comparison operation. Other machine-specific modes may also be used for the condition code. These modes are not used on machines that use @{b}cc0@{ub} (see see @{"Condition Code" Link "Condition Code"}). @{b}BLKmode@{ub} "Block" mode represents values that are aggregates to which none of the other modes apply. In RTL, only memory references can have this mode, and only if they appear in string-move or vector instructions. On machines which have no such instructions, @{b}BLKmode@{ub} will not appear in RTL. @{b}VOIDmode@{ub} Void mode means the absence of a mode or an unspecified mode. For example, RTL expressions of code @{b}const_int@{ub} have mode @{b}VOIDmode@{ub} because they can be taken to have whatever mode the context requires. In debugging dumps of RTL, @{b}VOIDmode@{ub} is expressed by the absence of any mode. @{b}SCmode, DCmode, XCmode, TCmode@{ub} These modes stand for a complex number represented as a pair of floating point values. The floating point values are in @{b}SFmode@{ub}, @{b}DFmode@{ub}, @{b}XFmode@{ub}, and @{b}TFmode@{ub}, respectively. @{b}CQImode, CHImode, CSImode, CDImode, CTImode, COImode@{ub} These modes stand for a complex number represented as a pair of integer values. The integer values are in @{b}QImode@{ub}, @{b}HImode@{ub}, @{b}SImode@{ub}, @{b}DImode@{ub}, @{b}TImode@{ub}, and @{b}OImode@{ub}, respectively. The machine description defines @{b}Pmode@{ub} as a C macro which expands into the machine mode used for addresses. Normally this is the mode whose size is @{b}BITS_PER_WORD@{ub}, @{b}SImode@{ub} on 32-bit machines. The only modes which a machine description @{i}must@{ui} support are @{b}QImode@{ub}, and the modes corresponding to @{b}BITS_PER_WORD@{ub}, @{b}FLOAT_TYPE_SIZE@{ub} and @{b}DOUBLE_TYPE_SIZE@{ub}. The compiler will attempt to use @{b}DImode@{ub} for 8-byte structures and unions, but this can be prevented by overriding the definition of @{b}MAX_FIXED_MODE_SIZE@{ub}. Alternatively, you can have the compiler use @{b}TImode@{ub} for 16-byte structures and unions. Likewise, you can arrange for the C type @{b}short int@{ub} to avoid using @{b}HImode@{ub}. Very few explicit references to machine modes remain in the compiler and these few references will soon be removed. Instead, the machine modes are divided into mode classes. These are represented by the enumeration type @{b}enum mode_class@{ub} defined in @{b}machmode.h@{ub}. The possible mode classes are: @{b}MODE_INT@{ub} Integer modes. By default these are @{b}QImode@{ub}, @{b}HImode@{ub}, @{b}SImode@{ub}, @{b}DImode@{ub}, and @{b}TImode@{ub}. @{b}MODE_PARTIAL_INT@{ub} The "partial integer" modes, @{b}PSImode@{ub} and @{b}PDImode@{ub}. @{b}MODE_FLOAT@{ub} floating point modes. By default these are @{b}SFmode@{ub}, @{b}DFmode@{ub}, @{b}XFmode@{ub} and @{b}TFmode@{ub}. @{b}MODE_COMPLEX_INT@{ub} Complex integer modes. (These are not currently implemented). @{b}MODE_COMPLEX_FLOAT@{ub} Complex floating point modes. By default these are @{b}SCmode@{ub}, @{b}DCmode@{ub}, @{b}XCmode@{ub}, and @{b}TCmode@{ub}. @{b}MODE_FUNCTION@{ub} Algol or Pascal function variables including a static chain. (These are not currently implemented). @{b}MODE_CC@{ub} Modes representing condition code values. These are @{b}CCmode@{ub} plus any modes listed in the @{b}EXTRA_CC_MODES@{ub} macro. See @{"Jump Patterns" Link "Jump Patterns"}, also see @{"Condition Code" Link "Condition Code"}. @{b}MODE_RANDOM@{ub} This is a catchall mode class for modes which don't fit into the above classes. Currently @{b}VOIDmode@{ub} and @{b}BLKmode@{ub} are in @{b}MODE_RANDOM@{ub}. Here are some C macros that relate to machine modes: @{b}GET_MODE (@{I}X@{ui})@{ub} Returns the machine mode of the RTX @{I}X@{ui}. @{b}PUT_MODE (@{I}X@{ui}, @{I}NEWMODE@{ui})@{ub} Alters the machine mode of the RTX @{I}X@{ui} to be @{I}NEWMODE@{ui}. @{b}NUM_MACHINE_MODES@{ub} Stands for the number of machine modes available on the target machine. This is one greater than the largest numeric value of any machine mode. @{b}GET_MODE_NAME (@{I}M@{ui})@{ub} Returns the name of mode @{I}M@{ui} as a string. @{b}GET_MODE_CLASS (@{I}M@{ui})@{ub} Returns the mode class of mode @{I}M@{ui}. @{b}GET_MODE_WIDER_MODE (@{I}M@{ui})@{ub} Returns the next wider natural mode. For example, the expression @{b}GET_MODE_WIDER_MODE (QImode)@{ub} returns @{b}HImode@{ub}. @{b}GET_MODE_SIZE (@{I}M@{ui})@{ub} Returns the size in bytes of a datum of mode @{I}M@{ui}. @{b}GET_MODE_BITSIZE (@{I}M@{ui})@{ub} Returns the size in bits of a datum of mode @{I}M@{ui}. @{b}GET_MODE_MASK (@{I}M@{ui})@{ub} Returns a bitmask containing 1 for all bits in a word that fit within mode @{I}M@{ui}. This macro can only be used for modes whose bitsize is less than or equal to @{b}HOST_BITS_PER_INT@{ub}. @{b}GET_MODE_ALIGNMENT (@{I}M)@{ui})@{ub} Return the required alignment, in bits, for an object of mode @{I}M@{ui}. @{b}GET_MODE_UNIT_SIZE (@{I}M@{ui})@{ub} Returns the size in bytes of the subunits of a datum of mode @{I}M@{ui}. This is the same as @{b}GET_MODE_SIZE@{ub} except in the case of complex modes. For them, the unit size is the size of the real or imaginary part. @{b}GET_MODE_NUNITS (@{I}M@{ui})@{ub} Returns the number of units contained in a mode, i.e., @{b}GET_MODE_SIZE@{ub} divided by @{b}GET_MODE_UNIT_SIZE@{ub}. @{b}GET_CLASS_NARROWEST_MODE (@{I}C@{ui})@{ub} Returns the narrowest mode in mode class @{I}C@{ui}. The global variables @{b}byte_mode@{ub} and @{b}word_mode@{ub} contain modes whose classes are @{b}MODE_INT@{ub} and whose bitsizes are either @{b}BITS_PER_UNIT@{ub} or @{b}BITS_PER_WORD@{ub}, respectively. On 32-bit machines, these are @{b}QImode@{ub} and @{b}SImode@{ub}, respectively. @EndNode @Node "Constants" "gcc.guide/Constants" @Next "Regs and Memory" @Prev "Machine Modes" @Toc "RTL" Constant Expression Types ========================= The simplest RTL expressions are those that represent constant values. @{b}(const_int @{I}I@{ui})@{ub} This type of expression represents the integer value @{I}I@{ui}. @{I}I@{ui} is customarily accessed with the macro @{b}INTVAL@{ub} as in @{b}INTVAL (@{I}EXP@{ui})@{ub}, which is equivalent to @{b}XWINT (@{I}EXP@{ui}, 0)@{ub}. There is only one expression object for the integer value zero; it is the value of the variable @{b}const0_rtx@{ub}. Likewise, the only expression for integer value one is found in @{b}const1_rtx@{ub}, the only expression for integer value two is found in @{b}const2_rtx@{ub}, and the only expression for integer value negative one is found in @{b}constm1_rtx@{ub}. Any attempt to create an expression of code @{b}const_int@{ub} and value zero, one, two or negative one will return @{b}const0_rtx@{ub}, @{b}const1_rtx@{ub}, @{b}const2_rtx@{ub} or @{b}constm1_rtx@{ub} as appropriate. Similarly, there is only one object for the integer whose value is @{b}STORE_FLAG_VALUE@{ub}. It is found in @{b}const_true_rtx@{ub}. If @{b}STORE_FLAG_VALUE@{ub} is one, @{b}const_true_rtx@{ub} and @{b}const1_rtx@{ub} will point to the same object. If @{b}STORE_FLAG_VALUE@{ub} is -1, @{b}const_true_rtx@{ub} and @{b}constm1_rtx@{ub} will point to the same object. @{b}(const_double:@{I}M@{ui} @{I}ADDR@{ui} @{I}I0@{ui} @{I}I1@{ui} ...)@{ub} Represents either a floating-point constant of mode @{I}M@{ui} or an integer constant too large to fit into @{b}HOST_BITS_PER_WIDE_INT@{ub} bits but small enough to fit within twice that number of bits (GNU CC does not provide a mechanism to represent even larger constants). In the latter case, @{I}M@{ui} will be @{b}VOIDmode@{ub}. @{I}ADDR@{ui} is used to contain the @{b}mem@{ub} expression that corresponds to the location in memory that at which the constant can be found. If it has not been allocated a memory location, but is on the chain of all @{b}const_double@{ub} expressions in this compilation (maintained using an undisplayed field), @{I}ADDR@{ui} contains @{b}const0_rtx@{ub}. If it is not on the chain, @{I}ADDR@{ui} contains @{b}cc0_rtx@{ub}. @{I}ADDR@{ui} is customarily accessed with the macro @{b}CONST_DOUBLE_MEM@{ub} and the chain field via @{b}CONST_DOUBLE_CHAIN@{ub}. If @{I}M@{ui} is @{b}VOIDmode@{ub}, the bits of the value are stored in @{I}I0@{ui} and @{I}I1@{ui}. @{I}I0@{ui} is customarily accessed with the macro @{b}CONST_DOUBLE_LOW@{ub} and @{I}I1@{ui} with @{b}CONST_DOUBLE_HIGH@{ub}. If the constant is floating point (regardless of its precision), then the number of integers used to store the value depends on the size of @{b}REAL_VALUE_TYPE@{ub} (see @{"Cross-compilation" Link "Cross-compilation"}). The integers represent a floating point number, but not precisely in the target machine's or host machine's floating point format. To convert them to the precise bit pattern used by the target machine, use the macro @{b}REAL_VALUE_TO_TARGET_DOUBLE@{ub} and friends (see @{"Data Output" Link "Data Output"}). The macro @{b}CONST0_RTX (@{I}MODE@{ui})@{ub} refers to an expression with value 0 in mode @{I}MODE@{ui}. If mode @{I}MODE@{ui} is of mode class @{b}MODE_INT@{ub}, it returns @{b}const0_rtx@{ub}. Otherwise, it returns a @{b}CONST_DOUBLE@{ub} expression in mode @{I}MODE@{ui}. Similarly, the macro @{b}CONST1_RTX (@{I}MODE@{ui})@{ub} refers to an expression with value 1 in mode @{I}MODE@{ui} and similarly for @{b}CONST2_RTX@{ub}. @{b}(const_string @{I}STR@{ui})@{ub} Represents a constant string with value @{I}STR@{ui}. Currently this is used only for insn attributes (see @{"Insn Attributes" Link "Insn Attributes"}) since constant strings in C are placed in memory. @{b}(symbol_ref:@{I}MODE@{ui} @{I}SYMBOL@{ui})@{ub} Represents the value of an assembler label for data. @{I}SYMBOL@{ui} is a string that describes the name of the assembler label. If it starts with a @{b}*@{ub}, the label is the rest of @{I}SYMBOL@{ui} not including the @{b}*@{ub}. Otherwise, the label is @{I}SYMBOL@{ui}, usually prefixed with @{b}_@{ub}. The @{b}symbol_ref@{ub} contains a mode, which is usually @{b}Pmode@{ub}. Usually that is the only mode for which a symbol is directly valid. @{b}(label_ref @{I}LABEL@{ui})@{ub} Represents the value of an assembler label for code. It contains one operand, an expression, which must be a @{b}code_label@{ub} that appears in the instruction sequence to identify the place where the label should go. The reason for using a distinct expression type for code label references is so that jump optimization can distinguish them. @{b}(const:@{I}M@{ui} @{I}EXP@{ui})@{ub} Represents a constant that is the result of an assembly-time arithmetic computation. The operand, @{I}EXP@{ui}, is an expression that contains only constants (@{b}const_int@{ub}, @{b}symbol_ref@{ub} and @{b}label_ref@{ub} expressions) combined with @{b}plus@{ub} and @{b}minus@{ub}. However, not all combinations are valid, since the assembler cannot do arbitrary arithmetic on relocatable symbols. @{I}M@{ui} should be @{b}Pmode@{ub}. @{b}(high:@{I}M@{ui} @{I}EXP@{ui})@{ub} Represents the high-order bits of @{I}EXP@{ui}, usually a @{b}symbol_ref@{ub}. The number of bits is machine-dependent and is normally the number of bits specified in an instruction that initializes the high order bits of a register. It is used with @{b}lo_sum@{ub} to represent the typical two-instruction sequence used in RISC machines to reference a global memory location. @{I}M@{ui} should be @{b}Pmode@{ub}. @EndNode @Node "Regs and Memory" "gcc.guide/Regs and Memory" @Next "Arithmetic" @Prev "Constants" @Toc "RTL" Registers and Memory ==================== Here are the RTL expression types for describing access to machine registers and to main memory. @{b}(reg:@{I}M@{ui} @{I}N@{ui})@{ub} For small values of the integer @{I}N@{ui} (those that are less than @{b}FIRST_PSEUDO_REGISTER@{ub}), this stands for a reference to machine register number @{I}N@{ui}: a @{i}hard register@{ui}. For larger values of @{I}N@{ui}, it stands for a temporary value or @{i}pseudo register@{ui}. The compiler's strategy is to generate code assuming an unlimited number of such pseudo registers, and later convert them into hard registers or into memory references. @{I}M@{ui} is the machine mode of the reference. It is necessary because machines can generally refer to each register in more than one mode. For example, a register may contain a full word but there may be instructions to refer to it as a half word or as a single byte, as well as instructions to refer to it as a floating point number of various precisions. Even for a register that the machine can access in only one mode, the mode must always be specified. The symbol @{b}FIRST_PSEUDO_REGISTER@{ub} is defined by the machine description, since the number of hard registers on the machine is an invariant characteristic of the machine. Note, however, that not all of the machine registers must be general registers. All the machine registers that can be used for storage of data are given hard register numbers, even those that can be used only in certain instructions or can hold only certain types of data. A hard register may be accessed in various modes throughout one function, but each pseudo register is given a natural mode and is accessed only in that mode. When it is necessary to describe an access to a pseudo register using a nonnatural mode, a @{b}subreg@{ub} expression is used. A @{b}reg@{ub} expression with a machine mode that specifies more than one word of data may actually stand for several consecutive registers. If in addition the register number specifies a hardware register, then it actually represents several consecutive hardware registers starting with the specified one. Each pseudo register number used in a function's RTL code is represented by a unique @{b}reg@{ub} expression. Some pseudo register numbers, those within the range of @{b}FIRST_VIRTUAL_REGISTER@{ub} to @{b}LAST_VIRTUAL_REGISTER@{ub} only appear during the RTL generation phase and are eliminated before the optimization phases. These represent locations in the stack frame that cannot be determined until RTL generation for the function has been completed. The following virtual register numbers are defined: @{b}VIRTUAL_INCOMING_ARGS_REGNUM@{ub} This points to the first word of the incoming arguments passed on the stack. Normally these arguments are placed there by the caller, but the callee may have pushed some arguments that were previously passed in registers. When RTL generation is complete, this virtual register is replaced by the sum of the register given by @{b}ARG_POINTER_REGNUM@{ub} and the value of @{b}FIRST_PARM_OFFSET@{ub}. @{b}VIRTUAL_STACK_VARS_REGNUM@{ub} If @{b}FRAME_GROWS_DOWNWARD@{ub} is defined, this points to immediately above the first variable on the stack. Otherwise, it points to the first variable on the stack. @{b}VIRTUAL_STACK_VARS_REGNUM@{ub} is replaced with the sum of the register given by @{b}FRAME_POINTER_REGNUM@{ub} and the value @{b}STARTING_FRAME_OFFSET@{ub}. @{b}VIRTUAL_STACK_DYNAMIC_REGNUM@{ub} This points to the location of dynamically allocated memory on the stack immediately after the stack pointer has been adjusted by the amount of memory desired. This virtual register is replaced by the sum of the register given by @{b}STACK_POINTER_REGNUM@{ub} and the value @{b}STACK_DYNAMIC_OFFSET@{ub}. @{b}VIRTUAL_OUTGOING_ARGS_REGNUM@{ub} This points to the location in the stack at which outgoing arguments should be written when the stack is pre-pushed (arguments pushed using push insns should always use @{b}STACK_POINTER_REGNUM@{ub}). This virtual register is replaced by the sum of the register given by @{b}STACK_POINTER_REGNUM@{ub} and the value @{b}STACK_POINTER_OFFSET@{ub}. @{b}(subreg:@{I}M@{ui} @{I}REG@{ui} @{I}WORDNUM@{ui})@{ub} @{b}subreg@{ub} expressions are used to refer to a register in a machine mode other than its natural one, or to refer to one register of a multi-word @{b}reg@{ub} that actually refers to several registers. Each pseudo-register has a natural mode. If it is necessary to operate on it in a different mode--for example, to perform a fullword move instruction on a pseudo-register that contains a single byte--the pseudo-register must be enclosed in a @{b}subreg@{ub}. In such a case, @{I}WORDNUM@{ui} is zero. Usually @{I}M@{ui} is at least as narrow as the mode of @{I}REG@{ui}, in which case it is restricting consideration to only the bits of @{I}REG@{ui} that are in @{I}M@{ui}. Sometimes @{I}M@{ui} is wider than the mode of @{I}REG@{ui}. These @{b}subreg@{ub} expressions are often called @{i}paradoxical@{ui}. They are used in cases where we want to refer to an object in a wider mode but do not care what value the additional bits have. The reload pass ensures that paradoxical references are only made to hard registers. The other use of @{b}subreg@{ub} is to extract the individual registers of a multi-register value. Machine modes such as @{b}DImode@{ub} and @{b}TImode@{ub} can indicate values longer than a word, values which usually require two or more consecutive registers. To access one of the registers, use a @{b}subreg@{ub} with mode @{b}SImode@{ub} and a @{I}WORDNUM@{ui} that says which register. Storing in a non-paradoxical @{b}subreg@{ub} has undefined results for bits belonging to the same word as the @{b}subreg@{ub}. This laxity makes it easier to generate efficient code for such instructions. To represent an instruction that preserves all the bits outside of those in the @{b}subreg@{ub}, use @{b}strict_low_part@{ub} around the @{b}subreg@{ub}. The compilation parameter @{b}WORDS_BIG_ENDIAN@{ub}, if set to 1, says that word number zero is the most significant part; otherwise, it is the least significant part. Between the combiner pass and the reload pass, it is possible to have a paradoxical @{b}subreg@{ub} which contains a @{b}mem@{ub} instead of a @{b}reg@{ub} as its first operand. After the reload pass, it is also possible to have a non-paradoxical @{b}subreg@{ub} which contains a @{b}mem@{ub}; this usually occurs when the @{b}mem@{ub} is a stack slot which replaced a pseudo register. Note that it is not valid to access a @{b}DFmode@{ub} value in @{b}SFmode@{ub} using a @{b}subreg@{ub}. On some machines the most significant part of a @{b}DFmode@{ub} value does not have the same format as a single-precision floating value. It is also not valid to access a single word of a multi-word value in a hard register when less registers can hold the value than would be expected from its size. For example, some 32-bit machines have floating-point registers that can hold an entire @{b}DFmode@{ub} value. If register 10 were such a register @{b}(subreg:SI (reg:DF 10) 1)@{ub} would be invalid because there is no way to convert that reference to a single machine register. The reload pass prevents @{b}subreg@{ub} expressions such as these from being formed. The first operand of a @{b}subreg@{ub} expression is customarily accessed with the @{b}SUBREG_REG@{ub} macro and the second operand is customarily accessed with the @{b}SUBREG_WORD@{ub} macro. @{b}(scratch:@{I}M@{ui})@{ub} This represents a scratch register that will be required for the execution of a single instruction and not used subsequently. It is converted into a @{b}reg@{ub} by either the local register allocator or the reload pass. @{b}scratch@{ub} is usually present inside a @{b}clobber@{ub} operation (see @{"Side Effects" Link "Side Effects"}). @{b}(cc0)@{ub} This refers to the machine's condition code register. It has no operands and may not have a machine mode. There are two ways to use it: @{b}*@{ub} To stand for a complete set of condition code flags. This is best on most machines, where each comparison sets the entire series of flags. With this technique, @{b}(cc0)@{ub} may be validly used in only two contexts: as the destination of an assignment (in test and compare instructions) and in comparison operators comparing against zero (@{b}const_int@{ub} with value zero; that is to say, @{b}const0_rtx@{ub}). @{b}*@{ub} To stand for a single flag that is the result of a single condition. This is useful on machines that have only a single flag bit, and in which comparison instructions must specify the condition to test. With this technique, @{b}(cc0)@{ub} may be validly used in only two contexts: as the destination of an assignment (in test and compare instructions) where the source is a comparison operator, and as the first operand of @{b}if_then_else@{ub} (in a conditional branch). There is only one expression object of code @{b}cc0@{ub}; it is the value of the variable @{b}cc0_rtx@{ub}. Any attempt to create an expression of code @{b}cc0@{ub} will return @{b}cc0_rtx@{ub}. Instructions can set the condition code implicitly. On many machines, nearly all instructions set the condition code based on the value that they compute or store. It is not necessary to record these actions explicitly in the RTL because the machine description includes a prescription for recognizing the instructions that do so (by means of the macro @{b}NOTICE_UPDATE_CC@{ub}). See @{"Condition Code" Link "Condition Code"}. Only instructions whose sole purpose is to set the condition code, and instructions that use the condition code, need mention @{b}(cc0)@{ub}. On some machines, the condition code register is given a register number and a @{b}reg@{ub} is used instead of @{b}(cc0)@{ub}. This is usually the preferable approach if only a small subset of instructions modify the condition code. Other machines store condition codes in general registers; in such cases a pseudo register should be used. Some machines, such as the Sparc and RS/6000, have two sets of arithmetic instructions, one that sets and one that does not set the condition code. This is best handled by normally generating the instruction that does not set the condition code, and making a pattern that both performs the arithmetic and sets the condition code register (which would not be @{b}(cc0)@{ub} in this case). For examples, search for @{b}addcc@{ub} and @{b}andcc@{ub} in @{b}sparc.md@{ub}. @{b}(pc)@{ub} This represents the machine's program counter. It has no operands and may not have a machine mode. @{b}(pc)@{ub} may be validly used only in certain specific contexts in jump instructions. There is only one expression object of code @{b}pc@{ub}; it is the value of the variable @{b}pc_rtx@{ub}. Any attempt to create an expression of code @{b}pc@{ub} will return @{b}pc_rtx@{ub}. All instructions that do not jump alter the program counter implicitly by incrementing it, but there is no need to mention this in the RTL. @{b}(mem:@{I}M@{ui} @{I}ADDR@{ui})@{ub} This RTX represents a reference to main memory at an address represented by the expression @{I}ADDR@{ui}. @{I}M@{ui} specifies how large a unit of memory is accessed. @EndNode @Node "Arithmetic" "gcc.guide/Arithmetic" @Next "Comparisons" @Prev "Regs and Memory" @Toc "RTL" RTL Expressions for Arithmetic ============================== Unless otherwise specified, all the operands of arithmetic expressions must be valid for mode @{I}M@{ui}. An operand is valid for mode @{I}M@{ui} if it has mode @{I}M@{ui}, or if it is a @{b}const_int@{ub} or @{b}const_double@{ub} and @{I}M@{ui} is a mode of class @{b}MODE_INT@{ub}. For commutative binary operations, constants should be placed in the second operand. @{b}(plus:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Represents the sum of the values represented by @{I}X@{ui} and @{I}Y@{ui} carried out in machine mode @{I}M@{ui}. @{b}(lo_sum:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Like @{b}plus@{ub}, except that it represents that sum of @{I}X@{ui} and the low-order bits of @{I}Y@{ui}. The number of low order bits is machine-dependent but is normally the number of bits in a @{b}Pmode@{ub} item minus the number of bits set by the @{b}high@{ub} code (see @{"Constants" Link "Constants"}). @{I}M@{ui} should be @{b}Pmode@{ub}. @{b}(minus:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Like @{b}plus@{ub} but represents subtraction. @{b}(compare:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Represents the result of subtracting @{I}Y@{ui} from @{I}X@{ui} for purposes of comparison. The result is computed without overflow, as if with infinite precision. Of course, machines can't really subtract with infinite precision. However, they can pretend to do so when only the sign of the result will be used, which is the case when the result is stored in the condition code. And that is the only way this kind of expression may validly be used: as a value to be stored in the condition codes. The mode @{I}M@{ui} is not related to the modes of @{I}X@{ui} and @{I}Y@{ui}, but instead is the mode of the condition code value. If @{b}(cc0)@{ub} is used, it is @{b}VOIDmode@{ub}. Otherwise it is some mode in class @{b}MODE_CC@{ub}, often @{b}CCmode@{ub}. See @{"Condition Code" Link "Condition Code"}. Normally, @{I}X@{ui} and @{I}Y@{ui} must have the same mode. Otherwise, @{b}compare@{ub} is valid only if the mode of @{I}X@{ui} is in class @{b}MODE_INT@{ub} and @{I}Y@{ui} is a @{b}const_int@{ub} or @{b}const_double@{ub} with mode @{b}VOIDmode@{ub}. The mode of @{I}X@{ui} determines what mode the comparison is to be done in; thus it must not be @{b}VOIDmode@{ub}. If one of the operands is a constant, it should be placed in the second operand and the comparison code adjusted as appropriate. A @{b}compare@{ub} specifying two @{b}VOIDmode@{ub} constants is not valid since there is no way to know in what mode the comparison is to be performed; the comparison must either be folded during the compilation or the first operand must be loaded into a register while its mode is still known. @{b}(neg:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the negation (subtraction from zero) of the value represented by @{I}X@{ui}, carried out in mode @{I}M@{ui}. @{b}(mult:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Represents the signed product of the values represented by @{I}X@{ui} and @{I}Y@{ui} carried out in machine mode @{I}M@{ui}. Some machines support a multiplication that generates a product wider than the operands. Write the pattern for this as (mult:@{I}M@{ui} (sign_extend:@{I}M@{ui} @{I}X@{ui}) (sign_extend:@{I}M@{ui} @{I}Y@{ui})) where @{I}M@{ui} is wider than the modes of @{I}X@{ui} and @{I}Y@{ui}, which need not be the same. Write patterns for unsigned widening multiplication similarly using @{b}zero_extend@{ub}. @{b}(div:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Represents the quotient in signed division of @{I}X@{ui} by @{I}Y@{ui}, carried out in machine mode @{I}M@{ui}. If @{I}M@{ui} is a floating point mode, it represents the exact quotient; otherwise, the integerized quotient. Some machines have division instructions in which the operands and quotient widths are not all the same; you should represent such instructions using @{b}truncate@{ub} and @{b}sign_extend@{ub} as in, (truncate:@{I}M1@{ui} (div:@{I}M2@{ui} @{I}X@{ui} (sign_extend:@{I}M2@{ui} @{I}Y@{ui}))) @{b}(udiv:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Like @{b}div@{ub} but represents unsigned division. @{b}(mod:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} @{b}(umod:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Like @{b}div@{ub} and @{b}udiv@{ub} but represent the remainder instead of the quotient. @{b}(smin:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} @{b}(smax:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Represents the smaller (for @{b}smin@{ub}) or larger (for @{b}smax@{ub}) of @{I}X@{ui} and @{I}Y@{ui}, interpreted as signed integers in mode @{I}M@{ui}. @{b}(umin:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} @{b}(umax:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Like @{b}smin@{ub} and @{b}smax@{ub}, but the values are interpreted as unsigned integers. @{b}(not:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the bitwise complement of the value represented by @{I}X@{ui}, carried out in mode @{I}M@{ui}, which must be a fixed-point machine mode. @{b}(and:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Represents the bitwise logical-and of the values represented by @{I}X@{ui} and @{I}Y@{ui}, carried out in machine mode @{I}M@{ui}, which must be a fixed-point machine mode. @{b}(ior:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Represents the bitwise inclusive-or of the values represented by @{I}X@{ui} and @{I}Y@{ui}, carried out in machine mode @{I}M@{ui}, which must be a fixed-point mode. @{b}(xor:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Represents the bitwise exclusive-or of the values represented by @{I}X@{ui} and @{I}Y@{ui}, carried out in machine mode @{I}M@{ui}, which must be a fixed-point mode. @{b}(ashift:@{I}M@{ui} @{I}X@{ui} @{I}C@{ui})@{ub} Represents the result of arithmetically shifting @{I}X@{ui} left by @{I}C@{ui} places. @{I}X@{ui} have mode @{I}M@{ui}, a fixed-point machine mode. @{I}C@{ui} be a fixed-point mode or be a constant with mode @{b}VOIDmode@{ub}; which mode is determined by the mode called for in the machine description entry for the left-shift instruction. For example, on the Vax, the mode of @{I}C@{ui} is @{b}QImode@{ub} regardless of @{I}M@{ui}. @{b}(lshiftrt:@{I}M@{ui} @{I}X@{ui} @{I}C@{ui})@{ub} @{b}(ashiftrt:@{I}M@{ui} @{I}X@{ui} @{I}C@{ui})@{ub} Like @{b}ashift@{ub} but for right shift. Unlike the case for left shift, these two operations are distinct. @{b}(rotate:@{I}M@{ui} @{I}X@{ui} @{I}C@{ui})@{ub} @{b}(rotatert:@{I}M@{ui} @{I}X@{ui} @{I}C@{ui})@{ub} Similar but represent left and right rotate. If @{I}C@{ui} is a constant, use @{b}rotate@{ub}. @{b}(abs:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the absolute value of @{I}X@{ui}, computed in mode @{I}M@{ui}. @{b}(sqrt:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the square root of @{I}X@{ui}, computed in mode @{I}M@{ui}. Most often @{I}M@{ui} will be a floating point mode. @{b}(ffs:@{I}M@{ui} @{I}X@{ui})@{ub} Represents one plus the index of the least significant 1-bit in @{I}X@{ui}, represented as an integer of mode @{I}M@{ui}. (The value is zero if @{I}X@{ui} is zero.) The mode of @{I}X@{ui} need not be @{I}M@{ui}; depending on the target machine, various mode combinations may be valid. @EndNode @Node "Comparisons" "gcc.guide/Comparisons" @Next "Bit Fields" @Prev "Arithmetic" @Toc "RTL" Comparison Operations ===================== Comparison operators test a relation on two operands and are considered to represent a machine-dependent nonzero value described by, but not necessarily equal to, @{b}STORE_FLAG_VALUE@{ub} (see @{"Misc" Link "Misc"}) if the relation holds, or zero if it does not. The mode of the comparison operation is independent of the mode of the data being compared. If the comparison operation is being tested (e.g., the first operand of an @{b}if_then_else@{ub}), the mode must be @{b}VOIDmode@{ub}. If the comparison operation is producing data to be stored in some variable, the mode must be in class @{b}MODE_INT@{ub}. All comparison operations producing data must use the same mode, which is machine-specific. There are two ways that comparison operations may be used. The comparison operators may be used to compare the condition codes @{b}(cc0)@{ub} against zero, as in @{b}(eq (cc0) (const_int 0))@{ub}. Such a construct actually refers to the result of the preceding instruction in which the condition codes were set. The instructing setting the condition code must be adjacent to the instruction using the condition code; only @{b}note@{ub} insns may separate them. Alternatively, a comparison operation may directly compare two data objects. The mode of the comparison is determined by the operands; they must both be valid for a common machine mode. A comparison with both operands constant would be invalid as the machine mode could not be deduced from it, but such a comparison should never exist in RTL due to constant folding. In the example above, if @{b}(cc0)@{ub} were last set to @{b}(compare @{I}X@{ui} @{I}Y@{ui})@{ub}, the comparison operation is identical to @{b}(eq @{I}X@{ui} @{I}Y@{ui})@{ub}. Usually only one style of comparisons is supported on a particular machine, but the combine pass will try to merge the operations to produce the @{b}eq@{ub} shown in case it exists in the context of the particular insn involved. Inequality comparisons come in two flavors, signed and unsigned. Thus, there are distinct expression codes @{b}gt@{ub} and @{b}gtu@{ub} for signed and unsigned greater-than. These can produce different results for the same pair of integer values: for example, 1 is signed greater-than -1 but not unsigned greater-than, because -1 when regarded as unsigned is actually @{b}0xffffffff@{ub} which is greater than 1. The signed comparisons are also used for floating point values. Floating point comparisons are distinguished by the machine modes of the operands. @{b}(eq:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} 1 if the values represented by @{I}X@{ui} and @{I}Y@{ui} are equal, otherwise 0. @{b}(ne:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} 1 if the values represented by @{I}X@{ui} and @{I}Y@{ui} are not equal, otherwise 0. @{b}(gt:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} 1 if the @{I}X@{ui} is greater than @{I}Y@{ui}. If they are fixed-point, the comparison is done in a signed sense. @{b}(gtu:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Like @{b}gt@{ub} but does unsigned comparison, on fixed-point numbers only. @{b}(lt:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} @{b}(ltu:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Like @{b}gt@{ub} and @{b}gtu@{ub} but test for "less than". @{b}(ge:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} @{b}(geu:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Like @{b}gt@{ub} and @{b}gtu@{ub} but test for "greater than or equal". @{b}(le:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} @{b}(leu:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} Like @{b}gt@{ub} and @{b}gtu@{ub} but test for "less than or equal". @{b}(if_then_else @{I}COND@{ui} @{I}THEN@{ui} @{I}ELSE@{ui})@{ub} This is not a comparison operation but is listed here because it is always used in conjunction with a comparison operation. To be precise, @{I}COND@{ui} is a comparison expression. This expression represents a choice, according to @{I}COND@{ui}, between the value represented by @{I}THEN@{ui} and the one represented by @{I}ELSE@{ui}. On most machines, @{b}if_then_else@{ub} expressions are valid only to express conditional jumps. @{b}(cond [@{I}TEST1@{ui} @{I}VALUE1@{ui} @{I}TEST2@{ui} @{I}VALUE2@{ui} ...] @{I}DEFAULT@{ui})@{ub} Similar to @{b}if_then_else@{ub}, but more general. Each of @{I}TEST1@{ui}, @{I}TEST2@{ui}, ... is performed in turn. The result of this expression is the @{I}VALUE@{ui} corresponding to the first non-zero test, or @{I}DEFAULT@{ui} if none of the tests are non-zero expressions. This is currently not valid for instruction patterns and is supported only for insn attributes. See @{"Insn Attributes" Link "Insn Attributes"}. @EndNode @Node "Bit Fields" "gcc.guide/Bit Fields" @Next "Conversions" @Prev "Comparisons" @Toc "RTL" Bit Fields ========== Special expression codes exist to represent bitfield instructions. These types of expressions are lvalues in RTL; they may appear on the left side of an assignment, indicating insertion of a value into the specified bit field. @{b}(sign_extract:@{I}M@{ui} @{I}LOC@{ui} @{I}SIZE@{ui} @{I}POS@{ui})@{ub} This represents a reference to a sign-extended bit field contained or starting in @{I}LOC@{ui} (a memory or register reference). The bit field is @{I}SIZE@{ui} bits wide and starts at bit @{I}POS@{ui}. The compilation option @{b}BITS_BIG_ENDIAN@{ub} says which end of the memory unit @{I}POS@{ui} counts from. If @{I}LOC@{ui} is in memory, its mode must be a single-byte integer mode. If @{I}LOC@{ui} is in a register, the mode to use is specified by the operand of the @{b}insv@{ub} or @{b}extv@{ub} pattern (see @{"Standard Names" Link "Standard Names"}) and is usually a full-word integer mode. The mode of @{I}POS@{ui} is machine-specific and is also specified in the @{b}insv@{ub} or @{b}extv@{ub} pattern. The mode @{I}M@{ui} is the same as the mode that would be used for @{I}LOC@{ui} if it were a register. @{b}(zero_extract:@{I}M@{ui} @{I}LOC@{ui} @{I}SIZE@{ui} @{I}POS@{ui})@{ub} Like @{b}sign_extract@{ub} but refers to an unsigned or zero-extended bit field. The same sequence of bits are extracted, but they are filled to an entire word with zeros instead of by sign-extension. @EndNode @Node "Conversions" "gcc.guide/Conversions" @Next "RTL Declarations" @Prev "Bit Fields" @Toc "RTL" Conversions =========== All conversions between machine modes must be represented by explicit conversion operations. For example, an expression which is the sum of a byte and a full word cannot be written as @{b}(plus:SI (reg:QI 34) (reg:SI 80))@{ub} because the @{b}plus@{ub} operation requires two operands of the same machine mode. Therefore, the byte-sized operand is enclosed in a conversion operation, as in (plus:SI (sign_extend:SI (reg:QI 34)) (reg:SI 80)) The conversion operation is not a mere placeholder, because there may be more than one way of converting from a given starting mode to the desired final mode. The conversion operation code says how to do it. For all conversion operations, @{I}X@{ui} must not be @{b}VOIDmode@{ub} because the mode in which to do the conversion would not be known. The conversion must either be done at compile-time or @{I}X@{ui} must be placed into a register. @{b}(sign_extend:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the result of sign-extending the value @{I}X@{ui} to machine mode @{I}M@{ui}. @{I}M@{ui} must be a fixed-point mode and @{I}X@{ui} a fixed-point value of a mode narrower than @{I}M@{ui}. @{b}(zero_extend:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the result of zero-extending the value @{I}X@{ui} to machine mode @{I}M@{ui}. @{I}M@{ui} must be a fixed-point mode and @{I}X@{ui} a fixed-point value of a mode narrower than @{I}M@{ui}. @{b}(float_extend:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the result of extending the value @{I}X@{ui} to machine mode @{I}M@{ui}. @{I}M@{ui} must be a floating point mode and @{I}X@{ui} a floating point value of a mode narrower than @{I}M@{ui}. @{b}(truncate:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the result of truncating the value @{I}X@{ui} to machine mode @{I}M@{ui}. @{I}M@{ui} must be a fixed-point mode and @{I}X@{ui} a fixed-point value of a mode wider than @{I}M@{ui}. @{b}(float_truncate:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the result of truncating the value @{I}X@{ui} to machine mode @{I}M@{ui}. @{I}M@{ui} must be a floating point mode and @{I}X@{ui} a floating point value of a mode wider than @{I}M@{ui}. @{b}(float:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the result of converting fixed point value @{I}X@{ui}, regarded as signed, to floating point mode @{I}M@{ui}. @{b}(unsigned_float:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the result of converting fixed point value @{I}X@{ui}, regarded as unsigned, to floating point mode @{I}M@{ui}. @{b}(fix:@{I}M@{ui} @{I}X@{ui})@{ub} When @{I}M@{ui} is a fixed point mode, represents the result of converting floating point value @{I}X@{ui} to mode @{I}M@{ui}, regarded as signed. How rounding is done is not specified, so this operation may be used validly in compiling C code only for integer-valued operands. @{b}(unsigned_fix:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the result of converting floating point value @{I}X@{ui} to fixed point mode @{I}M@{ui}, regarded as unsigned. How rounding is done is not specified. @{b}(fix:@{I}M@{ui} @{I}X@{ui})@{ub} When @{I}M@{ui} is a floating point mode, represents the result of converting floating point value @{I}X@{ui} (valid for mode @{I}M@{ui}) to an integer, still represented in floating point mode @{I}M@{ui}, by rounding towards zero. @EndNode @Node "RTL Declarations" "gcc.guide/RTL Declarations" @Next "Side Effects" @Prev "Conversions" @Toc "RTL" Declarations ============ Declaration expression codes do not represent arithmetic operations but rather state assertions about their operands. @{b}(strict_low_part (subreg:@{I}M@{ui} (reg:@{I}N@{ui} @{I}R@{ui}) 0))@{ub} This expression code is used in only one context: as the destination operand of a @{b}set@{ub} expression. In addition, the operand of this expression must be a non-paradoxical @{b}subreg@{ub} expression. The presence of @{b}strict_low_part@{ub} says that the part of the register which is meaningful in mode @{I}N@{ui}, but is not part of mode @{I}M@{ui}, is not to be altered. Normally, an assignment to such a subreg is allowed to have undefined effects on the rest of the register when @{I}M@{ui} is less than a word. @EndNode @Node "Side Effects" "gcc.guide/Side Effects" @Next "Incdec" @Prev "RTL Declarations" @Toc "RTL" Side Effect Expressions ======================= The expression codes described so far represent values, not actions. But machine instructions never produce values; they are meaningful only for their side effects on the state of the machine. Special expression codes are used to represent side effects. The body of an instruction is always one of these side effect codes; the codes described above, which represent values, appear only as the operands of these. @{b}(set @{I}LVAL@{ui} @{I}X@{ui})@{ub} Represents the action of storing the value of @{I}X@{ui} into the place represented by @{I}LVAL@{ui}. @{I}LVAL@{ui} must be an expression representing a place that can be stored in: @{b}reg@{ub} (or @{b}subreg@{ub} or @{b}strict_low_part@{ub}), @{b}mem@{ub}, @{b}pc@{ub} or @{b}cc0@{ub}. If @{I}LVAL@{ui} is a @{b}reg@{ub}, @{b}subreg@{ub} or @{b}mem@{ub}, it has a machine mode; then @{I}X@{ui} must be valid for that mode. If @{I}LVAL@{ui} is a @{b}reg@{ub} whose machine mode is less than the full width of the register, then it means that the part of the register specified by the machine mode is given the specified value and the rest of the register receives an undefined value. Likewise, if @{I}LVAL@{ui} is a @{b}subreg@{ub} whose machine mode is narrower than the mode of the register, the rest of the register can be changed in an undefined way. If @{I}LVAL@{ui} is a @{b}strict_low_part@{ub} of a @{b}subreg@{ub}, then the part of the register specified by the machine mode of the @{b}subreg@{ub} is given the value @{I}X@{ui} and the rest of the register is not changed. If @{I}LVAL@{ui} is @{b}(cc0)@{ub}, it has no machine mode, and @{I}X@{ui} may be either a @{b}compare@{ub} expression or a value that may have any mode. The latter case represents a "test" instruction. The expression @{b}(set (cc0) (reg:@{I}M@{ui} @{I}N@{ui}))@{ub} is equivalent to @{b}(set (cc0) (compare (reg:@{I}M@{ui} @{I}N@{ui}) (const_int 0)))@{ub}. Use the former expression to save space during the compilation. If @{I}LVAL@{ui} is @{b}(pc)@{ub}, we have a jump instruction, and the possibilities for @{I}X@{ui} are very limited. It may be a @{b}label_ref@{ub} expression (unconditional jump). It may be an @{b}if_then_else@{ub} (conditional jump), in which case either the second or the third operand must be @{b}(pc)@{ub} (for the case which does not jump) and the other of the two must be a @{b}label_ref@{ub} (for the case which does jump). @{I}X@{ui} may also be a @{b}mem@{ub} or @{b}(plus:SI (pc) @{I}Y@{ui})@{ub}, where @{I}Y@{ui} may be a @{b}reg@{ub} or a @{b}mem@{ub}; these unusual patterns are used to represent jumps through branch tables. If @{I}LVAL@{ui} is neither @{b}(cc0)@{ub} nor @{b}(pc)@{ub}, the mode of @{I}LVAL@{ui} must not be @{b}VOIDmode@{ub} and the mode of @{I}X@{ui} must be valid for the mode of @{I}LVAL@{ui}. @{I}LVAL@{ui} is customarily accessed with the @{b}SET_DEST@{ub} macro and @{I}X@{ui} with the @{b}SET_SRC@{ub} macro. @{b}(return)@{ub} As the sole expression in a pattern, represents a return from the current function, on machines where this can be done with one instruction, such as Vaxes. On machines where a multi-instruction "epilogue" must be executed in order to return from the function, returning is done by jumping to a label which precedes the epilogue, and the @{b}return@{ub} expression code is never used. Inside an @{b}if_then_else@{ub} expression, represents the value to be placed in @{b}pc@{ub} to return to the caller. Note that an insn pattern of @{b}(return)@{ub} is logically equivalent to @{b}(set (pc) (return))@{ub}, but the latter form is never used. @{b}(call @{I}FUNCTION@{ui} @{I}NARGS@{ui})@{ub} Represents a function call. @{I}FUNCTION@{ui} is a @{b}mem@{ub} expression whose address is the address of the function to be called. @{I}NARGS@{ui} is an expression which can be used for two purposes: on some machines it represents the number of bytes of stack argument; on others, it represents the number of argument registers. Each machine has a standard machine mode which @{I}FUNCTION@{ui} must have. The machine description defines macro @{b}FUNCTION_MODE@{ub} to expand into the requisite mode name. The purpose of this mode is to specify what kind of addressing is allowed, on machines where the allowed kinds of addressing depend on the machine mode being addressed. @{b}(clobber @{I}X@{ui})@{ub} Represents the storing or possible storing of an unpredictable, undescribed value into @{I}X@{ui}, which must be a @{b}reg@{ub}, @{b}scratch@{ub} or @{b}mem@{ub} expression. One place this is used is in string instructions that store standard values into particular hard registers. It may not be worth the trouble to describe the values that are stored, but it is essential to inform the compiler that the registers will be altered, lest it attempt to keep data in them across the string instruction. If @{I}X@{ui} is @{b}(mem:BLK (const_int 0))@{ub}, it means that all memory locations must be presumed clobbered. Note that the machine description classifies certain hard registers as "call-clobbered". All function call instructions are assumed by default to clobber these registers, so there is no need to use @{b}clobber@{ub} expressions to indicate this fact. Also, each function call is assumed to have the potential to alter any memory location, unless the function is declared @{b}const@{ub}. If the last group of expressions in a @{b}parallel@{ub} are each a @{b}clobber@{ub} expression whose arguments are @{b}reg@{ub} or @{b}match_scratch@{ub} (see @{"RTL Template" Link "RTL Template"}) expressions, the combiner phase can add the appropriate @{b}clobber@{ub} expressions to an insn it has constructed when doing so will cause a pattern to be matched. This feature can be used, for example, on a machine that whose multiply and add instructions don't use an MQ register but which has an add-accumulate instruction that does clobber the MQ register. Similarly, a combined instruction might require a temporary register while the constituent instructions might not. When a @{b}clobber@{ub} expression for a register appears inside a @{b}parallel@{ub} with other side effects, the register allocator guarantees that the register is unoccupied both before and after that insn. However, the reload phase may allocate a register used for one of the inputs unless the @{b}&@{ub} constraint is specified for the selected alternative (see @{"Modifiers" Link "Modifiers"}). You can clobber either a specific hard register, a pseudo register, or a @{b}scratch@{ub} expression; in the latter two cases, GNU CC will allocate a hard register that is available there for use as a temporary. For instructions that require a temporary register, you should use @{b}scratch@{ub} instead of a pseudo-register because this will allow the combiner phase to add the @{b}clobber@{ub} when required. You do this by coding (@{b}clobber@{ub} (@{b}match_scratch@{ub} ...)). If you do clobber a pseudo register, use one which appears nowhere else--generate a new one each time. Otherwise, you may confuse CSE. There is one other known use for clobbering a pseudo register in a @{b}parallel@{ub}: when one of the input operands of the insn is also clobbered by the insn. In this case, using the same pseudo register in the clobber and elsewhere in the insn produces the expected results. @{b}(use @{I}X@{ui})@{ub} Represents the use of the value of @{I}X@{ui}. It indicates that the value in @{I}X@{ui} at this point in the program is needed, even though it may not be apparent why this is so. Therefore, the compiler will not attempt to delete previous instructions whose only effect is to store a value in @{I}X@{ui}. @{I}X@{ui} must be a @{b}reg@{ub} expression. During the delayed branch scheduling phase, @{I}X@{ui} may be an insn. This indicates that @{I}X@{ui} previously was located at this place in the code and its data dependencies need to be taken into account. These @{b}use@{ub} insns will be deleted before the delayed branch scheduling phase exits. @{b}(parallel [@{I}X0@{ui} @{I}X1@{ui} ...])@{ub} Represents several side effects performed in parallel. The square brackets stand for a vector; the operand of @{b}parallel@{ub} is a vector of expressions. @{I}X0@{ui}, @{I}X1@{ui} and so on are individual side effect expressions--expressions of code @{b}set@{ub}, @{b}call@{ub}, @{b}return@{ub}, @{b}clobber@{ub} or @{b}use@{ub}. "In parallel" means that first all the values used in the individual side-effects are computed, and second all the actual side-effects are performed. For example, (parallel [(set (reg:SI 1) (mem:SI (reg:SI 1))) (set (mem:SI (reg:SI 1)) (reg:SI 1))]) says unambiguously that the values of hard register 1 and the memory location addressed by it are interchanged. In both places where @{b}(reg:SI 1)@{ub} appears as a memory address it refers to the value in register 1 @{i}before@{ui} the execution of the insn. It follows that it is @{i}incorrect@{ui} to use @{b}parallel@{ub} and expect the result of one @{b}set@{ub} to be available for the next one. For example, people sometimes attempt to represent a jump-if-zero instruction this way: (parallel [(set (cc0) (reg:SI 34)) (set (pc) (if_then_else (eq (cc0) (const_int 0)) (label_ref ...) (pc)))]) But this is incorrect, because it says that the jump condition depends on the condition code value @{i}before@{ui} this instruction, not on the new value that is set by this instruction. Peephole optimization, which takes place together with final assembly code output, can produce insns whose patterns consist of a @{b}parallel@{ub} whose elements are the operands needed to output the resulting assembler code--often @{b}reg@{ub}, @{b}mem@{ub} or constant expressions. This would not be well-formed RTL at any other stage in compilation, but it is ok then because no further optimization remains to be done. However, the definition of the macro @{b}NOTICE_UPDATE_CC@{ub}, if any, must deal with such insns if you define any peephole optimizations. @{b}(sequence [@{I}INSNS@{ui} ...])@{ub} Represents a sequence of insns. Each of the @{I}INSNS@{ui} that appears in the vector is suitable for appearing in the chain of insns, so it must be an @{b}insn@{ub}, @{b}jump_insn@{ub}, @{b}call_insn@{ub}, @{b}code_label@{ub}, @{b}barrier@{ub} or @{b}note@{ub}. A @{b}sequence@{ub} RTX is never placed in an actual insn during RTL generation. It represents the sequence of insns that result from a @{b}define_expand@{ub} @{i}before@{ui} those insns are passed to @{b}emit_insn@{ub} to insert them in the chain of insns. When actually inserted, the individual sub-insns are separated out and the @{b}sequence@{ub} is forgotten. After delay-slot scheduling is completed, an insn and all the insns that reside in its delay slots are grouped together into a @{b}sequence@{ub}. The insn requiring the delay slot is the first insn in the vector; subsequent insns are to be placed in the delay slot. @{b}INSN_ANNULLED_BRANCH_P@{ub} is set on an insn in a delay slot to indicate that a branch insn should be used that will conditionally annul the effect of the insns in the delay slots. In such a case, @{b}INSN_FROM_TARGET_P@{ub} indicates that the insn is from the target of the branch and should be executed only if the branch is taken; otherwise the insn should be executed only if the branch is not taken. See @{"Delay Slots" Link "Delay Slots"}. These expression codes appear in place of a side effect, as the body of an insn, though strictly speaking they do not always describe side effects as such: @{b}(asm_input @{I}S@{ui})@{ub} Represents literal assembler code as described by the string @{I}S@{ui}. @{b}(unspec [@{I}OPERANDS@{ui} ...] @{I}INDEX@{ui})@{ub} @{b}(unspec_volatile [@{I}OPERANDS@{ui} ...] @{I}INDEX@{ui})@{ub} Represents a machine-specific operation on @{I}OPERANDS@{ui}. @{I}INDEX@{ui} selects between multiple machine-specific operations. @{b}unspec_volatile@{ub} is used for volatile operations and operations that may trap; @{b}unspec@{ub} is used for other operations. These codes may appear inside a @{b}pattern@{ub} of an insn, inside a @{b}parallel@{ub}, or inside an expression. @{b}(addr_vec:@{I}M@{ui} [@{I}LR0@{ui} @{I}LR1@{ui} ...])@{ub} Represents a table of jump addresses. The vector elements @{I}LR0@{ui}, etc., are @{b}label_ref@{ub} expressions. The mode @{I}M@{ui} specifies how much space is given to each address; normally @{I}M@{ui} would be @{b}Pmode@{ub}. @{b}(addr_diff_vec:@{I}M@{ui} @{I}BASE@{ui} [@{I}LR0@{ui} @{I}LR1@{ui} ...])@{ub} Represents a table of jump addresses expressed as offsets from @{I}BASE@{ui}. The vector elements @{I}LR0@{ui}, etc., are @{b}label_ref@{ub} expressions and so is @{I}BASE@{ui}. The mode @{I}M@{ui} specifies how much space is given to each address-difference. @EndNode @Node "Incdec" "gcc.guide/Incdec" @Next "Assembler" @Prev "Side Effects" @Toc "RTL" Embedded Side-Effects on Addresses ================================== Four special side-effect expression codes appear as memory addresses. @{b}(pre_dec:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the side effect of decrementing @{I}X@{ui} by a standard amount and represents also the value that @{I}X@{ui} has after being decremented. @{I}X@{ui} must be a @{b}reg@{ub} or @{b}mem@{ub}, but most machines allow only a @{b}reg@{ub}. @{I}M@{ui} must be the machine mode for pointers on the machine in use. The amount @{I}X@{ui} is decremented by is the length in bytes of the machine mode of the containing memory reference of which this expression serves as the address. Here is an example of its use: (mem:DF (pre_dec:SI (reg:SI 39))) This says to decrement pseudo register 39 by the length of a @{b}DFmode@{ub} value and use the result to address a @{b}DFmode@{ub} value. @{b}(pre_inc:@{I}M@{ui} @{I}X@{ui})@{ub} Similar, but specifies incrementing @{I}X@{ui} instead of decrementing it. @{b}(post_dec:@{I}M@{ui} @{I}X@{ui})@{ub} Represents the same side effect as @{b}pre_dec@{ub} but a different value. The value represented here is the value @{I}X@{ui} has @{i}before@{ui} being decremented. @{b}(post_inc:@{I}M@{ui} @{I}X@{ui})@{ub} Similar, but specifies incrementing @{I}X@{ui} instead of decrementing it. These embedded side effect expressions must be used with care. Instruction patterns may not use them. Until the @{b}flow@{ub} pass of the compiler, they may occur only to represent pushes onto the stack. The @{b}flow@{ub} pass finds cases where registers are incremented or decremented in one instruction and used as an address shortly before or after; these cases are then transformed to use pre- or post-increment or -decrement. If a register used as the operand of these expressions is used in another address in an insn, the original value of the register is used. Uses of the register outside of an address are not permitted within the same insn as a use in an embedded side effect expression because such insns behave differently on different machines and hence must be treated as ambiguous and disallowed. An instruction that can be represented with an embedded side effect could also be represented using @{b}parallel@{ub} containing an additional @{b}set@{ub} to describe how the address register is altered. This is not done because machines that allow these operations at all typically allow them wherever a memory address is called for. Describing them as additional parallel stores would require doubling the number of entries in the machine description. @EndNode @Node "Assembler" "gcc.guide/Assembler" @Next "Insns" @Prev "Incdec" @Toc "RTL" Assembler Instructions as Expressions ===================================== The RTX code @{b}asm_operands@{ub} represents a value produced by a user-specified assembler instruction. It is used to represent an @{b}asm@{ub} statement with arguments. An @{b}asm@{ub} statement with a single output operand, like this: asm ("foo %1,%2,%0" : "=a" (outputvar) : "g" (x + y), "di" (*z)); is represented using a single @{b}asm_operands@{ub} RTX which represents the value that is stored in @{b}outputvar@{ub}: (set @{I}RTX-FOR-OUTPUTVAR@{ui} (asm_operands "foo %1,%2,%0" "a" 0 [@{I}RTX-FOR-ADDITION-RESULT@{ui} @{I}RTX-FOR-*Z@{ui}] [(asm_input:@{I}M1@{ui} "g") (asm_input:@{I}M2@{ui} "di")])) Here the operands of the @{b}asm_operands@{ub} RTX are the assembler template string, the output-operand's constraint, the index-number of the output operand among the output operands specified, a vector of input operand RTX's, and a vector of input-operand modes and constraints. The mode @{I}M1@{ui} is the mode of the sum @{b}x+y@{ub}; @{I}M2@{ui} is that of @{b}*z@{ub}. When an @{b}asm@{ub} statement has multiple output values, its insn has several such @{b}set@{ub} RTX's inside of a @{b}parallel@{ub}. Each @{b}set@{ub} contains a @{b}asm_operands@{ub}; all of these share the same assembler template and vectors, but each contains the constraint for the respective output operand. They are also distinguished by the output-operand index number, which is 0, 1, ... for successive output operands. @EndNode @Node "Insns" "gcc.guide/Insns" @Next "Calls" @Prev "Assembler" @Toc "RTL" Insns ===== The RTL representation of the code for a function is a doubly-linked chain of objects called @{i}insns@{ui}. Insns are expressions with special codes that are used for no other purpose. Some insns are actual instructions; others represent dispatch tables for @{b}switch@{ub} statements; others represent labels to jump to or various sorts of declarative information. In addition to its own specific data, each insn must have a unique id-number that distinguishes it from all other insns in the current function (after delayed branch scheduling, copies of an insn with the same id-number may be present in multiple places in a function, but these copies will always be identical and will only appear inside a @{b}sequence@{ub}), and chain pointers to the preceding and following insns. These three fields occupy the same position in every insn, independent of the expression code of the insn. They could be accessed with @{b}XEXP@{ub} and @{b}XINT@{ub}, but instead three special macros are always used: @{b}INSN_UID (@{I}I@{ui})@{ub} Accesses the unique id of insn @{I}I@{ui}. @{b}PREV_INSN (@{I}I@{ui})@{ub} Accesses the chain pointer to the insn preceding @{I}I@{ui}. If @{I}I@{ui} is the first insn, this is a null pointer. @{b}NEXT_INSN (@{I}I@{ui})@{ub} Accesses the chain pointer to the insn following @{I}I@{ui}. If @{I}I@{ui} is the last insn, this is a null pointer. The first insn in the chain is obtained by calling @{b}get_insns@{ub}; the last insn is the result of calling @{b}get_last_insn@{ub}. Within the chain delimited by these insns, the @{b}NEXT_INSN@{ub} and @{b}PREV_INSN@{ub} pointers must always correspond: if @{I}INSN@{ui} is not the first insn, NEXT_INSN (PREV_INSN (@{I}INSN@{ui})) == @{I}INSN@{ui} is always true and if @{I}INSN@{ui} is not the last insn, PREV_INSN (NEXT_INSN (@{I}INSN@{ui})) == @{I}INSN@{ui} is always true. After delay slot scheduling, some of the insns in the chain might be @{b}sequence@{ub} expressions, which contain a vector of insns. The value of @{b}NEXT_INSN@{ub} in all but the last of these insns is the next insn in the vector; the value of @{b}NEXT_INSN@{ub} of the last insn in the vector is the same as the value of @{b}NEXT_INSN@{ub} for the @{b}sequence@{ub} in which it is contained. Similar rules apply for @{b}PREV_INSN@{ub}. This means that the above invariants are not necessarily true for insns inside @{b}sequence@{ub} expressions. Specifically, if @{I}INSN@{ui} is the first insn in a @{b}sequence@{ub}, @{b}NEXT_INSN (PREV_INSN (@{I}INSN@{ui}))@{ub} is the insn containing the @{b}sequence@{ub} expression, as is the value of @{b}PREV_INSN (NEXT_INSN (@{I}INSN@{ui}))@{ub} is @{I}INSN@{ui} is the last insn in the @{b}sequence@{ub} expression. You can use these expressions to find the containing @{b}sequence@{ub} expression. Every insn has one of the following six expression codes: @{b}insn@{ub} The expression code @{b}insn@{ub} is used for instructions that do not jump and do not do function calls. @{b}sequence@{ub} expressions are always contained in insns with code @{b}insn@{ub} even if one of those insns should jump or do function calls. Insns with code @{b}insn@{ub} have four additional fields beyond the three mandatory ones listed above. These four are described in a table below. @{b}jump_insn@{ub} The expression code @{b}jump_insn@{ub} is used for instructions that may jump (or, more generally, may contain @{b}label_ref@{ub} expressions). If there is an instruction to return from the current function, it is recorded as a @{b}jump_insn@{ub}. @{b}jump_insn@{ub} insns have the same extra fields as @{b}insn@{ub} insns, accessed in the same way and in addition contain a field @{b}JUMP_LABEL@{ub} which is defined once jump optimization has completed. For simple conditional and unconditional jumps, this field contains the @{b}code_label@{ub} to which this insn will (possibly conditionally) branch. In a more complex jump, @{b}JUMP_LABEL@{ub} records one of the labels that the insn refers to; the only way to find the others is to scan the entire body of the insn. Return insns count as jumps, but since they do not refer to any labels, they have zero in the @{b}JUMP_LABEL@{ub} field. @{b}call_insn@{ub} The expression code @{b}call_insn@{ub} is used for instructions that may do function calls. It is important to distinguish these instructions because they imply that certain registers and memory locations may be altered unpredictably. @{b}call_insn@{ub} insns have the same extra fields as @{b}insn@{ub} insns, accessed in the same way and in addition contain a field @{b}CALL_INSN_FUNCTION_USAGE@{ub}, which contains a list (chain of @{b}expr_list@{ub} expressions) containing @{b}use@{ub} and @{b}clobber@{ub} expressions that denote hard registers used or clobbered by the called function. A register specified in a @{b}clobber@{ub} in this list is modified @{i}after@{ui} the execution of the @{b}call_insn@{ub}, while a register in a @{b}clobber@{ub} in the body of the @{b}call_insn@{ub} is clobbered before the insn completes execution. @{b}clobber@{ub} expressions in this list augment registers specified in @{b}CALL_USED_REGISTERS@{ub} (see @{"Register Basics" Link "Register Basics"}). @{b}code_label@{ub} A @{b}code_label@{ub} insn represents a label that a jump insn can jump to. It contains two special fields of data in addition to the three standard ones. @{b}CODE_LABEL_NUMBER@{ub} is used to hold the @{i}label number@{ui}, a number that identifies this label uniquely among all the labels in the compilation (not just in the current function). Ultimately, the label is represented in the assembler output as an assembler label, usually of the form @{b} L@{I}N@{ui}@{ub} where @{I}N@{ui} is the label number. When a @{b}code_label@{ub} appears in an RTL expression, it normally appears within a @{b}label_ref@{ub} which represents the address of the label, as a number. The field @{b}LABEL_NUSES@{ub} is only defined once the jump optimization phase is completed and contains the number of times this label is referenced in the current function. @{b}barrier@{ub} Barriers are placed in the instruction stream when control cannot flow past them. They are placed after unconditional jump instructions to indicate that the jumps are unconditional and after calls to @{b}volatile@{ub} functions, which do not return (e.g., @{b}exit@{ub}). They contain no information beyond the three standard fields. @{b}note@{ub} @{b}note@{ub} insns are used to represent additional debugging and declarative information. They contain two nonstandard fields, an integer which is accessed with the macro @{b}NOTE_LINE_NUMBER@{ub} and a string accessed with @{b}NOTE_SOURCE_FILE@{ub}. If @{b}NOTE_LINE_NUMBER@{ub} is positive, the note represents the position of a source line and @{b}NOTE_SOURCE_FILE@{ub} is the source file name that the line came from. These notes control generation of line number data in the assembler output. Otherwise, @{b}NOTE_LINE_NUMBER@{ub} is not really a line number but a code with one of the following values (and @{b}NOTE_SOURCE_FILE@{ub} must contain a null pointer): @{b}NOTE_INSN_DELETED@{ub} Such a note is completely ignorable. Some passes of the compiler delete insns by altering them into notes of this kind. @{b}NOTE_INSN_BLOCK_BEG@{ub} @{b}NOTE_INSN_BLOCK_END@{ub} These types of notes indicate the position of the beginning and end of a level of scoping of variable names. They control the output of debugging information. @{b}NOTE_INSN_LOOP_BEG@{ub} @{b}NOTE_INSN_LOOP_END@{ub} These types of notes indicate the position of the beginning and end of a @{b}while@{ub} or @{b}for@{ub} loop. They enable the loop optimizer to find loops quickly. @{b}NOTE_INSN_LOOP_CONT@{ub} Appears at the place in a loop that @{b}continue@{ub} statements jump to. @{b}NOTE_INSN_LOOP_VTOP@{ub} This note indicates the place in a loop where the exit test begins for those loops in which the exit test has been duplicated. This position becomes another virtual start of the loop when considering loop invariants. @{b}NOTE_INSN_FUNCTION_END@{ub} Appears near the end of the function body, just before the label that @{b}return@{ub} statements jump to (on machine where a single instruction does not suffice for returning). This note may be deleted by jump optimization. @{b}NOTE_INSN_SETJMP@{ub} Appears following each call to @{b}setjmp@{ub} or a related function. These codes are printed symbolically when they appear in debugging dumps. The machine mode of an insn is normally @{b}VOIDmode@{ub}, but some phases use the mode for various purposes; for example, the reload pass sets it to @{b}HImode@{ub} if the insn needs reloading but not register elimination and @{b}QImode@{ub} if both are required. The common subexpression elimination pass sets the mode of an insn to @{b}QImode@{ub} when it is the first insn in a block that has already been processed. Here is a table of the extra fields of @{b}insn@{ub}, @{b}jump_insn@{ub} and @{b}call_insn@{ub} insns: @{b}PATTERN (@{I}I@{ui})@{ub} An expression for the side effect performed by this insn. This must be one of the following codes: @{b}set@{ub}, @{b}call@{ub}, @{b}use@{ub}, @{b}clobber@{ub}, @{b}return@{ub}, @{b}asm_input@{ub}, @{b}asm_output@{ub}, @{b}addr_vec@{ub}, @{b}addr_diff_vec@{ub}, @{b}trap_if@{ub}, @{b}unspec@{ub}, @{b}unspec_volatile@{ub}, @{b}parallel@{ub}, or @{b}sequence@{ub}. If it is a @{b}parallel@{ub}, each element of the @{b}parallel@{ub} must be one these codes, except that @{b}parallel@{ub} expressions cannot be nested and @{b}addr_vec@{ub} and @{b}addr_diff_vec@{ub} are not permitted inside a @{b}parallel@{ub} expression. @{b}INSN_CODE (@{I}I@{ui})@{ub} An integer that says which pattern in the machine description matches this insn, or -1 if the matching has not yet been attempted. Such matching is never attempted and this field remains -1 on an insn whose pattern consists of a single @{b}use@{ub}, @{b}clobber@{ub}, @{b}asm_input@{ub}, @{b}addr_vec@{ub} or @{b}addr_diff_vec@{ub} expression. Matching is also never attempted on insns that result from an @{b}asm@{ub} statement. These contain at least one @{b}asm_operands@{ub} expression. The function @{b}asm_noperands@{ub} returns a non-negative value for such insns. In the debugging output, this field is printed as a number followed by a symbolic representation that locates the pattern in the @{b}md@{ub} file as some small positive or negative offset from a named pattern. @{b}LOG_LINKS (@{I}I@{ui})@{ub} A list (chain of @{b}insn_list@{ub} expressions) giving information about dependencies between instructions within a basic block. Neither a jump nor a label may come between the related insns. @{b}REG_NOTES (@{I}I@{ui})@{ub} A list (chain of @{b}expr_list@{ub} and @{b}insn_list@{ub} expressions) giving miscellaneous information about the insn. It is often information pertaining to the registers used in this insn. The @{b}LOG_LINKS@{ub} field of an insn is a chain of @{b}insn_list@{ub} expressions. Each of these has two operands: the first is an insn, and the second is another @{b}insn_list@{ub} expression (the next one in the chain). The last @{b}insn_list@{ub} in the chain has a null pointer as second operand. The significant thing about the chain is which insns appear in it (as first operands of @{b}insn_list@{ub} expressions). Their order is not significant. This list is originally set up by the flow analysis pass; it is a null pointer until then. Flow only adds links for those data dependencies which can be used for instruction combination. For each insn, the flow analysis pass adds a link to insns which store into registers values that are used for the first time in this insn. The instruction scheduling pass adds extra links so that every dependence will be represented. Links represent data dependencies, antidependencies and output dependencies; the machine mode of the link distinguishes these three types: antidependencies have mode @{b}REG_DEP_ANTI@{ub}, output dependencies have mode @{b}REG_DEP_OUTPUT@{ub}, and data dependencies have mode @{b}VOIDmode@{ub}. The @{b}REG_NOTES@{ub} field of an insn is a chain similar to the @{b}LOG_LINKS@{ub} field but it includes @{b}expr_list@{ub} expressions in addition to @{b}insn_list@{ub} expressions. There are several kinds of register notes, which are distinguished by the machine mode, which in a register note is really understood as being an @{b}enum reg_note@{ub}. The first operand @{I}OP@{ui} of the note is data whose meaning depends on the kind of note. The macro @{b}REG_NOTE_KIND (@{I}X@{ui})@{ub} returns the kind of register note. Its counterpart, the macro @{b}PUT_REG_NOTE_KIND (@{I}X@{ui}, @{I}NEWKIND@{ui})@{ub} sets the register note type of @{I}X@{ui} to be @{I}NEWKIND@{ui}. Register notes are of three classes: They may say something about an input to an insn, they may say something about an output of an insn, or they may create a linkage between two insns. There are also a set of values that are only used in @{b}LOG_LINKS@{ub}. These register notes annotate inputs to an insn: @{b}REG_DEAD@{ub} The value in @{I}OP@{ui} dies in this insn; that is to say, altering the value immediately after this insn would not affect the future behavior of the program. This does not necessarily mean that the register @{I}OP@{ui} has no useful value after this insn since it may also be an output of the insn. In such a case, however, a @{b}REG_DEAD@{ub} note would be redundant and is usually not present until after the reload pass, but no code relies on this fact. @{b}REG_INC@{ub} The register @{I}OP@{ui} is incremented (or decremented; at this level there is no distinction) by an embedded side effect inside this insn. This means it appears in a @{b}post_inc@{ub}, @{b}pre_inc@{ub}, @{b}post_dec@{ub} or @{b}pre_dec@{ub} expression. @{b}REG_NONNEG@{ub} The register @{I}OP@{ui} is known to have a nonnegative value when this insn is reached. This is used so that decrement and branch until zero instructions, such as the m68k dbra, can be matched. The @{b}REG_NONNEG@{ub} note is added to insns only if the machine description has a @{b}decrement_and_branch_until_zero@{ub} pattern. @{b}REG_NO_CONFLICT@{ub} This insn does not cause a conflict between @{I}OP@{ui} and the item being set by this insn even though it might appear that it does. In other words, if the destination register and @{I}OP@{ui} could otherwise be assigned the same register, this insn does not prevent that assignment. Insns with this note are usually part of a block that begins with a @{b}clobber@{ub} insn specifying a multi-word pseudo register (which will be the output of the block), a group of insns that each set one word of the value and have the @{b}REG_NO_CONFLICT@{ub} note attached, and a final insn that copies the output to itself with an attached @{b}REG_EQUAL@{ub} note giving the expression being computed. This block is encapsulated with @{b}REG_LIBCALL@{ub} and @{b}REG_RETVAL@{ub} notes on the first and last insns, respectively. @{b}REG_LABEL@{ub} This insn uses @{I}OP@{ui}, a @{b}code_label@{ub}, but is not a @{b}jump_insn@{ub}. The presence of this note allows jump optimization to be aware that @{I}OP@{ui} is, in fact, being used. The following notes describe attributes of outputs of an insn: @{b}REG_EQUIV@{ub} @{b}REG_EQUAL@{ub} This note is only valid on an insn that sets only one register and indicates that that register will be equal to @{I}OP@{ui} at run time; the scope of this equivalence differs between the two types of notes. The value which the insn explicitly copies into the register may look different from @{I}OP@{ui}, but they will be equal at run time. If the output of the single @{b}set@{ub} is a @{b}strict_low_part@{ub} expression, the note refers to the register that is contained in @{b}SUBREG_REG@{ub} of the @{b}subreg@{ub} expression. For @{b}REG_EQUIV@{ub}, the register is equivalent to @{I}OP@{ui} throughout the entire function, and could validly be replaced in all its occurrences by @{I}OP@{ui}. ("Validly" here refers to the data flow of the program; simple replacement may make some insns invalid.) For example, when a constant is loaded into a register that is never assigned any other value, this kind of note is used. When a parameter is copied into a pseudo-register at entry to a function, a note of this kind records that the register is equivalent to the stack slot where the parameter was passed. Although in this case the register may be set by other insns, it is still valid to replace the register by the stack slot throughout the function. In the case of @{b}REG_EQUAL@{ub}, the register that is set by this insn will be equal to @{I}OP@{ui} at run time at the end of this insn but not necessarily elsewhere in the function. In this case, @{I}OP@{ui} is typically an arithmetic expression. For example, when a sequence of insns such as a library call is used to perform an arithmetic operation, this kind of note is attached to the insn that produces or copies the final value. These two notes are used in different ways by the compiler passes. @{b}REG_EQUAL@{ub} is used by passes prior to register allocation (such as common subexpression elimination and loop optimization) to tell them how to think of that value. @{b}REG_EQUIV@{ub} notes are used by register allocation to indicate that there is an available substitute expression (either a constant or a @{b}mem@{ub} expression for the location of a parameter on the stack) that may be used in place of a register if insufficient registers are available. Except for stack homes for parameters, which are indicated by a @{b}REG_EQUIV@{ub} note and are not useful to the early optimization passes and pseudo registers that are equivalent to a memory location throughout there entire life, which is not detected until later in the compilation, all equivalences are initially indicated by an attached @{b}REG_EQUAL@{ub} note. In the early stages of register allocation, a @{b}REG_EQUAL@{ub} note is changed into a @{b}REG_EQUIV@{ub} note if @{I}OP@{ui} is a constant and the insn represents the only set of its destination register. Thus, compiler passes prior to register allocation need only check for @{b}REG_EQUAL@{ub} notes and passes subsequent to register allocation need only check for @{b}REG_EQUIV@{ub} notes. @{b}REG_UNUSED@{ub} The register @{I}OP@{ui} being set by this insn will not be used in a subsequent insn. This differs from a @{b}REG_DEAD@{ub} note, which indicates that the value in an input will not be used subsequently. These two notes are independent; both may be present for the same register. @{b}REG_WAS_0@{ub} The single output of this insn contained zero before this insn. @{I}OP@{ui} is the insn that set it to zero. You can rely on this note if it is present and @{I}OP@{ui} has not been deleted or turned into a @{b}note@{ub}; its absence implies nothing. These notes describe linkages between insns. They occur in pairs: one insn has one of a pair of notes that points to a second insn, which has the inverse note pointing back to the first insn. @{b}REG_RETVAL@{ub} This insn copies the value of a multi-insn sequence (for example, a library call), and @{I}OP@{ui} is the first insn of the sequence (for a library call, the first insn that was generated to set up the arguments for the library call). Loop optimization uses this note to treat such a sequence as a single operation for code motion purposes and flow analysis uses this note to delete such sequences whose results are dead. A @{b}REG_EQUAL@{ub} note will also usually be attached to this insn to provide the expression being computed by the sequence. @{b}REG_LIBCALL@{ub} This is the inverse of @{b}REG_RETVAL@{ub}: it is placed on the first insn of a multi-insn sequence, and it points to the last one. @{b}REG_CC_SETTER@{ub} @{b}REG_CC_USER@{ub} On machines that use @{b}cc0@{ub}, the insns which set and use @{b}cc0@{ub} set and use @{b}cc0@{ub} are adjacent. However, when branch delay slot filling is done, this may no longer be true. In this case a @{b}REG_CC_USER@{ub} note will be placed on the insn setting @{b}cc0@{ub} to point to the insn using @{b}cc0@{ub} and a @{b}REG_CC_SETTER@{ub} note will be placed on the insn using @{b}cc0@{ub} to point to the insn setting @{b}cc0@{ub}. These values are only used in the @{b}LOG_LINKS@{ub} field, and indicate the type of dependency that each link represents. Links which indicate a data dependence (a read after write dependence) do not use any code, they simply have mode @{b}VOIDmode@{ub}, and are printed without any descriptive text. @{b}REG_DEP_ANTI@{ub} This indicates an anti dependence (a write after read dependence). @{b}REG_DEP_OUTPUT@{ub} This indicates an output dependence (a write after write dependence). For convenience, the machine mode in an @{b}insn_list@{ub} or @{b}expr_list@{ub} is printed using these symbolic codes in debugging dumps. The only difference between the expression codes @{b}insn_list@{ub} and @{b}expr_list@{ub} is that the first operand of an @{b}insn_list@{ub} is assumed to be an insn and is printed in debugging dumps as the insn's unique id; the first operand of an @{b}expr_list@{ub} is printed in the ordinary way as an expression. @EndNode @Node "Calls" "gcc.guide/Calls" @Next "Sharing" @Prev "Insns" @Toc "RTL" RTL Representation of Function-Call Insns ========================================= Insns that call subroutines have the RTL expression code @{b}call_insn@{ub}. These insns must satisfy special rules, and their bodies must use a special RTL expression code, @{b}call@{ub}. A @{b}call@{ub} expression has two operands, as follows: (call (mem:@{I}FM@{ui} @{I}ADDR@{ui}) @{I}NBYTES@{ui}) Here @{I}NBYTES@{ui} is an operand that represents the number of bytes of argument data being passed to the subroutine, @{I}FM@{ui} is a machine mode (which must equal as the definition of the @{b}FUNCTION_MODE@{ub} macro in the machine description) and @{I}ADDR@{ui} represents the address of the subroutine. For a subroutine that returns no value, the @{b}call@{ub} expression as shown above is the entire body of the insn, except that the insn might also contain @{b}use@{ub} or @{b}clobber@{ub} expressions. For a subroutine that returns a value whose mode is not @{b}BLKmode@{ub}, the value is returned in a hard register. If this register's number is @{I}R@{ui}, then the body of the call insn looks like this: (set (reg:@{I}M@{ui} @{I}R@{ui}) (call (mem:@{I}FM@{ui} @{I}ADDR@{ui}) @{I}NBYTES@{ui})) This RTL expression makes it clear (to the optimizer passes) that the appropriate register receives a useful value in this insn. When a subroutine returns a @{b}BLKmode@{ub} value, it is handled by passing to the subroutine the address of a place to store the value. So the call insn itself does not "return" any value, and it has the same RTL form as a call that returns nothing. On some machines, the call instruction itself clobbers some register, for example to contain the return address. @{b}call_insn@{ub} insns on these machines should have a body which is a @{b}parallel@{ub} that contains both the @{b}call@{ub} expression and @{b}clobber@{ub} expressions that indicate which registers are destroyed. Similarly, if the call instruction requires some register other than the stack pointer that is not explicitly mentioned it its RTL, a @{b}use@{ub} subexpression should mention that register. Functions that are called are assumed to modify all registers listed in the configuration macro @{b}CALL_USED_REGISTERS@{ub} (see @{"Register Basics" Link "Register Basics"}) and, with the exception of @{b}const@{ub} functions and library calls, to modify all of memory. Insns containing just @{b}use@{ub} expressions directly precede the @{b}call_insn@{ub} insn to indicate which registers contain inputs to the function. Similarly, if registers other than those in @{b}CALL_USED_REGISTERS@{ub} are clobbered by the called function, insns containing a single @{b}clobber@{ub} follow immediately after the call to indicate which registers. @EndNode @Node "Sharing" "gcc.guide/Sharing" @Next "Reading RTL" @Prev "Calls" @Toc "RTL" Structure Sharing Assumptions ============================= The compiler assumes that certain kinds of RTL expressions are unique; there do not exist two distinct objects representing the same value. In other cases, it makes an opposite assumption: that no RTL expression object of a certain kind appears in more than one place in the containing structure. These assumptions refer to a single function; except for the RTL objects that describe global variables and external functions, and a few standard objects such as small integer constants, no RTL objects are common to two functions. @{b}*@{ub} Each pseudo-register has only a single @{b}reg@{ub} object to represent it, and therefore only a single machine mode. @{b}*@{ub} For any symbolic label, there is only one @{b}symbol_ref@{ub} object referring to it. @{b}*@{ub} There is only one @{b}const_int@{ub} expression with value 0, only one with value 1, and only one with value -1. Some other integer values are also stored uniquely. @{b}*@{ub} There is only one @{b}pc@{ub} expression. @{b}*@{ub} There is only one @{b}cc0@{ub} expression. @{b}*@{ub} There is only one @{b}const_double@{ub} expression with value 0 for each floating point mode. Likewise for values 1 and 2. @{b}*@{ub} No @{b}label_ref@{ub} or @{b}scratch@{ub} appears in more than one place in the RTL structure; in other words, it is safe to do a tree-walk of all the insns in the function and assume that each time a @{b}label_ref@{ub} or @{b}scratch@{ub} is seen it is distinct from all others that are seen. @{b}*@{ub} Only one @{b}mem@{ub} object is normally created for each static variable or stack slot, so these objects are frequently shared in all the places they appear. However, separate but equal objects for these variables are occasionally made. @{b}*@{ub} When a single @{b}asm@{ub} statement has multiple output operands, a distinct @{b}asm_operands@{ub} expression is made for each output operand. However, these all share the vector which contains the sequence of input operands. This sharing is used later on to test whether two @{b}asm_operands@{ub} expressions come from the same statement, so all optimizations must carefully preserve the sharing if they copy the vector at all. @{b}*@{ub} No RTL object appears in more than one place in the RTL structure except as described above. Many passes of the compiler rely on this by assuming that they can modify RTL objects in place without unwanted side-effects on other insns. @{b}*@{ub} During initial RTL generation, shared structure is freely introduced. After all the RTL for a function has been generated, all shared structure is copied by @{b}unshare_all_rtl@{ub} in @{b}emit-rtl.c@{ub}, after which the above rules are guaranteed to be followed. @{b}*@{ub} During the combiner pass, shared structure within an insn can exist temporarily. However, the shared structure is copied before the combiner is finished with the insn. This is done by calling @{b}copy_rtx_if_shared@{ub}, which is a subroutine of @{b}unshare_all_rtl@{ub}. @EndNode @Node "Reading RTL" "gcc.guide/Reading RTL" @Prev "Sharing" @Toc "RTL" Reading RTL =========== To read an RTL object from a file, call @{b}read_rtx@{ub}. It takes one argument, a stdio stream, and returns a single RTL object. Reading RTL from a file is very slow. This is not currently a problem since reading RTL occurs only as part of building the compiler. People frequently have the idea of using RTL stored as text in a file as an interface between a language front end and the bulk of GNU CC. This idea is not feasible. GNU CC was designed to use RTL internally only. Correct RTL for a given program is very dependent on the particular target machine. And the RTL does not contain all the information about the program. The proper way to interface GNU CC to a new language front end is with the "tree" data structure. There is no manual for this data structure, but it is described in the files @{b}tree.h@{ub} and @{b}tree.def@{ub}. @EndNode @Node "Machine Desc" "gcc.guide/Machine Desc" @Next "Target Macros" @Prev "RTL" @Toc "Main" Machine Descriptions ******************** A machine description has two parts: a file of instruction patterns (@{b}.md@{ub} file) and a C header file of macro definitions. The @{b}.md@{ub} file for a target machine contains a pattern for each instruction that the target machine supports (or at least each instruction that is worth telling the compiler about). It may also contain comments. A semicolon causes the rest of the line to be a comment, unless the semicolon is inside a quoted string. See the next chapter for information on the C header file. @{" Patterns " Link "Patterns"} How to write instruction patterns. @{" Example " Link "Example"} An explained example of a @{b}define_insn@{ub} pattern. @{" RTL Template " Link "RTL Template"} The RTL template defines what insns match a pattern. @{" Output Template " Link "Output Template"} The output template says how to make assembler code from such an insn. @{" Output Statement " Link "Output Statement"} For more generality, write C code to output the assembler code. @{" Constraints " Link "Constraints"} When not all operands are general operands. @{" Standard Names " Link "Standard Names"} Names mark patterns to use for code generation. @{" Pattern Ordering " Link "Pattern Ordering"} When the order of patterns makes a difference. @{" Dependent Patterns " Link "Dependent Patterns"} Having one pattern may make you need another. @{" Jump Patterns " Link "Jump Patterns"} Special considerations for patterns for jump insns. @{" Insn Canonicalizations " Link "Insn Canonicalizations"} Canonicalization of Instructions @{" Peephole Definitions " Link "Peephole Definitions"} Defining machine-specific peephole optimizations. @{" Expander Definitions " Link "Expander Definitions"} Generating a sequence of several RTL insns for a standard operation. @{" Insn Splitting " Link "Insn Splitting"} Splitting Instructions into Multiple Instructions @{" Insn Attributes " Link "Insn Attributes"} Specifying the value of attributes for generated insns. @EndNode @Node "Patterns" "gcc.guide/Patterns" @Next "Example" @Toc "Machine Desc" Everything about Instruction Patterns ===================================== Each instruction pattern contains an incomplete RTL expression, with pieces to be filled in later, operand constraints that restrict how the pieces can be filled in, and an output pattern or C code to generate the assembler output, all wrapped up in a @{b}define_insn@{ub} expression. A @{b}define_insn@{ub} is an RTL expression containing four or five operands: 1. An optional name. The presence of a name indicate that this instruction pattern can perform a certain standard job for the RTL-generation pass of the compiler. This pass knows certain names and will use the instruction patterns with those names, if the names are defined in the machine description. The absence of a name is indicated by writing an empty string where the name should go. Nameless instruction patterns are never used for generating RTL code, but they may permit several simpler insns to be combined later on. Names that are not thus known and used in RTL-generation have no effect; they are equivalent to no name at all. 2. The @{i}RTL template@{ui} (see @{"RTL Template" Link "RTL Template"}) is a vector of incomplete RTL expressions which show what the instruction should look like. It is incomplete because it may contain @{b}match_operand@{ub}, @{b}match_operator@{ub}, and @{b}match_dup@{ub} expressions that stand for operands of the instruction. If the vector has only one element, that element is the template for the instruction pattern. If the vector has multiple elements, then the instruction pattern is a @{b}parallel@{ub} expression containing the elements described. 3. A condition. This is a string which contains a C expression that is the final test to decide whether an insn body matches this pattern. For a named pattern, the condition (if present) may not depend on the data in the insn being matched, but only the target-machine-type flags. The compiler needs to test these conditions during initialization in order to learn exactly which named instructions are available in a particular run. For nameless patterns, the condition is applied only when matching an individual insn, and only after the insn has matched the pattern's recognition template. The insn's operands may be found in the vector @{b}operands@{ub}. 4. The @{i}output template@{ui}: a string that says how to output matching insns as assembler code. @{b}%@{ub} in this string specifies where to substitute the value of an operand. See @{"Output Template" Link "Output Template"}. When simple substitution isn't general enough, you can specify a piece of C code to compute the output. See @{"Output Statement" Link "Output Statement"}. 5. Optionally, a vector containing the values of attributes for insns matching this pattern. See @{"Insn Attributes" Link "Insn Attributes"}. @EndNode @Node "Example" "gcc.guide/Example" @Next "RTL Template" @Prev "Patterns" @Toc "Machine Desc" Example of @{b}define_insn@{ub} ====================== Here is an actual example of an instruction pattern, for the 68000/68020. (define_insn "tstsi" [(set (cc0) (match_operand:SI 0 "general_operand" "rm"))] "" "* { if (TARGET_68020 || ! ADDRESS_REG_P (operands[0])) return \\"tstl %0\\"; return \\"cmpl #0,%0\\"; }") This is an instruction that sets the condition codes based on the value of a general operand. It has no condition, so any insn whose RTL description has the form shown may be handled according to this pattern. The name @{b}tstsi@{ub} means "test a @{b}SImode@{ub} value" and tells the RTL generation pass that, when it is necessary to test such a value, an insn to do so can be constructed using this pattern. The output control string is a piece of C code which chooses which output template to return based on the kind of operand and the specific type of CPU for which code is being generated. @{b}"rm"@{ub} is an operand constraint. Its meaning is explained below. @EndNode @Node "RTL Template" "gcc.guide/RTL Template" @Next "Output Template" @Prev "Example" @Toc "Machine Desc" RTL Template ============ The RTL template is used to define which insns match the particular pattern and how to find their operands. For named patterns, the RTL template also says how to construct an insn from specified operands. Construction involves substituting specified operands into a copy of the template. Matching involves determining the values that serve as the operands in the insn being matched. Both of these activities are controlled by special expression types that direct matching and substitution of the operands. @{b}(match_operand:@{I}M@{ui} @{I}N@{ui} @{I}PREDICATE@{ui} @{I}CONSTRAINT@{ui})@{ub} This expression is a placeholder for operand number @{I}N@{ui} of the insn. When constructing an insn, operand number @{I}N@{ui} will be substituted at this point. When matching an insn, whatever appears at this position in the insn will be taken as operand number @{I}N@{ui}; but it must satisfy @{I}PREDICATE@{ui} or this instruction pattern will not match at all. Operand numbers must be chosen consecutively counting from zero in each instruction pattern. There may be only one @{b}match_operand@{ub} expression in the pattern for each operand number. Usually operands are numbered in the order of appearance in @{b}match_operand@{ub} expressions. @{I}PREDICATE@{ui} is a string that is the name of a C function that accepts two arguments, an expression and a machine mode. During matching, the function will be called with the putative operand as the expression and @{I}M@{ui} as the mode argument (if @{I}M@{ui} is not specified, @{b}VOIDmode@{ub} will be used, which normally causes @{I}PREDICATE@{ui} to accept any mode). If it returns zero, this instruction pattern fails to match. @{I}PREDICATE@{ui} may be an empty string; then it means no test is to be done on the operand, so anything which occurs in this position is valid. Most of the time, @{I}PREDICATE@{ui} will reject modes other than @{I}M@{ui}--but not always. For example, the predicate @{b}address_operand@{ub} uses @{I}M@{ui} as the mode of memory ref that the address should be valid for. Many predicates accept @{b}const_int@{ub} nodes even though their mode is @{b}VOIDmode@{ub}. @{I}CONSTRAINT@{ui} controls reloading and the choice of the best register class to use for a value, as explained later (see @{"Constraints" Link "Constraints"}). People are often unclear on the difference between the constraint and the predicate. The predicate helps decide whether a given insn matches the pattern. The constraint plays no role in this decision; instead, it controls various decisions in the case of an insn which does match. On CISC machines, the most common @{I}PREDICATE@{ui} is @{b}"general_operand"@{ub}. This function checks that the putative operand is either a constant, a register or a memory reference, and that it is valid for mode @{I}M@{ui}. For an operand that must be a register, @{I}PREDICATE@{ui} should be @{b}"register_operand"@{ub}. Using @{b}"general_operand"@{ub} would be valid, since the reload pass would copy any non-register operands through registers, but this would make GNU CC do extra work, it would prevent invariant operands (such as constant) from being removed from loops, and it would prevent the register allocator from doing the best possible job. On RISC machines, it is usually most efficient to allow @{I}PREDICATE@{ui} to accept only objects that the constraints allow. For an operand that must be a constant, you must be sure to either use @{b}"immediate_operand"@{ub} for @{I}PREDICATE@{ui}, or make the instruction pattern's extra condition require a constant, or both. You cannot expect the constraints to do this work! If the constraints allow only constants, but the predicate allows something else, the compiler will crash when that case arises. @{b}(match_scratch:@{I}M@{ui} @{I}N@{ui} @{I}CONSTRAINT@{ui})@{ub} This expression is also a placeholder for operand number @{I}N@{ui} and indicates that operand must be a @{b}scratch@{ub} or @{b}reg@{ub} expression. When matching patterns, this is equivalent to (match_operand:@{I}M@{ui} @{I}N@{ui} "scratch_operand" @{I}PRED@{ui}) but, when generating RTL, it produces a (@{b}scratch@{ub}:@{I}M@{ui}) expression. If the last few expressions in a @{b}parallel@{ub} are @{b}clobber@{ub} expressions whose operands are either a hard register or @{b}match_scratch@{ub}, the combiner can add or delete them when necessary. See @{"Side Effects" Link "Side Effects"}. @{b}(match_dup @{I}N@{ui})@{ub} This expression is also a placeholder for operand number @{I}N@{ui}. It is used when the operand needs to appear more than once in the insn. In construction, @{b}match_dup@{ub} acts just like @{b}match_operand@{ub}: the operand is substituted into the insn being constructed. But in matching, @{b}match_dup@{ub} behaves differently. It assumes that operand number @{I}N@{ui} has already been determined by a @{b}match_operand@{ub} appearing earlier in the recognition template, and it matches only an identical-looking expression. @{b}(match_operator:@{I}M@{ui} @{I}N@{ui} @{I}PREDICATE@{ui} [@{I}OPERANDS@{ui} ...])@{ub} This pattern is a kind of placeholder for a variable RTL expression code. When constructing an insn, it stands for an RTL expression whose expression code is taken from that of operand @{I}N@{ui}, and whose operands are constructed from the patterns @{I}OPERANDS@{ui}. When matching an expression, it matches an expression if the function @{I}PREDICATE@{ui} returns nonzero on that expression @{i}and@{ui} the patterns @{I}OPERANDS@{ui} match the operands of the expression. Suppose that the function @{b}commutative_operator@{ub} is defined as follows, to match any expression whose operator is one of the commutative arithmetic operators of RTL and whose mode is @{I}MODE@{ui}: int commutative_operator (x, mode) rtx x; enum machine_mode mode; { enum rtx_code code = GET_CODE (x); if (GET_MODE (x) != mode) return 0; return (GET_RTX_CLASS (code) == 'c' || code == EQ || code == NE); } Then the following pattern will match any RTL expression consisting of a commutative operator applied to two general operands: (match_operator:SI 3 "commutative_operator" [(match_operand:SI 1 "general_operand" "g") (match_operand:SI 2 "general_operand" "g")]) Here the vector @{b} [@{I}OPERANDS@{ui} ...]@{ub} contains two patterns because the expressions to be matched all contain two operands. When this pattern does match, the two operands of the commutative operator are recorded as operands 1 and 2 of the insn. (This is done by the two instances of @{b}match_operand@{ub}.) Operand 3 of the insn will be the entire commutative expression: use @{b}GET_CODE (operands[3])@{ub} to see which commutative operator was used. The machine mode @{I}M@{ui} of @{b}match_operator@{ub} works like that of @{b}match_operand@{ub}: it is passed as the second argument to the predicate function, and that function is solely responsible for deciding whether the expression to be matched "has" that mode. When constructing an insn, argument 3 of the gen-function will specify the operation (i.e. the expression code) for the expression to be made. It should be an RTL expression, whose expression code is copied into a new expression whose operands are arguments 1 and 2 of the gen-function. The subexpressions of argument 3 are not used; only its expression code matters. When @{b}match_operator@{ub} is used in a pattern for matching an insn, it usually best if the operand number of the @{b}match_operator@{ub} is higher than that of the actual operands of the insn. This improves register allocation because the register allocator often looks at operands 1 and 2 of insns to see if it can do register tying. There is no way to specify constraints in @{b}match_operator@{ub}. The operand of the insn which corresponds to the @{b}match_operator@{ub} never has any constraints because it is never reloaded as a whole. However, if parts of its @{I}OPERANDS@{ui} are matched by @{b}match_operand@{ub} patterns, those parts may have constraints of their own. @{b}(match_op_dup:@{I}M@{ui} @{I}N@{ui}[@{I}OPERANDS@{ui} ...])@{ub} Like @{b}match_dup@{ub}, except that it applies to operators instead of operands. When constructing an insn, operand number @{I}N@{ui} will be substituted at this point. But in matching, @{b}match_op_dup@{ub} behaves differently. It assumes that operand number @{I}N@{ui} has already been determined by a @{b}match_operator@{ub} appearing earlier in the recognition template, and it matches only an identical-looking expression. @{b}(match_parallel @{I}N@{ui} @{I}PREDICATE@{ui} [@{I}SUBPAT@{ui} ...])@{ub} This pattern is a placeholder for an insn that consists of a @{b}parallel@{ub} expression with a variable number of elements. This expression should only appear at the top level of an insn pattern. When constructing an insn, operand number @{I}N@{ui} will be substituted at this point. When matching an insn, it matches if the body of the insn is a @{b}parallel@{ub} expression with at least as many elements as the vector of @{I}SUBPAT@{ui} expressions in the @{b}match_parallel@{ub}, if each @{I}SUBPAT@{ui} matches the corresponding element of the @{b}parallel@{ub}, @{i}and@{ui} the function @{I}PREDICATE@{ui} returns nonzero on the @{b}parallel@{ub} that is the body of the insn. It is the responsibility of the predicate to validate elements of the @{b}parallel@{ub} beyond those listed in the @{b}match_parallel@{ub}. A typical use of @{b}match_parallel@{ub} is to match load and store multiple expressions, which can contain a variable number of elements in a @{b}parallel@{ub}. For example, (define_insn "" [(match_parallel 0 "load_multiple_operation" [(set (match_operand:SI 1 "gpc_reg_operand" "=r") (match_operand:SI 2 "memory_operand" "m")) (use (reg:SI 179)) (clobber (reg:SI 179))])] "" "loadm 0,0,%1,%2") This example comes from @{b}a29k.md@{ub}. The function @{b}load_multiple_operations@{ub} is defined in @{b}a29k.c@{ub} and checks that subsequent elements in the @{b}parallel@{ub} are the same as the @{b}set@{ub} in the pattern, except that they are referencing subsequent registers and memory locations. An insn that matches this pattern might look like: (parallel [(set (reg:SI 20) (mem:SI (reg:SI 100))) (use (reg:SI 179)) (clobber (reg:SI 179)) (set (reg:SI 21) (mem:SI (plus:SI (reg:SI 100) (const_int 4)))) (set (reg:SI 22) (mem:SI (plus:SI (reg:SI 100) (const_int 8))))]) @{b}(match_par_dup @{I}N@{ui} [@{I}SUBPAT@{ui} ...])@{ub} Like @{b}match_op_dup@{ub}, but for @{b}match_parallel@{ub} instead of @{b}match_operator@{ub}. @{b}(address (match_operand:@{I}M@{ui} @{I}N@{ui} "address_operand" ""))@{ub} This complex of expressions is a placeholder for an operand number @{I}N@{ui} in a "load address" instruction: an operand which specifies a memory location in the usual way, but for which the actual operand value used is the address of the location, not the contents of the location. @{b}address@{ub} expressions never appear in RTL code, only in machine descriptions. And they are used only in machine descriptions that do not use the operand constraint feature. When operand constraints are in use, the letter @{b}p@{ub} in the constraint serves this purpose. @{I}M@{ui} is the machine mode of the @{i}memory location being addressed@{ui}, not the machine mode of the address itself. That mode is always the same on a given target machine (it is @{b}Pmode@{ub}, which normally is @{b}SImode@{ub}), so there is no point in mentioning it; thus, no machine mode is written in the @{b}address@{ub} expression. If some day support is added for machines in which addresses of different kinds of objects appear differently or are used differently (such as the PDP-10), different formats would perhaps need different machine modes and these modes might be written in the @{b}address@{ub} expression. @EndNode @Node "Output Template" "gcc.guide/Output Template" @Next "Output Statement" @Prev "RTL Template" @Toc "Machine Desc" Output Templates and Operand Substitution ========================================= The @{i}output template@{ui} is a string which specifies how to output the assembler code for an instruction pattern. Most of the template is a fixed string which is output literally. The character @{b}%@{ub} is used to specify where to substitute an operand; it can also be used to identify places where different variants of the assembler require different syntax. In the simplest case, a @{b}%@{ub} followed by a digit @{I}N@{ui} says to output operand @{I}N@{ui} at that point in the string. @{b}%@{ub} followed by a letter and a digit says to output an operand in an alternate fashion. Four letters have standard, built-in meanings described below. The machine description macro @{b}PRINT_OPERAND@{ub} can define additional letters with nonstandard meanings. @{b}%c@{I}DIGIT@{ui}@{ub} can be used to substitute an operand that is a constant value without the syntax that normally indicates an immediate operand. @{b}%n@{I}DIGIT@{ui}@{ub} is like @{b}%c@{I}DIGIT@{ui}@{ub} except that the value of the constant is negated before printing. @{b}%a@{I}DIGIT@{ui}@{ub} can be used to substitute an operand as if it were a memory reference, with the actual operand treated as the address. This may be useful when outputting a "load address" instruction, because often the assembler syntax for such an instruction requires you to write the operand as if it were a memory reference. @{b}%l@{I}DIGIT@{ui}@{ub} is used to substitute a @{b}label_ref@{ub} into a jump instruction. @{b}%=@{ub} outputs a number which is unique to each instruction in the entire compilation. This is useful for making local labels to be referred to more than once in a single template that generates multiple assembler instructions. @{b}%@{ub} followed by a punctuation character specifies a substitution that does not use an operand. Only one case is standard: @{b}%%@{ub} outputs a @{b}%@{ub} into the assembler code. Other nonstandard cases can be defined in the @{b}PRINT_OPERAND@{ub} macro. You must also define which punctuation characters are valid with the @{b}PRINT_OPERAND_PUNCT_VALID_P@{ub} macro. The template may generate multiple assembler instructions. Write the text for the instructions, with @{b}\\;@{ub} between them. When the RTL contains two operands which are required by constraint to match each other, the output template must refer only to the lower-numbered operand. Matching operands are not always identical, and the rest of the compiler arranges to put the proper RTL expression for printing into the lower-numbered operand. One use of nonstandard letters or punctuation following @{b}%@{ub} is to distinguish between different assembler languages for the same machine; for example, Motorola syntax versus MIT syntax for the 68000. Motorola syntax requires periods in most opcode names, while MIT syntax does not. For example, the opcode @{b}movel@{ub} in MIT syntax is @{b}move.l@{ub} in Motorola syntax. The same file of patterns is used for both kinds of output syntax, but the character sequence @{b}%.@{ub} is used in each place where Motorola syntax wants a period. The @{b}PRINT_OPERAND@{ub} macro for Motorola syntax defines the sequence to output a period; the macro for MIT syntax defines it to do nothing. As a special case, a template consisting of the single character @{b}#@{ub} instructs the compiler to first split the insn, and then output the resulting instructions separately. This helps eliminate redundancy in the output templates. If you have a @{b}define_insn@{ub} that needs to emit multiple assembler instructions, and there is an matching @{b}define_split@{ub} already defined, then you can simply use @{b}#@{ub} as the output template instead of writing an output template that emits the multiple assembler instructions. If @{b}ASSEMBLER_DIALECT@{ub} is defined, you can use @{b}{option0|option1|option2}@{ub} constructs in the templates. These describe multiple variants of assembler language syntax. See @{"Instruction Output" Link "Instruction Output"}. @EndNode @Node "Output Statement" "gcc.guide/Output Statement" @Next "Constraints" @Prev "Output Template" @Toc "Machine Desc" C Statements for Assembler Output ================================= Often a single fixed template string cannot produce correct and efficient assembler code for all the cases that are recognized by a single instruction pattern. For example, the opcodes may depend on the kinds of operands; or some unfortunate combinations of operands may require extra machine instructions. If the output control string starts with a @{b} @@{ub}, then it is actually a series of templates, each on a separate line. (Blank lines and leading spaces and tabs are ignored.) The templates correspond to the pattern's constraint alternatives (see @{"Multi-Alternative" Link "Multi-Alternative"}). For example, if a target machine has a two-address add instruction @{b}addr@{ub} to add into a register and another @{b}addm@{ub} to add a register to memory, you might write this pattern: (define_insn "addsi3" [(set (match_operand:SI 0 "general_operand" "=r,m") (plus:SI (match_operand:SI 1 "general_operand" "0,0") (match_operand:SI 2 "general_operand" "g,r")))] "" "@ addr %2,%0 addm %2,%0") If the output control string starts with a @{b}*@{ub}, then it is not an output template but rather a piece of C program that should compute a template. It should execute a @{b}return@{ub} statement to return the template-string you want. Most such templates use C string literals, which require doublequote characters to delimit them. To include these doublequote characters in the string, prefix each one with @{b}\\ @{ub}. The operands may be found in the array @{b}operands@{ub}, whose C data type is @{b}rtx []@{ub}. It is very common to select different ways of generating assembler code based on whether an immediate operand is within a certain range. Be careful when doing this, because the result of @{b}INTVAL@{ub} is an integer on the host machine. If the host machine has more bits in an @{b}int@{ub} than the target machine has in the mode in which the constant will be used, then some of the bits you get from @{b}INTVAL@{ub} will be superfluous. For proper results, you must carefully disregard the values of those bits. It is possible to output an assembler instruction and then go on to output or compute more of them, using the subroutine @{b}output_asm_insn@{ub}. This receives two arguments: a template-string and a vector of operands. The vector may be @{b}operands@{ub}, or it may be another array of @{b}rtx@{ub} that you declare locally and initialize yourself. When an insn pattern has multiple alternatives in its constraints, often the appearance of the assembler code is determined mostly by which alternative was matched. When this is so, the C code can test the variable @{b}which_alternative@{ub}, which is the ordinal number of the alternative that was actually satisfied (0 for the first, 1 for the second alternative, etc.). For example, suppose there are two opcodes for storing zero, @{b}clrreg@{ub} for registers and @{b}clrmem@{ub} for memory locations. Here is how a pattern could use @{b}which_alternative@{ub} to choose between them: (define_insn "" [(set (match_operand:SI 0 "general_operand" "=r,m") (const_int 0))] "" "* return (which_alternative == 0 ? \\"clrreg %0\\" : \\"clrmem %0\\"); ") The example above, where the assembler code to generate was @{i}solely@{ui} determined by the alternative, could also have been specified as follows, having the output control string start with a @{b} @@{ub}: (define_insn "" [(set (match_operand:SI 0 "general_operand" "=r,m") (const_int 0))] "" "@ clrreg %0 clrmem %0") @EndNode @Node "Constraints" "gcc.guide/Constraints" @Next "Standard Names" @Prev "Output Statement" @Toc "Machine Desc" Operand Constraints =================== Each @{b}match_operand@{ub} in an instruction pattern can specify a constraint for the type of operands allowed. Constraints can say whether an operand may be in a register, and which kinds of register; whether the operand can be a memory reference, and which kinds of address; whether the operand may be an immediate constant, and which possible values it may have. Constraints can also require two operands to match. @{" Simple Constraints " Link "Simple Constraints"} Basic use of constraints. @{" Multi-Alternative " Link "Multi-Alternative"} When an insn has two alternative constraint-patterns. @{" Class Preferences " Link "Class Preferences"} Constraints guide which hard register to put things in. @{" Modifiers " Link "Modifiers"} More precise control over effects of constraints. @{" Machine Constraints " Link "Machine Constraints"} Existing constraints for some particular machines. @{" No Constraints " Link "No Constraints"} Describing a clean machine without constraints. @EndNode @Node "Simple Constraints" "gcc.guide/Simple Constraints" @Next "Multi-Alternative" @Toc "Constraints" Simple Constraints ------------------ The simplest kind of constraint is a string full of letters, each of which describes one kind of operand that is permitted. Here are the letters that are allowed: @{b}m@{ub} A memory operand is allowed, with any kind of address that the machine supports in general. @{b}o@{ub} A memory operand is allowed, but only if the address is @{i}offsettable@{ui}. This means that adding a small integer (actually, the width in bytes of the operand, as determined by its machine mode) may be added to the address and the result is also a valid memory address. For example, an address which is constant is offsettable; so is an address that is the sum of a register and a constant (as long as a slightly larger constant is also within the range of address-offsets supported by the machine); but an autoincrement or autodecrement address is not offsettable. More complicated indirect/indexed addresses may or may not be offsettable depending on the other addressing modes that the machine supports. Note that in an output operand which can be matched by another operand, the constraint letter @{b}o@{ub} is valid only when accompanied by both @{b}<@{ub} (if the target machine has predecrement addressing) and @{b}>@{ub} (if the target machine has preincrement addressing). @{b}V@{ub} A memory operand that is not offsettable. In other words, anything that would fit the @{b}m@{ub} constraint but not the @{b}o@{ub} constraint. @{b}<@{ub} A memory operand with autodecrement addressing (either predecrement or postdecrement) is allowed. @{b}>@{ub} A memory operand with autoincrement addressing (either preincrement or postincrement) is allowed. @{b}r@{ub} A register operand is allowed provided that it is in a general register. @{b}d@{ub}, @{b}a@{ub}, @{b}f@{ub}, ... Other letters can be defined in machine-dependent fashion to stand for particular classes of registers. @{b}d@{ub}, @{b}a@{ub} and @{b}f@{ub} are defined on the 68000/68020 to stand for data, address and floating point registers. @{b}i@{ub} An immediate integer operand (one with constant value) is allowed. This includes symbolic constants whose values will be known only at assembly time. @{b}n@{ub} An immediate integer operand with a known numeric value is allowed. Many systems cannot support assembly-time constants for operands less than a word wide. Constraints for these operands should use @{b}n@{ub} rather than @{b}i@{ub}. @{b}I@{ub}, @{b}J@{ub}, @{b}K@{ub}, ... @{b}P@{ub} Other letters in the range @{b}I@{ub} through @{b}P@{ub} may be defined in a machine-dependent fashion to permit immediate integer operands with explicit integer values in specified ranges. For example, on the 68000, @{b}I@{ub} is defined to stand for the range of values 1 to 8. This is the range permitted as a shift count in the shift instructions. @{b}E@{ub} An immediate floating operand (expression code @{b}const_double@{ub}) is allowed, but only if the target floating point format is the same as that of the host machine (on which the compiler is running). @{b}F@{ub} An immediate floating operand (expression code @{b}const_double@{ub}) is allowed. @{b}G@{ub}, @{b}H@{ub} @{b}G@{ub} and @{b}H@{ub} may be defined in a machine-dependent fashion to permit immediate floating operands in particular ranges of values. @{b}s@{ub} An immediate integer operand whose value is not an explicit integer is allowed. This might appear strange; if an insn allows a constant operand with a value not known at compile time, it certainly must allow any known value. So why use @{b}s@{ub} instead of @{b}i@{ub}? Sometimes it allows better code to be generated. For example, on the 68000 in a fullword instruction it is possible to use an immediate operand; but if the immediate value is between -128 and 127, better code results from loading the value into a register and using the register. This is because the load into the register can be done with a @{b}moveq@{ub} instruction. We arrange for this to happen by defining the letter @{b}K@{ub} to mean "any integer outside the range -128 to 127", and then specifying @{b}Ks@{ub} in the operand constraints. @{b}g@{ub} Any register, memory or immediate integer operand is allowed, except for registers that are not general registers. @{b}X@{ub} Any operand whatsoever is allowed, even if it does not satisfy @{b}general_operand@{ub}. This is normally used in the constraint of a @{b}match_scratch@{ub} when certain alternatives will not actually require a scratch register. @{b}0@{ub}, @{b}1@{ub}, @{b}2@{ub}, ... @{b}9@{ub} An operand that matches the specified operand number is allowed. If a digit is used together with letters within the same alternative, the digit should come last. This is called a @{i}matching constraint@{ui} and what it really means is that the assembler has only a single operand that fills two roles considered separate in the RTL insn. For example, an add insn has two input operands and one output operand in the RTL, but on most CISC machines an add instruction really has only two operands, one of them an input-output operand: addl #35,r12 Matching constraints are used in these circumstances. More precisely, the two operands that match must include one input-only operand and one output-only operand. Moreover, the digit must be a smaller number than the number of the operand that uses it in the constraint. For operands to match in a particular case usually means that they are identical-looking RTL expressions. But in a few special cases specific kinds of dissimilarity are allowed. For example, @{b}*x@{ub} as an input operand will match @{b}*x++@{ub} as an output operand. For proper results in such cases, the output template should always use the output-operand's number when printing the operand. @{b}p@{ub} An operand that is a valid memory address is allowed. This is for "load address" and "push address" instructions. @{b}p@{ub} in the constraint must be accompanied by @{b}address_operand@{ub} as the predicate in the @{b}match_operand@{ub}. This predicate interprets the mode specified in the @{b}match_operand@{ub} as the mode of the memory reference for which the address would be valid. @{b}Q@{ub}, @{b}R@{ub}, @{b}S@{ub}, ... @{b}U@{ub} Letters in the range @{b}Q@{ub} through @{b}U@{ub} may be defined in a machine-dependent fashion to stand for arbitrary operand types. The machine description macro @{b}EXTRA_CONSTRAINT@{ub} is passed the operand as its first argument and the constraint letter as its second operand. A typical use for this would be to distinguish certain types of memory references that affect other insn operands. Do not define these constraint letters to accept register references (@{b}reg@{ub}); the reload pass does not expect this and would not handle it properly. In order to have valid assembler code, each operand must satisfy its constraint. But a failure to do so does not prevent the pattern from applying to an insn. Instead, it directs the compiler to modify the code so that the constraint will be satisfied. Usually this is done by copying an operand into a register. Contrast, therefore, the two instruction patterns that follow: (define_insn "" [(set (match_operand:SI 0 "general_operand" "=r") (plus:SI (match_dup 0) (match_operand:SI 1 "general_operand" "r")))] "" "...") which has two operands, one of which must appear in two places, and (define_insn "" [(set (match_operand:SI 0 "general_operand" "=r") (plus:SI (match_operand:SI 1 "general_operand" "0") (match_operand:SI 2 "general_operand" "r")))] "" "...") which has three operands, two of which are required by a constraint to be identical. If we are considering an insn of the form (insn @{I}N@{ui} @{I}PREV@{ui} @{I}NEXT@{ui} (set (reg:SI 3) (plus:SI (reg:SI 6) (reg:SI 109))) ...) the first pattern would not apply at all, because this insn does not contain two identical subexpressions in the right place. The pattern would say, "That does not look like an add instruction; try other patterns." The second pattern would say, "Yes, that's an add instruction, but there is something wrong with it." It would direct the reload pass of the compiler to generate additional insns to make the constraint true. The results might look like this: (insn @{I}N2@{ui} @{I}PREV@{ui} @{I}N@{ui} (set (reg:SI 3) (reg:SI 6)) ...) (insn @{I}N@{ui} @{I}N2@{ui} @{I}NEXT@{ui} (set (reg:SI 3) (plus:SI (reg:SI 3) (reg:SI 109))) ...) It is up to you to make sure that each operand, in each pattern, has constraints that can handle any RTL expression that could be present for that operand. (When multiple alternatives are in use, each pattern must, for each possible combination of operand expressions, have at least one alternative which can handle that combination of operands.) The constraints don't need to @{i}allow@{ui} any possible operand--when this is the case, they do not constrain--but they must at least point the way to reloading any possible operand so that it will fit. @{b}*@{ub} If the constraint accepts whatever operands the predicate permits, there is no problem: reloading is never necessary for this operand. For example, an operand whose constraints permit everything except registers is safe provided its predicate rejects registers. An operand whose predicate accepts only constant values is safe provided its constraints include the letter @{b}i@{ub}. If any possible constant value is accepted, then nothing less than @{b}i@{ub} will do; if the predicate is more selective, then the constraints may also be more selective. @{b}*@{ub} Any operand expression can be reloaded by copying it into a register. So if an operand's constraints allow some kind of register, it is certain to be safe. It need not permit all classes of registers; the compiler knows how to copy a register into another register of the proper class in order to make an instruction valid. @{b}*@{ub} A nonoffsettable memory reference can be reloaded by copying the address into a register. So if the constraint uses the letter @{b}o@{ub}, all memory references are taken care of. @{b}*@{ub} A constant operand can be reloaded by allocating space in memory to hold it as preinitialized data. Then the memory reference can be used in place of the constant. So if the constraint uses the letters @{b}o@{ub} or @{b}m@{ub}, constant operands are not a problem. @{b}*@{ub} If the constraint permits a constant and a pseudo register used in an insn was not allocated to a hard register and is equivalent to a constant, the register will be replaced with the constant. If the predicate does not permit a constant and the insn is re-recognized for some reason, the compiler will crash. Thus the predicate must always recognize any objects allowed by the constraint. If the operand's predicate can recognize registers, but the constraint does not permit them, it can make the compiler crash. When this operand happens to be a register, the reload pass will be stymied, because it does not know how to copy a register temporarily into memory. @EndNode @Node "Multi-Alternative" "gcc.guide/Multi-Alternative" @Next "Class Preferences" @Prev "Simple Constraints" @Toc "Constraints" Multiple Alternative Constraints -------------------------------- Sometimes a single instruction has multiple alternative sets of possible operands. For example, on the 68000, a logical-or instruction can combine register or an immediate value into memory, or it can combine any kind of operand into a register; but it cannot combine one memory location into another. These constraints are represented as multiple alternatives. An alternative can be described by a series of letters for each operand. The overall constraint for an operand is made from the letters for this operand from the first alternative, a comma, the letters for this operand from the second alternative, a comma, and so on until the last alternative. Here is how it is done for fullword logical-or on the 68000: (define_insn "iorsi3" [(set (match_operand:SI 0 "general_operand" "=m,d") (ior:SI (match_operand:SI 1 "general_operand" "%0,0") (match_operand:SI 2 "general_operand" "dKs,dmKs")))] ...) The first alternative has @{b}m@{ub} (memory) for operand 0, @{b}0@{ub} for operand 1 (meaning it must match operand 0), and @{b}dKs@{ub} for operand 2. The second alternative has @{b}d@{ub} (data register) for operand 0, @{b}0@{ub} for operand 1, and @{b}dmKs@{ub} for operand 2. The @{b}=@{ub} and @{b}%@{ub} in the constraints apply to all the alternatives; their meaning is explained in the next section (see @{"Class Preferences" Link "Class Preferences"}). If all the operands fit any one alternative, the instruction is valid. Otherwise, for each alternative, the compiler counts how many instructions must be added to copy the operands so that that alternative applies. The alternative requiring the least copying is chosen. If two alternatives need the same amount of copying, the one that comes first is chosen. These choices can be altered with the @{b}?@{ub} and @{b}!@{ub} characters: @{b}?@{ub} Disparage slightly the alternative that the @{b}?@{ub} appears in, as a choice when no alternative applies exactly. The compiler regards this alternative as one unit more costly for each @{b}?@{ub} that appears in it. @{b}!@{ub} Disparage severely the alternative that the @{b}!@{ub} appears in. This alternative can still be used if it fits without reloading, but if reloading is needed, some other alternative will be used. When an insn pattern has multiple alternatives in its constraints, often the appearance of the assembler code is determined mostly by which alternative was matched. When this is so, the C code for writing the assembler code can use the variable @{b}which_alternative@{ub}, which is the ordinal number of the alternative that was actually satisfied (0 for the first, 1 for the second alternative, etc.). See @{"Output Statement" Link "Output Statement"}. @EndNode @Node "Class Preferences" "gcc.guide/Class Preferences" @Next "Modifiers" @Prev "Multi-Alternative" @Toc "Constraints" Register Class Preferences -------------------------- The operand constraints have another function: they enable the compiler to decide which kind of hardware register a pseudo register is best allocated to. The compiler examines the constraints that apply to the insns that use the pseudo register, looking for the machine-dependent letters such as @{b}d@{ub} and @{b}a@{ub} that specify classes of registers. The pseudo register is put in whichever class gets the most "votes". The constraint letters @{b}g@{ub} and @{b}r@{ub} also vote: they vote in favor of a general register. The machine description says which registers are considered general. Of course, on some machines all registers are equivalent, and no register classes are defined. Then none of this complexity is relevant. @EndNode @Node "Modifiers" "gcc.guide/Modifiers" @Next "Machine Constraints" @Prev "Class Preferences" @Toc "Constraints" Constraint Modifier Characters ------------------------------ Here are constraint modifier characters. @{b}=@{ub} Means that this operand is write-only for this instruction: the previous value is discarded and replaced by output data. @{b}+@{ub} Means that this operand is both read and written by the instruction. When the compiler fixes up the operands to satisfy the constraints, it needs to know which operands are inputs to the instruction and which are outputs from it. @{b}=@{ub} identifies an output; @{b}+@{ub} identifies an operand that is both input and output; all other operands are assumed to be input only. @{b}&@{ub} Means (in a particular alternative) that this operand is written before the instruction is finished using the input operands. Therefore, this operand may not lie in a register that is used as an input operand or as part of any memory address. @{b}&@{ub} applies only to the alternative in which it is written. In constraints with multiple alternatives, sometimes one alternative requires @{b}&@{ub} while others do not. See, for example, the @{b}movdf@{ub} insn of the 68000. @{b}&@{ub} does not obviate the need to write @{b}=@{ub}. @{b}%@{ub} Declares the instruction to be commutative for this operand and the following operand. This means that the compiler may interchange the two operands if that is the cheapest way to make all operands fit the constraints. This is often used in patterns for addition instructions that really have only two operands: the result must go in one of the arguments. Here for example, is how the 68000 halfword-add instruction is defined: (define_insn "addhi3" [(set (match_operand:HI 0 "general_operand" "=m,r") (plus:HI (match_operand:HI 1 "general_operand" "%0,0") (match_operand:HI 2 "general_operand" "di,g")))] ...) @{b}#@{ub} Says that all following characters, up to the next comma, are to be ignored as a constraint. They are significant only for choosing register preferences. @{b}*@{ub} Says that the following character should be ignored when choosing register preferences. @{b}*@{ub} has no effect on the meaning of the constraint as a constraint, and no effect on reloading. Here is an example: the 68000 has an instruction to sign-extend a halfword in a data register, and can also sign-extend a value by copying it into an address register. While either kind of register is acceptable, the constraints on an address-register destination are less strict, so it is best if register allocation makes an address register its goal. Therefore, @{b}*@{ub} is used so that the @{b}d@{ub} constraint letter (for data register) is ignored when computing register preferences. (define_insn "extendhisi2" [(set (match_operand:SI 0 "general_operand" "=*d,a") (sign_extend:SI (match_operand:HI 1 "general_operand" "0,g")))] ...) @EndNode @Node "Machine Constraints" "gcc.guide/Machine Constraints" @Next "No Constraints" @Prev "Modifiers" @Toc "Constraints" Constraints for Particular Machines ----------------------------------- Whenever possible, you should use the general-purpose constraint letters in @{b}asm@{ub} arguments, since they will convey meaning more readily to people reading your code. Failing that, use the constraint letters that usually have very similar meanings across architectures. The most commonly used constraints are @{b}m@{ub} and @{b}r@{ub} (for memory and general-purpose registers respectively; see @{"Simple Constraints" Link "Simple Constraints"}), and @{b}I@{ub}, usually the letter indicating the most common immediate-constant format. For each machine architecture, the @{b}config/@{I}MACHINE@{ui}.h@{ub} file defines additional constraints. These constraints are used by the compiler itself for instruction generation, as well as for @{b}asm@{ub} statements; therefore, some of the constraints are not particularly interesting for @{b}asm@{ub}. The constraints are defined through these macros: @{b}REG_CLASS_FROM_LETTER@{ub} Register class constraints (usually lower case). @{b}CONST_OK_FOR_LETTER_P@{ub} Immediate constant constraints, for non-floating point constants of word size or smaller precision (usually upper case). @{b}CONST_DOUBLE_OK_FOR_LETTER_P@{ub} Immediate constant constraints, for all floating point constants and for constants of greater than word size precision (usually upper case). @{b}EXTRA_CONSTRAINT@{ub} Special cases of registers or memory. This macro is not required, and is only defined for some machines. Inspecting these macro definitions in the compiler source for your machine is the best way to be certain you have the right constraints. However, here is a summary of the machine-dependent constraints available on some particular machines. @{i}ARM family--@{b}arm.h@{ub}@{ui} @{b}f@{ub} Floating-point register @{b}F@{ub} One of the floating-point constants 0.0, 0.5, 1.0, 2.0, 3.0, 4.0, 5.0 or 10.0 @{b}G@{ub} Floating-point constant that would satisfy the constraint @{b}F@{ub} if it were negated @{b}I@{ub} Integer that is valid as an immediate operand in a data processing instruction. That is, an integer in the range 0 to 255 rotated by a multiple of 2 @{b}J@{ub} Integer in the range -4095 to 4095 @{b}K@{ub} Integer that satisfies constraint @{b}I@{ub} when inverted (ones complement) @{b}L@{ub} Integer that satisfies constraint @{b}I@{ub} when negated (twos complement) @{b}M@{ub} Integer in the range 0 to 32 @{b}Q@{ub} A memory reference where the exact address is in a single register (`@{b}m@{ub}' is preferable for @{b}asm@{ub} statements) @{b}R@{ub} An item in the constant pool @{b}S@{ub} A symbol in the text segment of the current file @{i}AMD 29000 family--@{b}a29k.h@{ub}@{ui} @{b}l@{ub} Local register 0 @{b}b@{ub} Byte Pointer (@{b}BP@{ub}) register @{b}q@{ub} @{b}Q@{ub} register @{b}h@{ub} Special purpose register @{b}A@{ub} First accumulator register @{b}a@{ub} Other accumulator register @{b}f@{ub} Floating point register @{b}I@{ub} Constant greater than 0, less than 0x100 @{b}J@{ub} Constant greater than 0, less than 0x10000 @{b}K@{ub} Constant whose high 24 bits are on (1) @{b}L@{ub} 16 bit constant whose high 8 bits are on (1) @{b}M@{ub} 32 bit constant whose high 16 bits are on (1) @{b}N@{ub} 32 bit negative constant that fits in 8 bits @{b}O@{ub} The constant 0x80000000 or, on the 29050, any 32 bit constant whose low 16 bits are 0. @{b}P@{ub} 16 bit negative constant that fits in 8 bits @{b}G@{ub} @{b}H@{ub} A floating point constant (in @{b}asm@{ub} statements, use the machine independent @{b}E@{ub} or @{b}F@{ub} instead) @{i}IBM RS6000--@{b}rs6000.h@{ub}@{ui} @{b}b@{ub} Address base register @{b}f@{ub} Floating point register @{b}h@{ub} @{b}MQ@{ub}, @{b}CTR@{ub}, or @{b}LINK@{ub} register @{b}q@{ub} @{b}MQ@{ub} register @{b}c@{ub} @{b}CTR@{ub} register @{b}l@{ub} @{b}LINK@{ub} register @{b}x@{ub} @{b}CR@{ub} register (condition register) number 0 @{b}y@{ub} @{b}CR@{ub} register (condition register) @{b}I@{ub} Signed 16 bit constant @{b}J@{ub} Constant whose low 16 bits are 0 @{b}K@{ub} Constant whose high 16 bits are 0 @{b}L@{ub} Constant suitable as a mask operand @{b}M@{ub} Constant larger than 31 @{b}N@{ub} Exact power of 2 @{b}O@{ub} Zero @{b}P@{ub} Constant whose negation is a signed 16 bit constant @{b}G@{ub} Floating point constant that can be loaded into a register with one instruction per word @{b}Q@{ub} Memory operand that is an offset from a register (@{b}m@{ub} is preferable for @{b}asm@{ub} statements) @{i}Intel 386--@{b}i386.h@{ub}@{ui} @{b}q@{ub} @{b}a@{ub}, @{b}b@{ub}, @{b}c@{ub}, or @{b}d@{ub} register @{b}A@{ub} @{b}a@{ub}, or @{b}d@{ub} register (for 64-bit ints) @{b}f@{ub} Floating point register @{b}t@{ub} First (top of stack) floating point register @{b}u@{ub} Second floating point register @{b}a@{ub} @{b}a@{ub} register @{b}b@{ub} @{b}b@{ub} register @{b}c@{ub} @{b}c@{ub} register @{b}d@{ub} @{b}d@{ub} register @{b}D@{ub} @{b}di@{ub} register @{b}S@{ub} @{b}si@{ub} register @{b}I@{ub} Constant in range 0 to 31 (for 32 bit shifts) @{b}J@{ub} Constant in range 0 to 63 (for 64 bit shifts) @{b}K@{ub} @{b}0xff@{ub} @{b}L@{ub} @{b}0xffff@{ub} @{b}M@{ub} 0, 1, 2, or 3 (shifts for @{b}lea@{ub} instruction) @{b}N@{ub} Constant in range 0 to 255 (for @{b}out@{ub} instruction) @{b}G@{ub} Standard 80387 floating point constant @{i}Intel 960--@{b}i960.h@{ub}@{ui} @{b}f@{ub} Floating point register (@{b}fp0@{ub} to @{b}fp3@{ub}) @{b}l@{ub} Local register (@{b}r0@{ub} to @{b}r15@{ub}) @{b}b@{ub} Global register (@{b}g0@{ub} to @{b}g15@{ub}) @{b}d@{ub} Any local or global register @{b}I@{ub} Integers from 0 to 31 @{b}J@{ub} 0 @{b}K@{ub} Integers from -31 to 0 @{b}G@{ub} Floating point 0 @{b}H@{ub} Floating point 1 @{i}MIPS--@{b}mips.h@{ub}@{ui} @{b}d@{ub} General-purpose integer register @{b}f@{ub} Floating-point register (if available) @{b}h@{ub} @{b}Hi@{ub} register @{b}l@{ub} @{b}Lo@{ub} register @{b}x@{ub} @{b}Hi@{ub} or @{b}Lo@{ub} register @{b}y@{ub} General-purpose integer register @{b}z@{ub} Floating-point status register @{b}I@{ub} Signed 16 bit constant (for arithmetic instructions) @{b}J@{ub} Zero @{b}K@{ub} Zero-extended 16-bit constant (for logic instructions) @{b}L@{ub} Constant with low 16 bits zero (can be loaded with @{b}lui@{ub}) @{b}M@{ub} 32 bit constant which requires two instructions to load (a constant which is not @{b}I@{ub}, @{b}K@{ub}, or @{b}L@{ub}) @{b}N@{ub} Negative 16 bit constant @{b}O@{ub} Exact power of two @{b}P@{ub} Positive 16 bit constant @{b}G@{ub} Floating point zero @{b}Q@{ub} Memory reference that can be loaded with more than one instruction (@{b}m@{ub} is preferable for @{b}asm@{ub} statements) @{b}R@{ub} Memory reference that can be loaded with one instruction (@{b}m@{ub} is preferable for @{b}asm@{ub} statements) @{b}S@{ub} Memory reference in external OSF/rose PIC format (@{b}m@{ub} is preferable for @{b}asm@{ub} statements) @{i}Motorola 680x0--@{b}m68k.h@{ub}@{ui} @{b}a@{ub} Address register @{b}d@{ub} Data register @{b}f@{ub} 68881 floating-point register, if available @{b}x@{ub} Sun FPA (floating-point) register, if available @{b}y@{ub} First 16 Sun FPA registers, if available @{b}I@{ub} Integer in the range 1 to 8 @{b}J@{ub} 16 bit signed number @{b}K@{ub} Signed number whose magnitude is greater than 0x80 @{b}L@{ub} Integer in the range -8 to -1 @{b}G@{ub} Floating point constant that is not a 68881 constant @{b}H@{ub} Floating point constant that can be used by Sun FPA @{i}SPARC--@{b}sparc.h@{ub}@{ui} @{b}f@{ub} Floating-point register @{b}I@{ub} Signed 13 bit constant @{b}J@{ub} Zero @{b}K@{ub} 32 bit constant with the low 12 bits clear (a constant that can be loaded with the @{b}sethi@{ub} instruction) @{b}G@{ub} Floating-point zero @{b}H@{ub} Signed 13 bit constant, sign-extended to 32 or 64 bits @{b}Q@{ub} Memory reference that can be loaded with one instruction (@{b}m@{ub} is more appropriate for @{b}asm@{ub} statements) @{b}S@{ub} Constant, or memory address @{b}T@{ub} Memory address aligned to an 8-byte boundary @{b}U@{ub} Even register @EndNode @Node "No Constraints" "gcc.guide/No Constraints" @Prev "Machine Constraints" @Toc "Constraints" Not Using Constraints --------------------- Some machines are so clean that operand constraints are not required. For example, on the Vax, an operand valid in one context is valid in any other context. On such a machine, every operand constraint would be @{b}g@{ub}, excepting only operands of "load address" instructions which are written as if they referred to a memory location's contents but actual refer to its address. They would have constraint @{b}p@{ub}. For such machines, instead of writing @{b}g@{ub} and @{b}p@{ub} for all the constraints, you can choose to write a description with empty constraints. Then you write @{b}""@{ub} for the constraint in every @{b}match_operand@{ub}. Address operands are identified by writing an @{b}address@{ub} expression around the @{b}match_operand@{ub}, not by their constraints. When the machine description has just empty constraints, certain parts of compilation are skipped, making the compiler faster. However, few machines actually do not need constraints; all machine descriptions now in existence use constraints. @EndNode @Node "Standard Names" "gcc.guide/Standard Names" @Next "Pattern Ordering" @Prev "Constraints" @Toc "Machine Desc" Standard Pattern Names For Generation ===================================== Here is a table of the instruction names that are meaningful in the RTL generation pass of the compiler. Giving one of these names to an instruction pattern tells the RTL generation pass that it can use the pattern in to accomplish a certain task. @{b}mov@{I}M@{ui}@{ub} Here @{I}M@{ui} stands for a two-letter machine mode name, in lower case. This instruction pattern moves data with that machine mode from operand 1 to operand 0. For example, @{b}movsi@{ub} moves full-word data. If operand 0 is a @{b}subreg@{ub} with mode @{I}M@{ui} of a register whose own mode is wider than @{I}M@{ui}, the effect of this instruction is to store the specified value in the part of the register that corresponds to mode @{I}M@{ui}. The effect on the rest of the register is undefined. This class of patterns is special in several ways. First of all, each of these names @{i}must@{ui} be defined, because there is no other way to copy a datum from one place to another. Second, these patterns are not used solely in the RTL generation pass. Even the reload pass can generate move insns to copy values from stack slots into temporary registers. When it does so, one of the operands is a hard register and the other is an operand that can need to be reloaded into a register. Therefore, when given such a pair of operands, the pattern must generate RTL which needs no reloading and needs no temporary registers--no registers other than the operands. For example, if you support the pattern with a @{b}define_expand@{ub}, then in such a case the @{b}define_expand@{ub} mustn't call @{b}force_reg@{ub} or any other such function which might generate new pseudo registers. This requirement exists even for subword modes on a RISC machine where fetching those modes from memory normally requires several insns and some temporary registers. Look in @{b}spur.md@{ub} to see how the requirement can be satisfied. During reload a memory reference with an invalid address may be passed as an operand. Such an address will be replaced with a valid address later in the reload pass. In this case, nothing may be done with the address except to use it as it stands. If it is copied, it will not be replaced with a valid address. No attempt should be made to make such an address into a valid address and no routine (such as @{b}change_address@{ub}) that will do so may be called. Note that @{b}general_operand@{ub} will fail when applied to such an address. The global variable @{b}reload_in_progress@{ub} (which must be explicitly declared if required) can be used to determine whether such special handling is required. The variety of operands that have reloads depends on the rest of the machine description, but typically on a RISC machine these can only be pseudo registers that did not get hard registers, while on other machines explicit memory references will get optional reloads. If a scratch register is required to move an object to or from memory, it can be allocated using @{b}gen_reg_rtx@{ub} prior to reload. But this is impossible during and after reload. If there are cases needing scratch registers after reload, you must define @{b}SECONDARY_INPUT_RELOAD_CLASS@{ub} and perhaps also @{b}SECONDARY_OUTPUT_RELOAD_CLASS@{ub} to detect them, and provide patterns @{b}reload_in@{I}M@{ui}@{ub} or @{b}reload_out@{I}M@{ui}@{ub} to handle them. See @{"Register Classes" Link "Register Classes"}. The constraints on a @{b}move@{I}M@{ui}@{ub} must permit moving any hard register to any other hard register provided that @{b}HARD_REGNO_MODE_OK@{ub} permits mode @{I}M@{ui} in both registers and @{b}REGISTER_MOVE_COST@{ub} applied to their classes returns a value of 2. It is obligatory to support floating point @{b}move@{I}M@{ui}@{ub} instructions into and out of any registers that can hold fixed point values, because unions and structures (which have modes @{b}SImode@{ub} or @{b}DImode@{ub}) can be in those registers and they may have floating point members. There may also be a need to support fixed point @{b}move@{I}M@{ui}@{ub} instructions in and out of floating point registers. Unfortunately, I have forgotten why this was so, and I don't know whether it is still true. If @{b}HARD_REGNO_MODE_OK@{ub} rejects fixed point values in floating point registers, then the constraints of the fixed point @{b}move@{I}M@{ui}@{ub} instructions must be designed to avoid ever trying to reload into a floating point register. @{b}reload_in@{I}M@{ui}@{ub} @{b}reload_out@{I}M@{ui}@{ub} Like @{b}mov@{I}M@{ui}@{ub}, but used when a scratch register is required to move between operand 0 and operand 1. Operand 2 describes the scratch register. See the discussion of the @{b}SECONDARY_RELOAD_CLASS@{ub} macro in see @{"Register Classes" Link "Register Classes"}. @{b}movstrict@{I}M@{ui}@{ub} Like @{b}mov@{I}M@{ui}@{ub} except that if operand 0 is a @{b}subreg@{ub} with mode @{I}M@{ui} of a register whose natural mode is wider, the @{b}movstrict@{I}M@{ui}@{ub} instruction is guaranteed not to alter any of the register except the part which belongs to mode @{I}M@{ui}. @{b}load_multiple@{ub} Load several consecutive memory locations into consecutive registers. Operand 0 is the first of the consecutive registers, operand 1 is the first memory location, and operand 2 is a constant: the number of consecutive registers. Define this only if the target machine really has such an instruction; do not define this if the most efficient way of loading consecutive registers from memory is to do them one at a time. On some machines, there are restrictions as to which consecutive registers can be stored into memory, such as particular starting or ending register numbers or only a range of valid counts. For those machines, use a @{b}define_expand@{ub} (see @{"Expander Definitions" Link "Expander Definitions"}) and make the pattern fail if the restrictions are not met. Write the generated insn as a @{b}parallel@{ub} with elements being a @{b}set@{ub} of one register from the appropriate memory location (you may also need @{b}use@{ub} or @{b}clobber@{ub} elements). Use a @{b}match_parallel@{ub} (see @{"RTL Template" Link "RTL Template"}) to recognize the insn. See @{b}a29k.md@{ub} and @{b}rs6000.md@{ub} for examples of the use of this insn pattern. @{b}store_multiple@{ub} Similar to @{b}load_multiple@{ub}, but store several consecutive registers into consecutive memory locations. Operand 0 is the first of the consecutive memory locations, operand 1 is the first register, and operand 2 is a constant: the number of consecutive registers. @{b}add@{I}M@{ui}3@{ub} Add operand 2 and operand 1, storing the result in operand 0. All operands must have mode @{I}M@{ui}. This can be used even on two-address machines, by means of constraints requiring operands 1 and 0 to be the same location. @{b}sub@{I}M@{ui}3@{ub}, @{b}mul@{I}M@{ui}3@{ub} @{b}div@{I}M@{ui}3@{ub}, @{b}udiv@{I}M@{ui}3@{ub}, @{b}mod@{I}M@{ui}3@{ub}, @{b}umod@{I}M@{ui}3@{ub} @{b}smin@{I}M@{ui}3@{ub}, @{b}smax@{I}M@{ui}3@{ub}, @{b}umin@{I}M@{ui}3@{ub}, @{b}umax@{I}M@{ui}3@{ub} @{b}and@{I}M@{ui}3@{ub}, @{b}ior@{I}M@{ui}3@{ub}, @{b}xor@{I}M@{ui}3@{ub} Similar, for other arithmetic operations. @{b}mulhisi3@{ub} Multiply operands 1 and 2, which have mode @{b}HImode@{ub}, and store a @{b}SImode@{ub} product in operand 0. @{b}mulqihi3@{ub}, @{b}mulsidi3@{ub} Similar widening-multiplication instructions of other widths. @{b}umulqihi3@{ub}, @{b}umulhisi3@{ub}, @{b}umulsidi3@{ub} Similar widening-multiplication instructions that do unsigned multiplication. @{b}mul@{I}M@{ui}3_highpart@{ub} Perform a signed multiplication of operands 1 and 2, which have mode @{I}M@{ui}, and store the most significant half of the product in operand 0. The least significant half of the product is discarded. @{b}umul@{I}M@{ui}3_highpart@{ub} Similar, but the multiplication is unsigned. @{b}divmod@{I}M@{ui}4@{ub} Signed division that produces both a quotient and a remainder. Operand 1 is divided by operand 2 to produce a quotient stored in operand 0 and a remainder stored in operand 3. For machines with an instruction that produces both a quotient and a remainder, provide a pattern for @{b}divmod@{I}M@{ui}4@{ub} but do not provide patterns for @{b}div@{I}M@{ui}3@{ub} and @{b}mod@{I}M@{ui}3@{ub}. This allows optimization in the relatively common case when both the quotient and remainder are computed. If an instruction that just produces a quotient or just a remainder exists and is more efficient than the instruction that produces both, write the output routine of @{b}divmod@{I}M@{ui}4@{ub} to call @{b}find_reg_note@{ub} and look for a @{b}REG_UNUSED@{ub} note on the quotient or remainder and generate the appropriate instruction. @{b}udivmod@{I}M@{ui}4@{ub} Similar, but does unsigned division. @{b}ashl@{I}M@{ui}3@{ub} Arithmetic-shift operand 1 left by a number of bits specified by operand 2, and store the result in operand 0. Here @{I}M@{ui} is the mode of operand 0 and operand 1; operand 2's mode is specified by the instruction pattern, and the compiler will convert the operand to that mode before generating the instruction. @{b}ashr@{I}M@{ui}3@{ub}, @{b}lshr@{I}M@{ui}3@{ub}, @{b}rotl@{I}M@{ui}3@{ub}, @{b}rotr@{I}M@{ui}3@{ub} Other shift and rotate instructions, analogous to the @{b}ashl@{I}M@{ui}3@{ub} instructions. @{b}neg@{I}M@{ui}2@{ub} Negate operand 1 and store the result in operand 0. @{b}abs@{I}M@{ui}2@{ub} Store the absolute value of operand 1 into operand 0. @{b}sqrt@{I}M@{ui}2@{ub} Store the square root of operand 1 into operand 0. The @{b}sqrt@{ub} built-in function of C always uses the mode which corresponds to the C data type @{b}double@{ub}. @{b}ffs@{I}M@{ui}2@{ub} Store into operand 0 one plus the index of the least significant 1-bit of operand 1. If operand 1 is zero, store zero. @{I}M@{ui} is the mode of operand 0; operand 1's mode is specified by the instruction pattern, and the compiler will convert the operand to that mode before generating the instruction. The @{b}ffs@{ub} built-in function of C always uses the mode which corresponds to the C data type @{b}int@{ub}. @{b}one_cmpl@{I}M@{ui}2@{ub} Store the bitwise-complement of operand 1 into operand 0. @{b}cmp@{I}M@{ui}@{ub} Compare operand 0 and operand 1, and set the condition codes. The RTL pattern should look like this: (set (cc0) (compare (match_operand:@{I}M@{ui} 0 ...) (match_operand:@{I}M@{ui} 1 ...))) @{b}tst@{I}M@{ui}@{ub} Compare operand 0 against zero, and set the condition codes. The RTL pattern should look like this: (set (cc0) (match_operand:@{I}M@{ui} 0 ...)) @{b}tst@{I}M@{ui}@{ub} patterns should not be defined for machines that do not use @{b}(cc0)@{ub}. Doing so would confuse the optimizer since it would no longer be clear which @{b}set@{ub} operations were comparisons. The @{b}cmp@{I}M@{ui}@{ub} patterns should be used instead. @{b}movstr@{I}M@{ui}@{ub} Block move instruction. The addresses of the destination and source strings are the first two operands, and both are in mode @{b}Pmode@{ub}. The number of bytes to move is the third operand, in mode @{I}M@{ui}. The fourth operand is the known shared alignment of the source and destination, in the form of a @{b}const_int@{ub} rtx. Thus, if the compiler knows that both source and destination are word-aligned, it may provide the value 4 for this operand. These patterns need not give special consideration to the possibility that the source and destination strings might overlap. @{b}cmpstr@{I}M@{ui}@{ub} Block compare instruction, with five operands. Operand 0 is the output; it has mode @{I}M@{ui}. The remaining four operands are like the operands of @{b}movstr@{I}M@{ui}@{ub}. The two memory blocks specified are compared byte by byte in lexicographic order. The effect of the instruction is to store a value in operand 0 whose sign indicates the result of the comparison. Compute the length of a string, with three operands. Operand 0 is the result (of mode @{I}M@{ui}), operand 1 is a @{b}mem@{ub} referring to the first character of the string, operand 2 is the character to search for (normally zero), and operand 3 is a constant describing the known alignment of the beginning of the string. @{b}float@{I}M@{ui} @{I}N@{ui}2@{ub} Convert signed integer operand 1 (valid for fixed point mode @{I}M@{ui}) to floating point mode @{I}N@{ui} and store in operand 0 (which has mode @{I}N@{ui}). @{b}floatuns@{I}M@{ui} @{I}N@{ui}2@{ub} Convert unsigned integer operand 1 (valid for fixed point mode @{I}M@{ui}) to floating point mode @{I}N@{ui} and store in operand 0 (which has mode @{I}N@{ui}). @{b}fix@{I}M@{ui} @{I}N@{ui}2@{ub} Convert operand 1 (valid for floating point mode @{I}M@{ui}) to fixed point mode @{I}N@{ui} as a signed number and store in operand 0 (which has mode @{I}N@{ui}). This instruction's result is defined only when the value of operand 1 is an integer. @{b}fixuns@{I}M@{ui} @{I}N@{ui}2@{ub} Convert operand 1 (valid for floating point mode @{I}M@{ui}) to fixed point mode @{I}N@{ui} as an unsigned number and store in operand 0 (which has mode @{I}N@{ui}). This instruction's result is defined only when the value of operand 1 is an integer. @{b}ftrunc@{I}M@{ui}2@{ub} Convert operand 1 (valid for floating point mode @{I}M@{ui}) to an integer value, still represented in floating point mode @{I}M@{ui}, and store it in operand 0 (valid for floating point mode @{I}M@{ui}). @{b}fix_trunc@{I}M@{ui} @{I}N@{ui}2@{ub} Like @{b}fix@{I}M@{ui} @{I}N@{ui}2@{ub} but works for any floating point value of mode @{I}M@{ui} by converting the value to an integer. @{b}fixuns_trunc@{I}M@{ui} @{I}N@{ui}2@{ub} Like @{b}fixuns@{I}M@{ui} @{I}N@{ui}2@{ub} but works for any floating point value of mode @{I}M@{ui} by converting the value to an integer. @{b}trunc@{I}M@{ui} @{I}N@{ui}@{ub} Truncate operand 1 (valid for mode @{I}M@{ui}) to mode @{I}N@{ui} and store in operand 0 (which has mode @{I}N@{ui}). Both modes must be fixed point or both floating point. @{b}extend@{I}M@{ui} @{I}N@{ui}@{ub} Sign-extend operand 1 (valid for mode @{I}M@{ui}) to mode @{I}N@{ui} and store in operand 0 (which has mode @{I}N@{ui}). Both modes must be fixed point or both floating point. @{b}zero_extend@{I}M@{ui} @{I}N@{ui}@{ub} Zero-extend operand 1 (valid for mode @{I}M@{ui}) to mode @{I}N@{ui} and store in operand 0 (which has mode @{I}N@{ui}). Both modes must be fixed point. @{b}extv@{ub} Extract a bit field from operand 1 (a register or memory operand), where operand 2 specifies the width in bits and operand 3 the starting bit, and store it in operand 0. Operand 0 must have mode @{b}word_mode@{ub}. Operand 1 may have mode @{b}byte_mode@{ub} or @{b}word_mode@{ub}; often @{b}word_mode@{ub} is allowed only for registers. Operands 2 and 3 must be valid for @{b}word_mode@{ub}. The RTL generation pass generates this instruction only with constants for operands 2 and 3. The bit-field value is sign-extended to a full word integer before it is stored in operand 0. @{b}extzv@{ub} Like @{b}extv@{ub} except that the bit-field value is zero-extended. @{b}insv@{ub} Store operand 3 (which must be valid for @{b}word_mode@{ub}) into a bit field in operand 0, where operand 1 specifies the width in bits and operand 2 the starting bit. Operand 0 may have mode @{b}byte_mode@{ub} or @{b}word_mode@{ub}; often @{b}word_mode@{ub} is allowed only for registers. Operands 1 and 2 must be valid for @{b}word_mode@{ub}. The RTL generation pass generates this instruction only with constants for operands 1 and 2. @{b}mov@{I}MODE@{ui}cc@{ub} Conditionally move operand 2 or operand 3 into operand 0 according to the comparison in operand 1. If the comparison is true, operand 2 is moved into operand 0, otherwise operand 3 is moved. The mode of the operands being compared need not be the same as the operands being moved. Some machines, sparc64 for example, have instructions that conditionally move an integer value based on the floating point condition codes and vice versa. If the machine does not have conditional move instructions, do not define these patterns. @{b} s@{I}COND@{ui}@{ub} Store zero or nonzero in the operand according to the condition codes. Value stored is nonzero iff the condition @{I}COND@{ui} is true. @{I}COND@{ui} is the name of a comparison operation expression code, such as @{b}eq@{ub}, @{b}lt@{ub} or @{b}leu@{ub}. You specify the mode that the operand must have when you write the @{b}match_operand@{ub} expression. The compiler automatically sees which mode you have used and supplies an operand of that mode. The value stored for a true condition must have 1 as its low bit, or else must be negative. Otherwise the instruction is not suitable and you should omit it from the machine description. You describe to the compiler exactly which value is stored by defining the macro @{b}STORE_FLAG_VALUE@{ub} (see @{"Misc" Link "Misc"}). If a description cannot be found that can be used for all the @{b} s@{I}COND@{ui}@{ub} patterns, you should omit those operations from the machine description. These operations may fail, but should do so only in relatively uncommon cases; if they would fail for common cases involving integer comparisons, it is best to omit these patterns. If these operations are omitted, the compiler will usually generate code that copies the constant one to the target and branches around an assignment of zero to the target. If this code is more efficient than the potential instructions used for the @{b} s@{I}COND@{ui}@{ub} pattern followed by those required to convert the result into a 1 or a zero in @{b}SImode@{ub}, you should omit the @{b} s@{I}COND@{ui}@{ub} operations from the machine description. @{b} b@{I}COND@{ui}@{ub} Conditional branch instruction. Operand 0 is a @{b}label_ref@{ub} that refers to the label to jump to. Jump if the condition codes meet condition @{I}COND@{ui}. Some machines do not follow the model assumed here where a comparison instruction is followed by a conditional branch instruction. In that case, the @{b}cmp@{I}M@{ui}@{ub} (and @{b}tst@{I}M@{ui}@{ub}) patterns should simply store the operands away and generate all the required insns in a @{b}define_expand@{ub} (see @{"Expander Definitions" Link "Expander Definitions"}) for the conditional branch operations. All calls to expand @{b} b@{I}COND@{ui}@{ub} patterns are immediately preceded by calls to expand either a @{b}cmp@{I}M@{ui}@{ub} pattern or a @{b}tst@{I}M@{ui}@{ub} pattern. Machines that use a pseudo register for the condition code value, or where the mode used for the comparison depends on the condition being tested, should also use the above mechanism. See @{"Jump Patterns" Link "Jump Patterns"} The above discussion also applies to the @{b}mov@{I}MODE@{ui}cc@{ub} and @{b} s@{I}COND@{ui}@{ub} patterns. @{b}call@{ub} Subroutine call instruction returning no value. Operand 0 is the function to call; operand 1 is the number of bytes of arguments pushed (in mode @{b}SImode@{ub}, except it is normally a @{b}const_int@{ub}); operand 2 is the number of registers used as operands. On most machines, operand 2 is not actually stored into the RTL pattern. It is supplied for the sake of some RISC machines which need to put this information into the assembler code; they can put it in the RTL instead of operand 1. Operand 0 should be a @{b}mem@{ub} RTX whose address is the address of the function. Note, however, that this address can be a @{b}symbol_ref@{ub} expression even if it would not be a legitimate memory address on the target machine. If it is also not a valid argument for a call instruction, the pattern for this operation should be a @{b}define_expand@{ub} (see @{"Expander Definitions" Link "Expander Definitions"}) that places the address into a register and uses that register in the call instruction. @{b}call_value@{ub} Subroutine call instruction returning a value. Operand 0 is the hard register in which the value is returned. There are three more operands, the same as the three operands of the @{b}call@{ub} instruction (but with numbers increased by one). Subroutines that return @{b}BLKmode@{ub} objects use the @{b}call@{ub} insn. @{b}call_pop@{ub}, @{b}call_value_pop@{ub} Similar to @{b}call@{ub} and @{b}call_value@{ub}, except used if defined and if @{b}RETURN_POPS_ARGS@{ub} is non-zero. They should emit a @{b}parallel@{ub} that contains both the function call and a @{b}set@{ub} to indicate the adjustment made to the frame pointer. For machines where @{b}RETURN_POPS_ARGS@{ub} can be non-zero, the use of these patterns increases the number of functions for which the frame pointer can be eliminated, if desired. @{b}untyped_call@{ub} Subroutine call instruction returning a value of any type. Operand 0 is the function to call; operand 1 is a memory location where the result of calling the function is to be stored; operand 2 is a @{b}parallel@{ub} expression where each element is a @{b}set@{ub} expression that indicates the saving of a function return value into the result block. This instruction pattern should be defined to support @{b}__builtin_apply@{ub} on machines where special instructions are needed to call a subroutine with arbitrary arguments or to save the value returned. This instruction pattern is required on machines that have multiple registers that can hold a return value (i.e. @{b}FUNCTION_VALUE_REGNO_P@{ub} is true for more than one register). @{b}return@{ub} Subroutine return instruction. This instruction pattern name should be defined only if a single instruction can do all the work of returning from a function. Like the @{b}mov@{I}M@{ui}@{ub} patterns, this pattern is also used after the RTL generation phase. In this case it is to support machines where multiple instructions are usually needed to return from a function, but some class of functions only requires one instruction to implement a return. Normally, the applicable functions are those which do not need to save any registers or allocate stack space. For such machines, the condition specified in this pattern should only be true when @{b}reload_completed@{ub} is non-zero and the function's epilogue would only be a single instruction. For machines with register windows, the routine @{b}leaf_function_p@{ub} may be used to determine if a register window push is required. Machines that have conditional return instructions should define patterns such as (define_insn "" [(set (pc) (if_then_else (match_operator 0 "comparison_operator" [(cc0) (const_int 0)]) (return) (pc)))] "@{I}CONDITION@{ui}" "...") where @{I}CONDITION@{ui} would normally be the same condition specified on the named @{b}return@{ub} pattern. @{b}untyped_return@{ub} Untyped subroutine return instruction. This instruction pattern should be defined to support @{b}__builtin_return@{ub} on machines where special instructions are needed to return a value of any type. Operand 0 is a memory location where the result of calling a function with @{b}__builtin_apply@{ub} is stored; operand 1 is a @{b}parallel@{ub} expression where each element is a @{b}set@{ub} expression that indicates the restoring of a function return value from the result block. @{b}nop@{ub} No-op instruction. This instruction pattern name should always be defined to output a no-op in assembler code. @{b}(const_int 0)@{ub} will do as an RTL pattern. @{b}indirect_jump@{ub} An instruction to jump to an address which is operand zero. This pattern name is mandatory on all machines. @{b}casesi@{ub} Instruction to jump through a dispatch table, including bounds checking. This instruction takes five operands: 1. The index to dispatch on, which has mode @{b}SImode@{ub}. 2. The lower bound for indices in the table, an integer constant. 3. The total range of indices in the table--the largest index minus the smallest one (both inclusive). 4. A label that precedes the table itself. 5. A label to jump to if the index has a value outside the bounds. (If the machine-description macro @{b}CASE_DROPS_THROUGH@{ub} is defined, then an out-of-bounds index drops through to the code following the jump table instead of jumping to this label. In that case, this label is not actually used by the @{b}casesi@{ub} instruction, but it is always provided as an operand.) The table is a @{b}addr_vec@{ub} or @{b}addr_diff_vec@{ub} inside of a @{b}jump_insn@{ub}. The number of elements in the table is one plus the difference between the upper bound and the lower bound. @{b}tablejump@{ub} Instruction to jump to a variable address. This is a low-level capability which can be used to implement a dispatch table when there is no @{b}casesi@{ub} pattern. This pattern requires two operands: the address or offset, and a label which should immediately precede the jump table. If the macro @{b}CASE_VECTOR_PC_RELATIVE@{ub} is defined then the first operand is an offset which counts from the address of the table; otherwise, it is an absolute address to jump to. In either case, the first operand has mode @{b}Pmode@{ub}. The @{b}tablejump@{ub} insn is always the last insn before the jump table it uses. Its assembler code normally has no need to use the second operand, but you should incorporate it in the RTL pattern so that the jump optimizer will not delete the table as unreachable code. @{b}save_stack_block@{ub} @{b}save_stack_function@{ub} @{b}save_stack_nonlocal@{ub} @{b}restore_stack_block@{ub} @{b}restore_stack_function@{ub} @{b}restore_stack_nonlocal@{ub} Most machines save and restore the stack pointer by copying it to or from an object of mode @{b}Pmode@{ub}. Do not define these patterns on such machines. Some machines require special handling for stack pointer saves and restores. On those machines, define the patterns corresponding to the non-standard cases by using a @{b}define_expand@{ub} (see @{"Expander Definitions" Link "Expander Definitions"}) that produces the required insns. The three types of saves and restores are: 1. @{b}save_stack_block@{ub} saves the stack pointer at the start of a block that allocates a variable-sized object, and @{b}restore_stack_block@{ub} restores the stack pointer when the block is exited. 2. @{b}save_stack_function@{ub} and @{b}restore_stack_function@{ub} do a similar job for the outermost block of a function and are used when the function allocates variable-sized objects or calls @{b}alloca@{ub}. Only the epilogue uses the restored stack pointer, allowing a simpler save or restore sequence on some machines. 3. @{b}save_stack_nonlocal@{ub} is used in functions that contain labels branched to by nested functions. It saves the stack pointer in such a way that the inner function can use @{b}restore_stack_nonlocal@{ub} to restore the stack pointer. The compiler generates code to restore the frame and argument pointer registers, but some machines require saving and restoring additional data such as register window information or stack backchains. Place insns in these patterns to save and restore any such required data. When saving the stack pointer, operand 0 is the save area and operand 1 is the stack pointer. The mode used to allocate the save area is the mode of operand 0. You must specify an integral mode, or @{b}VOIDmode@{ub} if no save area is needed for a particular type of save (either because no save is needed or because a machine-specific save area can be used). Operand 0 is the stack pointer and operand 1 is the save area for restore operations. If @{b}save_stack_block@{ub} is defined, operand 0 must not be @{b}VOIDmode@{ub} since these saves can be arbitrarily nested. A save area is a @{b}mem@{ub} that is at a constant offset from @{b}virtual_stack_vars_rtx@{ub} when the stack pointer is saved for use by nonlocal gotos and a @{b}reg@{ub} in the other two cases. @{b}allocate_stack@{ub} Subtract (or add if @{b}STACK_GROWS_DOWNWARD@{ub} is undefined) operand 0 from the stack pointer to create space for dynamically allocated data. Do not define this pattern if all that must be done is the subtraction. Some machines require other operations such as stack probes or maintaining the back chain. Define this pattern to emit those operations in addition to updating the stack pointer. @EndNode @Node "Pattern Ordering" "gcc.guide/Pattern Ordering" @Next "Dependent Patterns" @Prev "Standard Names" @Toc "Machine Desc" When the Order of Patterns Matters ================================== Sometimes an insn can match more than one instruction pattern. Then the pattern that appears first in the machine description is the one used. Therefore, more specific patterns (patterns that will match fewer things) and faster instructions (those that will produce better code when they do match) should usually go first in the description. In some cases the effect of ordering the patterns can be used to hide a pattern when it is not valid. For example, the 68000 has an instruction for converting a fullword to floating point and another for converting a byte to floating point. An instruction converting an integer to floating point could match either one. We put the pattern to convert the fullword first to make sure that one will be used rather than the other. (Otherwise a large integer might be generated as a single-byte immediate quantity, which would not work.) Instead of using this pattern ordering it would be possible to make the pattern for convert-a-byte smart enough to deal properly with any constant value. @EndNode @Node "Dependent Patterns" "gcc.guide/Dependent Patterns" @Next "Jump Patterns" @Prev "Pattern Ordering" @Toc "Machine Desc" Interdependence of Patterns =========================== Every machine description must have a named pattern for each of the conditional branch names @{b} b@{I}COND@{ui}@{ub}. The recognition template must always have the form (set (pc) (if_then_else (@{I}COND@{ui} (cc0) (const_int 0)) (label_ref (match_operand 0 "" "")) (pc))) In addition, every machine description must have an anonymous pattern for each of the possible reverse-conditional branches. Their templates look like (set (pc) (if_then_else (@{I}COND@{ui} (cc0) (const_int 0)) (pc) (label_ref (match_operand 0 "" "")))) They are necessary because jump optimization can turn direct-conditional branches into reverse-conditional branches. It is often convenient to use the @{b}match_operator@{ub} construct to reduce the number of patterns that must be specified for branches. For example, (define_insn "" [(set (pc) (if_then_else (match_operator 0 "comparison_operator" [(cc0) (const_int 0)]) (pc) (label_ref (match_operand 1 "" ""))))] "@{I}CONDITION@{ui}" "...") In some cases machines support instructions identical except for the machine mode of one or more operands. For example, there may be "sign-extend halfword" and "sign-extend byte" instructions whose patterns are (set (match_operand:SI 0 ...) (extend:SI (match_operand:HI 1 ...))) (set (match_operand:SI 0 ...) (extend:SI (match_operand:QI 1 ...))) Constant integers do not specify a machine mode, so an instruction to extend a constant value could match either pattern. The pattern it actually will match is the one that appears first in the file. For correct results, this must be the one for the widest possible mode (@{b}HImode@{ub}, here). If the pattern matches the @{b}QImode@{ub} instruction, the results will be incorrect if the constant value does not actually fit that mode. Such instructions to extend constants are rarely generated because they are optimized away, but they do occasionally happen in nonoptimized compilations. If a constraint in a pattern allows a constant, the reload pass may replace a register with a constant permitted by the constraint in some cases. Similarly for memory references. Because of this substitution, you should not provide separate patterns for increment and decrement instructions. Instead, they should be generated from the same pattern that supports register-register add insns by examining the operands and generating the appropriate machine instruction. @EndNode @Node "Jump Patterns" "gcc.guide/Jump Patterns" @Next "Insn Canonicalizations" @Prev "Dependent Patterns" @Toc "Machine Desc" Defining Jump Instruction Patterns ================================== For most machines, GNU CC assumes that the machine has a condition code. A comparison insn sets the condition code, recording the results of both signed and unsigned comparison of the given operands. A separate branch insn tests the condition code and branches or not according its value. The branch insns come in distinct signed and unsigned flavors. Many common machines, such as the Vax, the 68000 and the 32000, work this way. Some machines have distinct signed and unsigned compare instructions, and only one set of conditional branch instructions. The easiest way to handle these machines is to treat them just like the others until the final stage where assembly code is written. At this time, when outputting code for the compare instruction, peek ahead at the following branch using @{b}next_cc0_user (insn)@{ub}. (The variable @{b}insn@{ub} refers to the insn being output, in the output-writing code in an instruction pattern.) If the RTL says that is an unsigned branch, output an unsigned compare; otherwise output a signed compare. When the branch itself is output, you can treat signed and unsigned branches identically. The reason you can do this is that GNU CC always generates a pair of consecutive RTL insns, possibly separated by @{b}note@{ub} insns, one to set the condition code and one to test it, and keeps the pair inviolate until the end. To go with this technique, you must define the machine-description macro @{b}NOTICE_UPDATE_CC@{ub} to do @{b}CC_STATUS_INIT@{ub}; in other words, no compare instruction is superfluous. Some machines have compare-and-branch instructions and no condition code. A similar technique works for them. When it is time to "output" a compare instruction, record its operands in two static variables. When outputting the branch-on-condition-code instruction that follows, actually output a compare-and-branch instruction that uses the remembered operands. It also works to define patterns for compare-and-branch instructions. In optimizing compilation, the pair of compare and branch instructions will be combined according to these patterns. But this does not happen if optimization is not requested. So you must use one of the solutions above in addition to any special patterns you define. In many RISC machines, most instructions do not affect the condition code and there may not even be a separate condition code register. On these machines, the restriction that the definition and use of the condition code be adjacent insns is not necessary and can prevent important optimizations. For example, on the IBM RS/6000, there is a delay for taken branches unless the condition code register is set three instructions earlier than the conditional branch. The instruction scheduler cannot perform this optimization if it is not permitted to separate the definition and use of the condition code register. On these machines, do not use @{b}(cc0)@{ub}, but instead use a register to represent the condition code. If there is a specific condition code register in the machine, use a hard register. If the condition code or comparison result can be placed in any general register, or if there are multiple condition registers, use a pseudo register. On some machines, the type of branch instruction generated may depend on the way the condition code was produced; for example, on the 68k and Sparc, setting the condition code directly from an add or subtract instruction does not clear the overflow bit the way that a test instruction does, so a different branch instruction must be used for some conditional branches. For machines that use @{b}(cc0)@{ub}, the set and use of the condition code must be adjacent (separated only by @{b}note@{ub} insns) allowing flags in @{b}cc_status@{ub} to be used. (See @{"Condition Code" Link "Condition Code"}.) Also, the comparison and branch insns can be located from each other by using the functions @{b}prev_cc0_setter@{ub} and @{b}next_cc0_user@{ub}. However, this is not true on machines that do not use @{b}(cc0)@{ub}. On those machines, no assumptions can be made about the adjacency of the compare and branch insns and the above methods cannot be used. Instead, we use the machine mode of the condition code register to record different formats of the condition code register. Registers used to store the condition code value should have a mode that is in class @{b}MODE_CC@{ub}. Normally, it will be @{b}CCmode@{ub}. If additional modes are required (as for the add example mentioned above in the Sparc), define the macro @{b}EXTRA_CC_MODES@{ub} to list the additional modes required (see @{"Condition Code" Link "Condition Code"}). Also define @{b}EXTRA_CC_NAMES@{ub} to list the names of those modes and @{b}SELECT_CC_MODE@{ub} to choose a mode given an operand of a compare. If it is known during RTL generation that a different mode will be required (for example, if the machine has separate compare instructions for signed and unsigned quantities, like most IBM processors), they can be specified at that time. If the cases that require different modes would be made by instruction combination, the macro @{b}SELECT_CC_MODE@{ub} determines which machine mode should be used for the comparison result. The patterns should be written using that mode. To support the case of the add on the Sparc discussed above, we have the pattern (define_insn "" [(set (reg:CC_NOOV 0) (compare:CC_NOOV (plus:SI (match_operand:SI 0 "register_operand" "%r") (match_operand:SI 1 "arith_operand" "rI")) (const_int 0)))] "" "...") The @{b}SELECT_CC_MODE@{ub} macro on the Sparc returns @{b}CC_NOOVmode@{ub} for comparisons whose argument is a @{b}plus@{ub}. @EndNode @Node "Insn Canonicalizations" "gcc.guide/Insn Canonicalizations" @Next "Peephole Definitions" @Prev "Jump Patterns" @Toc "Machine Desc" Canonicalization of Instructions ================================ There are often cases where multiple RTL expressions could represent an operation performed by a single machine instruction. This situation is most commonly encountered with logical, branch, and multiply-accumulate instructions. In such cases, the compiler attempts to convert these multiple RTL expressions into a single canonical form to reduce the number of insn patterns required. In addition to algebraic simplifications, following canonicalizations are performed: @{b}*@{ub} For commutative and comparison operators, a constant is always made the second operand. If a machine only supports a constant as the second operand, only patterns that match a constant in the second operand need be supplied. For these operators, if only one operand is a @{b}neg@{ub}, @{b}not@{ub}, @{b}mult@{ub}, @{b}plus@{ub}, or @{b}minus@{ub} expression, it will be the first operand. @{b}*@{ub} For the @{b}compare@{ub} operator, a constant is always the second operand on machines where @{b}cc0@{ub} is used (see @{"Jump Patterns" Link "Jump Patterns"}). On other machines, there are rare cases where the compiler might want to construct a @{b}compare@{ub} with a constant as the first operand. However, these cases are not common enough for it to be worthwhile to provide a pattern matching a constant as the first operand unless the machine actually has such an instruction. An operand of @{b}neg@{ub}, @{b}not@{ub}, @{b}mult@{ub}, @{b}plus@{ub}, or @{b}minus@{ub} is made the first operand under the same conditions as above. @{b}*@{ub} @{b}(minus @{I}X@{ui} (const_int @{I}N@{ui}))@{ub} is converted to @{b}(plus @{I}X@{ui} (const_int @{I}-N@{ui}))@{ub}. @{b}*@{ub} Within address computations (i.e., inside @{b}mem@{ub}), a left shift is converted into the appropriate multiplication by a power of two. De`Morgan's Law is used to move bitwise negation inside a bitwise logical-and or logical-or operation. If this results in only one operand being a @{b}not@{ub} expression, it will be the first one. A machine that has an instruction that performs a bitwise logical-and of one operand with the bitwise negation of the other should specify the pattern for that instruction as (define_insn "" [(set (match_operand:@{I}M@{ui} 0 ...) (and:@{I}M@{ui} (not:@{I}M@{ui} (match_operand:@{I}M@{ui} 1 ...)) (match_operand:@{I}M@{ui} 2 ...)))] "..." "...") Similarly, a pattern for a "NAND" instruction should be written (define_insn "" [(set (match_operand:@{I}M@{ui} 0 ...) (ior:@{I}M@{ui} (not:@{I}M@{ui} (match_operand:@{I}M@{ui} 1 ...)) (not:@{I}M@{ui} (match_operand:@{I}M@{ui} 2 ...))))] "..." "...") In both cases, it is not necessary to include patterns for the many logically equivalent RTL expressions. @{b}*@{ub} The only possible RTL expressions involving both bitwise exclusive-or and bitwise negation are @{b}(xor:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui})@{ub} and @{b}(not:@{I}M@{ui} (xor:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui}))@{ub}. @{b}*@{ub} The sum of three items, one of which is a constant, will only appear in the form (plus:@{I}M@{ui} (plus:@{I}M@{ui} @{I}X@{ui} @{I}Y@{ui}) @{I}CONSTANT@{ui}) @{b}*@{ub} On machines that do not use @{b}cc0@{ub}, @{b}(compare @{I}X@{ui} (const_int 0))@{ub} will be converted to @{I}X@{ui}. @{b}*@{ub} Equality comparisons of a group of bits (usually a single bit) with zero will be written using @{b}zero_extract@{ub} rather than the equivalent @{b}and@{ub} or @{b}sign_extract@{ub} operations. @EndNode @Node "Peephole Definitions" "gcc.guide/Peephole Definitions" @Next "Expander Definitions" @Prev "Insn Canonicalizations" @Toc "Machine Desc" Machine-Specific Peephole Optimizers ==================================== In addition to instruction patterns the @{b}md@{ub} file may contain definitions of machine-specific peephole optimizations. The combiner does not notice certain peephole optimizations when the data flow in the program does not suggest that it should try them. For example, sometimes two consecutive insns related in purpose can be combined even though the second one does not appear to use a register computed in the first one. A machine-specific peephole optimizer can detect such opportunities. A definition looks like this: (define_peephole [@{I}INSN-PATTERN-1@{ui} @{I}INSN-PATTERN-2@{ui} ...] "@{I}CONDITION@{ui}" "@{I}TEMPLATE@{ui}" "@{I}OPTIONAL INSN-ATTRIBUTES@{ui}") The last string operand may be omitted if you are not using any machine-specific information in this machine description. If present, it must obey the same rules as in a @{b}define_insn@{ub}. In this skeleton, @{I}INSN-PATTERN-1@{ui} and so on are patterns to match consecutive insns. The optimization applies to a sequence of insns when @{I}INSN-PATTERN-1@{ui} matches the first one, @{I}INSN-PATTERN-2@{ui} matches the next, and so on. Each of the insns matched by a peephole must also match a @{b}define_insn@{ub}. Peepholes are checked only at the last stage just before code generation, and only optionally. Therefore, any insn which would match a peephole but no @{b}define_insn@{ub} will cause a crash in code generation in an unoptimized compilation, or at various optimization stages. The operands of the insns are matched with @{b}match_operands@{ub}, @{b}match_operator@{ub}, and @{b}match_dup@{ub}, as usual. What is not usual is that the operand numbers apply to all the insn patterns in the definition. So, you can check for identical operands in two insns by using @{b}match_operand@{ub} in one insn and @{b}match_dup@{ub} in the other. The operand constraints used in @{b}match_operand@{ub} patterns do not have any direct effect on the applicability of the peephole, but they will be validated afterward, so make sure your constraints are general enough to apply whenever the peephole matches. If the peephole matches but the constraints are not satisfied, the compiler will crash. It is safe to omit constraints in all the operands of the peephole; or you can write constraints which serve as a double-check on the criteria previously tested. Once a sequence of insns matches the patterns, the @{I}CONDITION@{ui} is checked. This is a C expression which makes the final decision whether to perform the optimization (we do so if the expression is nonzero). If @{I}CONDITION@{ui} is omitted (in other words, the string is empty) then the optimization is applied to every sequence of insns that matches the patterns. The defined peephole optimizations are applied after register allocation is complete. Therefore, the peephole definition can check which operands have ended up in which kinds of registers, just by looking at the operands. The way to refer to the operands in @{I}CONDITION@{ui} is to write @{b}operands[@{I}I@{ui}]@{ub} for operand number @{I}I@{ui} (as matched by @{b}(match_operand @{I}I@{ui} ...)@{ub}). Use the variable @{b}insn@{ub} to refer to the last of the insns being matched; use @{b}prev_active_insn@{ub} to find the preceding insns. When optimizing computations with intermediate results, you can use @{I}CONDITION@{ui} to match only when the intermediate results are not used elsewhere. Use the C expression @{b}dead_or_set_p (@{I}INSN@{ui}, @{I}OP@{ui})@{ub}, where @{I}INSN@{ui} is the insn in which you expect the value to be used for the last time (from the value of @{b}insn@{ub}, together with use of @{b}prev_nonnote_insn@{ub}), and @{I}OP@{ui} is the intermediate value (from @{b}operands[@{I}I@{ui}]@{ub}). Applying the optimization means replacing the sequence of insns with one new insn. The @{I}TEMPLATE@{ui} controls ultimate output of assembler code for this combined insn. It works exactly like the template of a @{b}define_insn@{ub}. Operand numbers in this template are the same ones used in matching the original sequence of insns. The result of a defined peephole optimizer does not need to match any of the insn patterns in the machine description; it does not even have an opportunity to match them. The peephole optimizer definition itself serves as the insn pattern to control how the insn is output. Defined peephole optimizers are run as assembler code is being output, so the insns they produce are never combined or rearranged in any way. Here is an example, taken from the 68000 machine description: (define_peephole [(set (reg:SI 15) (plus:SI (reg:SI 15) (const_int 4))) (set (match_operand:DF 0 "register_operand" "=f") (match_operand:DF 1 "register_operand" "ad"))] "FP_REG_P (operands[0]) && ! FP_REG_P (operands[1])" "* { rtx xoperands[2]; xoperands[1] = gen_rtx (REG, SImode, REGNO (operands[1]) + 1); #ifdef MOTOROLA output_asm_insn (\\"move.l %1,(sp)\\", xoperands); output_asm_insn (\\"move.l %1,-(sp)\\", operands); return \\"fmove.d (sp)+,%0\\"; #else output_asm_insn (\\"movel %1,sp@\\", xoperands); output_asm_insn (\\"movel %1,sp@-\\", operands); return \\"fmoved sp@+,%0\\"; #endif } ") The effect of this optimization is to change jbsr _foobar addql #4,sp movel d1,sp@- movel d0,sp@- fmoved sp@+,fp0 into jbsr _foobar movel d1,sp@ movel d0,sp@- fmoved sp@+,fp0 @{I}INSN-PATTERN-1@{ui} and so on look @{i}almost@{ui} like the second operand of @{b}define_insn@{ub}. There is one important difference: the second operand of @{b}define_insn@{ub} consists of one or more RTX's enclosed in square brackets. Usually, there is only one: then the same action can be written as an element of a @{b}define_peephole@{ub}. But when there are multiple actions in a @{b}define_insn@{ub}, they are implicitly enclosed in a @{b}parallel@{ub}. Then you must explicitly write the @{b}parallel@{ub}, and the square brackets within it, in the @{b}define_peephole@{ub}. Thus, if an insn pattern looks like this, (define_insn "divmodsi4" [(set (match_operand:SI 0 "general_operand" "=d") (div:SI (match_operand:SI 1 "general_operand" "0") (match_operand:SI 2 "general_operand" "dmsK"))) (set (match_operand:SI 3 "general_operand" "=d") (mod:SI (match_dup 1) (match_dup 2)))] "TARGET_68020" "divsl%.l %2,%3:%0") then the way to mention this insn in a peephole is as follows: (define_peephole [... (parallel [(set (match_operand:SI 0 "general_operand" "=d") (div:SI (match_operand:SI 1 "general_operand" "0") (match_operand:SI 2 "general_operand" "dmsK"))) (set (match_operand:SI 3 "general_operand" "=d") (mod:SI (match_dup 1) (match_dup 2)))]) ...] ...) @EndNode @Node "Expander Definitions" "gcc.guide/Expander Definitions" @Next "Insn Splitting" @Prev "Peephole Definitions" @Toc "Machine Desc" Defining RTL Sequences for Code Generation ========================================== On some target machines, some standard pattern names for RTL generation cannot be handled with single insn, but a sequence of RTL insns can represent them. For these target machines, you can write a @{b}define_expand@{ub} to specify how to generate the sequence of RTL. A @{b}define_expand@{ub} is an RTL expression that looks almost like a @{b}define_insn@{ub}; but, unlike the latter, a @{b}define_expand@{ub} is used only for RTL generation and it can produce more than one RTL insn. A @{b}define_expand@{ub} RTX has four operands: @{b}*@{ub} The name. Each @{b}define_expand@{ub} must have a name, since the only use for it is to refer to it by name. @{b}*@{ub} The RTL template. This is just like the RTL template for a @{b}define_peephole@{ub} in that it is a vector of RTL expressions each being one insn. @{b}*@{ub} The condition, a string containing a C expression. This expression is used to express how the availability of this pattern depends on subclasses of target machine, selected by command-line options when GNU CC is run. This is just like the condition of a @{b}define_insn@{ub} that has a standard name. Therefore, the condition (if present) may not depend on the data in the insn being matched, but only the target-machine-type flags. The compiler needs to test these conditions during initialization in order to learn exactly which named instructions are available in a particular run. @{b}*@{ub} The preparation statements, a string containing zero or more C statements which are to be executed before RTL code is generated from the RTL template. Usually these statements prepare temporary registers for use as internal operands in the RTL template, but they can also generate RTL insns directly by calling routines such as @{b}emit_insn@{ub}, etc. Any such insns precede the ones that come from the RTL template. Every RTL insn emitted by a @{b}define_expand@{ub} must match some @{b}define_insn@{ub} in the machine description. Otherwise, the compiler will crash when trying to generate code for the insn or trying to optimize it. The RTL template, in addition to controlling generation of RTL insns, also describes the operands that need to be specified when this pattern is used. In particular, it gives a predicate for each operand. A true operand, which needs to be specified in order to generate RTL from the pattern, should be described with a @{b}match_operand@{ub} in its first occurrence in the RTL template. This enters information on the operand's predicate into the tables that record such things. GNU CC uses the information to preload the operand into a register if that is required for valid RTL code. If the operand is referred to more than once, subsequent references should use @{b}match_dup@{ub}. The RTL template may also refer to internal "operands" which are temporary registers or labels used only within the sequence made by the @{b}define_expand@{ub}. Internal operands are substituted into the RTL template with @{b}match_dup@{ub}, never with @{b}match_operand@{ub}. The values of the internal operands are not passed in as arguments by the compiler when it requests use of this pattern. Instead, they are computed within the pattern, in the preparation statements. These statements compute the values and store them into the appropriate elements of @{b}operands@{ub} so that @{b}match_dup@{ub} can find them. There are two special macros defined for use in the preparation statements: @{b}DONE@{ub} and @{b}FAIL@{ub}. Use them with a following semicolon, as a statement. @{b}DONE@{ub} Use the @{b}DONE@{ub} macro to end RTL generation for the pattern. The only RTL insns resulting from the pattern on this occasion will be those already emitted by explicit calls to @{b}emit_insn@{ub} within the preparation statements; the RTL template will not be generated. @{b}FAIL@{ub} Make the pattern fail on this occasion. When a pattern fails, it means that the pattern was not truly available. The calling routines in the compiler will try other strategies for code generation using other patterns. Failure is currently supported only for binary (addition, multiplication, shifting, etc.) and bitfield (@{b}extv@{ub}, @{b}extzv@{ub}, and @{b}insv@{ub}) operations. Here is an example, the definition of left-shift for the SPUR chip: (define_expand "ashlsi3" [(set (match_operand:SI 0 "register_operand" "") (ashift:SI (match_operand:SI 1 "register_operand" "") (match_operand:SI 2 "nonmemory_operand" "")))] "" " { if (GET_CODE (operands[2]) != CONST_INT || (unsigned) INTVAL (operands[2]) > 3) FAIL; }") This example uses @{b}define_expand@{ub} so that it can generate an RTL insn for shifting when the shift-count is in the supported range of 0 to 3 but fail in other cases where machine insns aren't available. When it fails, the compiler tries another strategy using different patterns (such as, a library call). If the compiler were able to handle nontrivial condition-strings in patterns with names, then it would be possible to use a @{b}define_insn@{ub} in that case. Here is another case (zero-extension on the 68000) which makes more use of the power of @{b}define_expand@{ub}: (define_expand "zero_extendhisi2" [(set (match_operand:SI 0 "general_operand" "") (const_int 0)) (set (strict_low_part (subreg:HI (match_dup 0) 0)) (match_operand:HI 1 "general_operand" ""))] "" "operands[1] = make_safe_from (operands[1], operands[0]);") Here two RTL insns are generated, one to clear the entire output operand and the other to copy the input operand into its low half. This sequence is incorrect if the input operand refers to [the old value of] the output operand, so the preparation statement makes sure this isn't so. The function @{b}make_safe_from@{ub} copies the @{b}operands[1]@{ub} into a temporary register if it refers to @{b}operands[0]@{ub}. It does this by emitting another RTL insn. Finally, a third example shows the use of an internal operand. Zero-extension on the SPUR chip is done by @{b}and@{ub}-ing the result against a halfword mask. But this mask cannot be represented by a @{b}const_int@{ub} because the constant value is too large to be legitimate on this machine. So it must be copied into a register with @{b}force_reg@{ub} and then the register used in the @{b}and@{ub}. (define_expand "zero_extendhisi2" [(set (match_operand:SI 0 "register_operand" "") (and:SI (subreg:SI (match_operand:HI 1 "register_operand" "") 0) (match_dup 2)))] "" "operands[2] = force_reg (SImode, gen_rtx (CONST_INT, VOIDmode, 65535)); ") @{i}Note:@{ui} If the @{b}define_expand@{ub} is used to serve a standard binary or unary arithmetic operation or a bitfield operation, then the last insn it generates must not be a @{b}code_label@{ub}, @{b}barrier@{ub} or @{b}note@{ub}. It must be an @{b}insn@{ub}, @{b}jump_insn@{ub} or @{b}call_insn@{ub}. If you don't need a real insn at the end, emit an insn to copy the result of the operation into itself. Such an insn will generate no code, but it can avoid problems in the compiler. @EndNode @Node "Insn Splitting" "gcc.guide/Insn Splitting" @Next "Insn Attributes" @Prev "Expander Definitions" @Toc "Machine Desc" Defining How to Split Instructions ================================== There are two cases where you should specify how to split a pattern into multiple insns. On machines that have instructions requiring delay slots (see @{"Delay Slots" Link "Delay Slots"}) or that have instructions whose output is not available for multiple cycles (see @{"Function Units" Link "Function Units"}), the compiler phases that optimize these cases need to be able to move insns into one-instruction delay slots. However, some insns may generate more than one machine instruction. These insns cannot be placed into a delay slot. Often you can rewrite the single insn as a list of individual insns, each corresponding to one machine instruction. The disadvantage of doing so is that it will cause the compilation to be slower and require more space. If the resulting insns are too complex, it may also suppress some optimizations. The compiler splits the insn if there is a reason to believe that it might improve instruction or delay slot scheduling. The insn combiner phase also splits putative insns. If three insns are merged into one insn with a complex expression that cannot be matched by some @{b}define_insn@{ub} pattern, the combiner phase attempts to split the complex pattern into two insns that are recognized. Usually it can break the complex pattern into two patterns by splitting out some subexpression. However, in some other cases, such as performing an addition of a large constant in two insns on a RISC machine, the way to split the addition into two insns is machine-dependent. The @{b}define_split@{ub} definition tells the compiler how to split a complex insn into several simpler insns. It looks like this: (define_split [@{I}INSN-PATTERN@{ui}] "@{I}CONDITION@{ui}" [@{I}NEW-INSN-PATTERN-1@{ui} @{I}NEW-INSN-PATTERN-2@{ui} ...] "@{I}PREPARATION STATEMENTS@{ui}") @{I}INSN-PATTERN@{ui} is a pattern that needs to be split and @{I}CONDITION@{ui} is the final condition to be tested, as in a @{b}define_insn@{ub}. When an insn matching @{I}INSN-PATTERN@{ui} and satisfying @{I}CONDITION@{ui} is found, it is replaced in the insn list with the insns given by @{I}NEW-INSN-PATTERN-1@{ui}, @{I}NEW-INSN-PATTERN-2@{ui}, etc. The @{I}PREPARATION STATEMENTS@{ui} are similar to those statements that are specified for @{b}define_expand@{ub} (see @{"Expander Definitions" Link "Expander Definitions"}) and are executed before the new RTL is generated to prepare for the generated code or emit some insns whose pattern is not fixed. Unlike those in @{b}define_expand@{ub}, however, these statements must not generate any new pseudo-registers. Once reload has completed, they also must not allocate any space in the stack frame. Patterns are matched against @{I}INSN-PATTERN@{ui} in two different circumstances. If an insn needs to be split for delay slot scheduling or insn scheduling, the insn is already known to be valid, which means that it must have been matched by some @{b}define_insn@{ub} and, if @{b}reload_completed@{ub} is non-zero, is known to satisfy the constraints of that @{b}define_insn@{ub}. In that case, the new insn patterns must also be insns that are matched by some @{b}define_insn@{ub} and, if @{b}reload_completed@{ub} is non-zero, must also satisfy the constraints of those definitions. As an example of this usage of @{b}define_split@{ub}, consider the following example from @{b}a29k.md@{ub}, which splits a @{b}sign_extend@{ub} from @{b}HImode@{ub} to @{b}SImode@{ub} into a pair of shift insns: (define_split [(set (match_operand:SI 0 "gen_reg_operand" "") (sign_extend:SI (match_operand:HI 1 "gen_reg_operand" "")))] "" [(set (match_dup 0) (ashift:SI (match_dup 1) (const_int 16))) (set (match_dup 0) (ashiftrt:SI (match_dup 0) (const_int 16)))] " { operands[1] = gen_lowpart (SImode, operands[1]); }") When the combiner phase tries to split an insn pattern, it is always the case that the pattern is @{i}not@{ui} matched by any @{b}define_insn@{ub}. The combiner pass first tries to split a single @{b}set@{ub} expression and then the same @{b}set@{ub} expression inside a @{b}parallel@{ub}, but followed by a @{b}clobber@{ub} of a pseudo-reg to use as a scratch register. In these cases, the combiner expects exactly two new insn patterns to be generated. It will verify that these patterns match some @{b}define_insn@{ub} definitions, so you need not do this test in the @{b}define_split@{ub} (of course, there is no point in writing a @{b}define_split@{ub} that will never produce insns that match). Here is an example of this use of @{b}define_split@{ub}, taken from @{b}rs6000.md@{ub}: (define_split [(set (match_operand:SI 0 "gen_reg_operand" "") (plus:SI (match_operand:SI 1 "gen_reg_operand" "") (match_operand:SI 2 "non_add_cint_operand" "")))] "" [(set (match_dup 0) (plus:SI (match_dup 1) (match_dup 3))) (set (match_dup 0) (plus:SI (match_dup 0) (match_dup 4)))] " { int low = INTVAL (operands[2]) & 0xffff; int high = (unsigned) INTVAL (operands[2]) >> 16; if (low & 0x8000) high++, low |= 0xffff0000; operands[3] = gen_rtx (CONST_INT, VOIDmode, high << 16); operands[4] = gen_rtx (CONST_INT, VOIDmode, low); }") Here the predicate @{b}non_add_cint_operand@{ub} matches any @{b}const_int@{ub} that is @{i}not@{ui} a valid operand of a single add insn. The add with the smaller displacement is written so that it can be substituted into the address of a subsequent operation. An example that uses a scratch register, from the same file, generates an equality comparison of a register and a large constant: (define_split [(set (match_operand:CC 0 "cc_reg_operand" "") (compare:CC (match_operand:SI 1 "gen_reg_operand" "") (match_operand:SI 2 "non_short_cint_operand" ""))) (clobber (match_operand:SI 3 "gen_reg_operand" ""))] "find_single_use (operands[0], insn, 0) && (GET_CODE (*find_single_use (operands[0], insn, 0)) == EQ || GET_CODE (*find_single_use (operands[0], insn, 0)) == NE)" [(set (match_dup 3) (xor:SI (match_dup 1) (match_dup 4))) (set (match_dup 0) (compare:CC (match_dup 3) (match_dup 5)))] " { /* Get the constant we are comparing against, C, and see what it looks like sign-extended to 16 bits. Then see what constant could be XOR'ed with C to get the sign-extended value. */ int c = INTVAL (operands[2]); int sextc = (c << 16) >> 16; int xorv = c ^ sextc; operands[4] = gen_rtx (CONST_INT, VOIDmode, xorv); operands[5] = gen_rtx (CONST_INT, VOIDmode, sextc); }") To avoid confusion, don't write a single @{b}define_split@{ub} that accepts some insns that match some @{b}define_insn@{ub} as well as some insns that don't. Instead, write two separate @{b}define_split@{ub} definitions, one for the insns that are valid and one for the insns that are not valid. @EndNode @Node "Insn Attributes" "gcc.guide/Insn Attributes" @Prev "Insn Splitting" @Toc "Machine Desc" Instruction Attributes ====================== In addition to describing the instruction supported by the target machine, the @{b}md@{ub} file also defines a group of @{i}attributes@{ui} and a set of values for each. Every generated insn is assigned a value for each attribute. One possible attribute would be the effect that the insn has on the machine's condition code. This attribute can then be used by @{b}NOTICE_UPDATE_CC@{ub} to track the condition codes. @{" Defining Attributes " Link "Defining Attributes"} Specifying attributes and their values. @{" Expressions " Link "Expressions"} Valid expressions for attribute values. @{" Tagging Insns " Link "Tagging Insns"} Assigning attribute values to insns. @{" Attr Example " Link "Attr Example"} An example of assigning attributes. @{" Insn Lengths " Link "Insn Lengths"} Computing the length of insns. @{" Constant Attributes " Link "Constant Attributes"} Defining attributes that are constant. @{" Delay Slots " Link "Delay Slots"} Defining delay slots required for a machine. @{" Function Units " Link "Function Units"} Specifying information for insn scheduling. @EndNode @Node "Defining Attributes" "gcc.guide/Defining Attributes" @Next "Expressions" @Toc "Insn Attributes" Defining Attributes and their Values ------------------------------------ The @{b}define_attr@{ub} expression is used to define each attribute required by the target machine. It looks like: (define_attr @{I}NAME@{ui} @{I}LIST-OF-VALUES@{ui} @{I}DEFAULT@{ui}) @{I}NAME@{ui} is a string specifying the name of the attribute being defined. @{I}LIST-OF-VALUES@{ui} is either a string that specifies a comma-separated list of values that can be assigned to the attribute, or a null string to indicate that the attribute takes numeric values. @{I}DEFAULT@{ui} is an attribute expression that gives the value of this attribute for insns that match patterns whose definition does not include an explicit value for this attribute. See @{"Attr Example" Link "Attr Example"}, for more information on the handling of defaults. See @{"Constant Attributes" Link "Constant Attributes"}, for information on attributes that do not depend on any particular insn. For each defined attribute, a number of definitions are written to the @{b}insn-attr.h@{ub} file. For cases where an explicit set of values is specified for an attribute, the following are defined: @{b}*@{ub} A @{b}#define@{ub} is written for the symbol @{b}HAVE_ATTR_@{I}NAME@{ui}@{ub}. @{b}*@{ub} An enumeral class is defined for @{b}attr_@{I}NAME@{ui}@{ub} with elements of the form @{b}@{I}UPPER-NAME@{ui}_@{I}UPPER-VALUE@{ui}@{ub} where the attribute name and value are first converted to upper case. @{b}*@{ub} A function @{b}get_attr_@{I}NAME@{ui}@{ub} is defined that is passed an insn and returns the attribute value for that insn. For example, if the following is present in the @{b}md@{ub} file: (define_attr "type" "branch,fp,load,store,arith" ...) the following lines will be written to the file @{b}insn-attr.h@{ub}. #define HAVE_ATTR_type enum attr_type {TYPE_BRANCH, TYPE_FP, TYPE_LOAD, TYPE_STORE, TYPE_ARITH}; extern enum attr_type get_attr_type (); If the attribute takes numeric values, no @{b}enum@{ub} type will be defined and the function to obtain the attribute's value will return @{b}int@{ub}. @EndNode @Node "Expressions" "gcc.guide/Expressions" @Next "Tagging Insns" @Prev "Defining Attributes" @Toc "Insn Attributes" Attribute Expressions --------------------- RTL expressions used to define attributes use the codes described above plus a few specific to attribute definitions, to be discussed below. Attribute value expressions must have one of the following forms: @{b}(const_int @{I}I@{ui})@{ub} The integer @{I}I@{ui} specifies the value of a numeric attribute. @{I}I@{ui} must be non-negative. The value of a numeric attribute can be specified either with a @{b}const_int@{ub} or as an integer represented as a string in @{b}const_string@{ub}, @{b}eq_attr@{ub} (see below), and @{b}set_attr@{ub} (see @{"Tagging Insns" Link "Tagging Insns"}) expressions. @{b}(const_string @{I}VALUE@{ui})@{ub} The string @{I}VALUE@{ui} specifies a constant attribute value. If @{I}VALUE@{ui} is specified as @{b}"*"@{ub}, it means that the default value of the attribute is to be used for the insn containing this expression. @{b}"*"@{ub} obviously cannot be used in the @{I}DEFAULT@{ui} expression of a @{b}define_attr@{ub}. If the attribute whose value is being specified is numeric, @{I}VALUE@{ui} must be a string containing a non-negative integer (normally @{b}const_int@{ub} would be used in this case). Otherwise, it must contain one of the valid values for the attribute. @{b}(if_then_else @{I}TEST@{ui} @{I}TRUE-VALUE@{ui} @{I}FALSE-VALUE@{ui})@{ub} @{I}TEST@{ui} specifies an attribute test, whose format is defined below. The value of this expression is @{I}TRUE-VALUE@{ui} if @{I}TEST@{ui} is true, otherwise it is @{I}FALSE-VALUE@{ui}. @{b}(cond [@{I}TEST1@{ui} @{I}VALUE1@{ui} ...] @{I}DEFAULT@{ui})@{ub} The first operand of this expression is a vector containing an even number of expressions and consisting of pairs of @{I}TEST@{ui} and @{I}VALUE@{ui} expressions. The value of the @{b}cond@{ub} expression is that of the @{I}VALUE@{ui} corresponding to the first true @{I}TEST@{ui} expression. If none of the @{I}TEST@{ui} expressions are true, the value of the @{b}cond@{ub} expression is that of the @{I}DEFAULT@{ui} expression. @{I}TEST@{ui} expressions can have one of the following forms: @{b}(const_int @{I}I@{ui})@{ub} This test is true if @{I}I@{ui} is non-zero and false otherwise. @{b}(not @{I}TEST@{ui})@{ub} @{b}(ior @{I}TEST1@{ui} @{I}TEST2@{ui})@{ub} @{b}(and @{I}TEST1@{ui} @{I}TEST2@{ui})@{ub} These tests are true if the indicated logical function is true. @{b}(match_operand:@{I}M@{ui} @{I}N@{ui} @{I}PRED@{ui} @{I}CONSTRAINTS@{ui})@{ub} This test is true if operand @{I}N@{ui} of the insn whose attribute value is being determined has mode @{I}M@{ui} (this part of the test is ignored if @{I}M@{ui} is @{b}VOIDmode@{ub}) and the function specified by the string @{I}PRED@{ui} returns a non-zero value when passed operand @{I}N@{ui} and mode @{I}M@{ui} (this part of the test is ignored if @{I}PRED@{ui} is the null string). The @{I}CONSTRAINTS@{ui} operand is ignored and should be the null string. @{b}(le @{I}ARITH1@{ui} @{I}ARITH2@{ui})@{ub} @{b}(leu @{I}ARITH1@{ui} @{I}ARITH2@{ui})@{ub} @{b}(lt @{I}ARITH1@{ui} @{I}ARITH2@{ui})@{ub} @{b}(ltu @{I}ARITH1@{ui} @{I}ARITH2@{ui})@{ub} @{b}(gt @{I}ARITH1@{ui} @{I}ARITH2@{ui})@{ub} @{b}(gtu @{I}ARITH1@{ui} @{I}ARITH2@{ui})@{ub} @{b}(ge @{I}ARITH1@{ui} @{I}ARITH2@{ui})@{ub} @{b}(geu @{I}ARITH1@{ui} @{I}ARITH2@{ui})@{ub} @{b}(ne @{I}ARITH1@{ui} @{I}ARITH2@{ui})@{ub} @{b}(eq @{I}ARITH1@{ui} @{I}ARITH2@{ui})@{ub} These tests are true if the indicated comparison of the two arithmetic expressions is true. Arithmetic expressions are formed with @{b}plus@{ub}, @{b}minus@{ub}, @{b}mult@{ub}, @{b}div@{ub}, @{b}mod@{ub}, @{b}abs@{ub}, @{b}neg@{ub}, @{b}and@{ub}, @{b}ior@{ub}, @{b}xor@{ub}, @{b}not@{ub}, @{b}ashift@{ub}, @{b}lshiftrt@{ub}, and @{b}ashiftrt@{ub} expressions. @{b}const_int@{ub} and @{b}symbol_ref@{ub} are always valid terms (see @{"Insn Lengths" Link "Insn Lengths"},for additional forms). @{b}symbol_ref@{ub} is a string denoting a C expression that yields an @{b}int@{ub} when evaluated by the @{b}get_attr_...@{ub} routine. It should normally be a global variable. @{b}(eq_attr @{I}NAME@{ui} @{I}VALUE@{ui})@{ub} @{I}NAME@{ui} is a string specifying the name of an attribute. @{I}VALUE@{ui} is a string that is either a valid value for attribute @{I}NAME@{ui}, a comma-separated list of values, or @{b}!@{ub} followed by a value or list. If @{I}VALUE@{ui} does not begin with a @{b}!@{ub}, this test is true if the value of the @{I}NAME@{ui} attribute of the current insn is in the list specified by @{I}VALUE@{ui}. If @{I}VALUE@{ui} begins with a @{b}!@{ub}, this test is true if the attribute's value is @{i}not@{ui} in the specified list. For example, (eq_attr "type" "load,store") is equivalent to (ior (eq_attr "type" "load") (eq_attr "type" "store")) If @{I}NAME@{ui} specifies an attribute of @{b}alternative@{ub}, it refers to the value of the compiler variable @{b}which_alternative@{ub} (see @{"Output Statement" Link "Output Statement"}) and the values must be small integers. For example, (eq_attr "alternative" "2,3") is equivalent to (ior (eq (symbol_ref "which_alternative") (const_int 2)) (eq (symbol_ref "which_alternative") (const_int 3))) Note that, for most attributes, an @{b}eq_attr@{ub} test is simplified in cases where the value of the attribute being tested is known for all insns matching a particular pattern. This is by far the most common case. @{b}(attr_flag @{I}NAME@{ui})@{ub} The value of an @{b}attr_flag@{ub} expression is true if the flag specified by @{I}NAME@{ui} is true for the @{b}insn@{ub} currently being scheduled. @{I}NAME@{ui} is a string specifying one of a fixed set of flags to test. Test the flags @{b}forward@{ub} and @{b}backward@{ub} to determine the direction of a conditional branch. Test the flags @{b}very_likely@{ub}, @{b}likely@{ub}, @{b}very_unlikely@{ub}, and @{b}unlikely@{ub} to determine if a conditional branch is expected to be taken. If the @{b}very_likely@{ub} flag is true, then the @{b}likely@{ub} flag is also true. Likewise for the @{b}very_unlikely@{ub} and @{b}unlikely@{ub} flags. This example describes a conditional branch delay slot which can be nullified for forward branches that are taken (annul-true) or for backward branches which are not taken (annul-false). (define_delay (eq_attr "type" "cbranch") [(eq_attr "in_branch_delay" "true") (and (eq_attr "in_branch_delay" "true") (attr_flag "forward")) (and (eq_attr "in_branch_delay" "true") (attr_flag "backward"))]) The @{b}forward@{ub} and @{b}backward@{ub} flags are false if the current @{b}insn@{ub} being scheduled is not a conditional branch. The @{b}very_likely@{ub} and @{b}likely@{ub} flags are true if the @{b}insn@{ub} being scheduled is not a conditional branch. The The @{b}very_unlikely@{ub} and @{b}unlikely@{ub} flags are false if the @{b}insn@{ub} being scheduled is not a conditional branch. @{b}attr_flag@{ub} is only used during delay slot scheduling and has no meaning to other passes of the compiler. @EndNode @Node "Tagging Insns" "gcc.guide/Tagging Insns" @Next "Attr Example" @Prev "Expressions" @Toc "Insn Attributes" Assigning Attribute Values to Insns ----------------------------------- The value assigned to an attribute of an insn is primarily determined by which pattern is matched by that insn (or which @{b}define_peephole@{ub} generated it). Every @{b}define_insn@{ub} and @{b}define_peephole@{ub} can have an optional last argument to specify the values of attributes for matching insns. The value of any attribute not specified in a particular insn is set to the default value for that attribute, as specified in its @{b}define_attr@{ub}. Extensive use of default values for attributes permits the specification of the values for only one or two attributes in the definition of most insn patterns, as seen in the example in the next section. The optional last argument of @{b}define_insn@{ub} and @{b}define_peephole@{ub} is a vector of expressions, each of which defines the value for a single attribute. The most general way of assigning an attribute's value is to use a @{b}set@{ub} expression whose first operand is an @{b}attr@{ub} expression giving the name of the attribute being set. The second operand of the @{b}set@{ub} is an attribute expression (see @{"Expressions" Link "Expressions"}) giving the value of the attribute. When the attribute value depends on the @{b}alternative@{ub} attribute (i.e., which is the applicable alternative in the constraint of the insn), the @{b}set_attr_alternative@{ub} expression can be used. It allows the specification of a vector of attribute expressions, one for each alternative. When the generality of arbitrary attribute expressions is not required, the simpler @{b}set_attr@{ub} expression can be used, which allows specifying a string giving either a single attribute value or a list of attribute values, one for each alternative. The form of each of the above specifications is shown below. In each case, @{I}NAME@{ui} is a string specifying the attribute to be set. @{b}(set_attr @{I}NAME@{ui} @{I}VALUE-STRING@{ui})@{ub} @{I}VALUE-STRING@{ui} is either a string giving the desired attribute value, or a string containing a comma-separated list giving the values for succeeding alternatives. The number of elements must match the number of alternatives in the constraint of the insn pattern. Note that it may be useful to specify @{b}*@{ub} for some alternative, in which case the attribute will assume its default value for insns matching that alternative. @{b}(set_attr_alternative @{I}NAME@{ui} [@{I}VALUE1@{ui} @{I}VALUE2@{ui} ...])@{ub} Depending on the alternative of the insn, the value will be one of the specified values. This is a shorthand for using a @{b}cond@{ub} with tests on the @{b}alternative@{ub} attribute. @{b}(set (attr @{I}NAME@{ui}) @{I}VALUE@{ui})@{ub} The first operand of this @{b}set@{ub} must be the special RTL expression @{b}attr@{ub}, whose sole operand is a string giving the name of the attribute being set. @{I}VALUE@{ui} is the value of the attribute. The following shows three different ways of representing the same attribute value specification: (set_attr "type" "load,store,arith") (set_attr_alternative "type" [(const_string "load") (const_string "store") (const_string "arith")]) (set (attr "type") (cond [(eq_attr "alternative" "1") (const_string "load") (eq_attr "alternative" "2") (const_string "store")] (const_string "arith"))) The @{b}define_asm_attributes@{ub} expression provides a mechanism to specify the attributes assigned to insns produced from an @{b}asm@{ub} statement. It has the form: (define_asm_attributes [@{I}ATTR-SETS@{ui}]) where @{I}ATTR-SETS@{ui} is specified the same as for both the @{b}define_insn@{ub} and the @{b}define_peephole@{ub} expressions. These values will typically be the "worst case" attribute values. For example, they might indicate that the condition code will be clobbered. A specification for a @{b}length@{ub} attribute is handled specially. The way to compute the length of an @{b}asm@{ub} insn is to multiply the length specified in the expression @{b}define_asm_attributes@{ub} by the number of machine instructions specified in the @{b}asm@{ub} statement, determined by counting the number of semicolons and newlines in the string. Therefore, the value of the @{b}length@{ub} attribute specified in a @{b}define_asm_attributes@{ub} should be the maximum possible length of a single machine instruction. @EndNode @Node "Attr Example" "gcc.guide/Attr Example" @Next "Insn Lengths" @Prev "Tagging Insns" @Toc "Insn Attributes" Example of Attribute Specifications ----------------------------------- The judicious use of defaulting is important in the efficient use of insn attributes. Typically, insns are divided into @{i}types@{ui} and an attribute, customarily called @{b}type@{ub}, is used to represent this value. This attribute is normally used only to define the default value for other attributes. An example will clarify this usage. Assume we have a RISC machine with a condition code and in which only full-word operations are performed in registers. Let us assume that we can divide all insns into loads, stores, (integer) arithmetic operations, floating point operations, and branches. Here we will concern ourselves with determining the effect of an insn on the condition code and will limit ourselves to the following possible effects: The condition code can be set unpredictably (clobbered), not be changed, be set to agree with the results of the operation, or only changed if the item previously set into the condition code has been modified. Here is part of a sample @{b}md@{ub} file for such a machine: (define_attr "type" "load,store,arith,fp,branch" (const_string "arith")) (define_attr "cc" "clobber,unchanged,set,change0" (cond [(eq_attr "type" "load") (const_string "change0") (eq_attr "type" "store,branch") (const_string "unchanged") (eq_attr "type" "arith") (if_then_else (match_operand:SI 0 "" "") (const_string "set") (const_string "clobber"))] (const_string "clobber"))) (define_insn "" [(set (match_operand:SI 0 "general_operand" "=r,r,m") (match_operand:SI 1 "general_operand" "r,m,r"))] "" "@ move %0,%1 load %0,%1 store %0,%1" [(set_attr "type" "arith,load,store")]) Note that we assume in the above example that arithmetic operations performed on quantities smaller than a machine word clobber the condition code since they will set the condition code to a value corresponding to the full-word result. @EndNode @Node "Insn Lengths" "gcc.guide/Insn Lengths" @Next "Constant Attributes" @Prev "Attr Example" @Toc "Insn Attributes" Computing the Length of an Insn ------------------------------- For many machines, multiple types of branch instructions are provided, each for different length branch displacements. In most cases, the assembler will choose the correct instruction to use. However, when the assembler cannot do so, GCC can when a special attribute, the @{b}length@{ub} attribute, is defined. This attribute must be defined to have numeric values by specifying a null string in its @{b}define_attr@{ub}. In the case of the @{b}length@{ub} attribute, two additional forms of arithmetic terms are allowed in test expressions: @{b}(match_dup @{I}N@{ui})@{ub} This refers to the address of operand @{I}N@{ui} of the current insn, which must be a @{b}label_ref@{ub}. @{b}(pc)@{ub} This refers to the address of the @{i}current@{ui} insn. It might have been more consistent with other usage to make this the address of the @{i}next@{ui} insn but this would be confusing because the length of the current insn is to be computed. For normal insns, the length will be determined by value of the @{b}length@{ub} attribute. In the case of @{b}addr_vec@{ub} and @{b}addr_diff_vec@{ub} insn patterns, the length is computed as the number of vectors multiplied by the size of each vector. Lengths are measured in addressable storage units (bytes). The following macros can be used to refine the length computation: @{b}FIRST_INSN_ADDRESS@{ub} When the @{b}length@{ub} insn attribute is used, this macro specifies the value to be assigned to the address of the first insn in a function. If not specified, 0 is used. @{b}ADJUST_INSN_LENGTH (@{I}INSN@{ui}, @{I}LENGTH@{ui})@{ub} If defined, modifies the length assigned to instruction @{I}INSN@{ui} as a function of the context in which it is used. @{I}LENGTH@{ui} is an lvalue that contains the initially computed length of the insn and should be updated with the correct length of the insn. If updating is required, @{I}INSN@{ui} must not be a varying-length insn. This macro will normally not be required. A case in which it is required is the ROMP. On this machine, the size of an @{b}addr_vec@{ub} insn must be increased by two to compensate for the fact that alignment may be required. The routine that returns @{b}get_attr_length@{ub} (the value of the @{b}length@{ub} attribute) can be used by the output routine to determine the form of the branch instruction to be written, as the example below illustrates. As an example of the specification of variable-length branches, consider the IBM 360. If we adopt the convention that a register will be set to the starting address of a function, we can jump to labels within 4k of the start using a four-byte instruction. Otherwise, we need a six-byte sequence to load the address from memory and then branch to it. On such a machine, a pattern for a branch instruction might be specified as follows: (define_insn "jump" [(set (pc) (label_ref (match_operand 0 "" "")))] "" "* { return (get_attr_length (insn) == 4 ? \\"b %l0\\" : \\"l r15,=a(%l0); br r15\\"); }" [(set (attr "length") (if_then_else (lt (match_dup 0) (const_int 4096)) (const_int 4) (const_int 6)))]) @EndNode @Node "Constant Attributes" "gcc.guide/Constant Attributes" @Next "Delay Slots" @Prev "Insn Lengths" @Toc "Insn Attributes" Constant Attributes ------------------- A special form of @{b}define_attr@{ub}, where the expression for the default value is a @{b}const@{ub} expression, indicates an attribute that is constant for a given run of the compiler. Constant attributes may be used to specify which variety of processor is used. For example, (define_attr "cpu" "m88100,m88110,m88000" (const (cond [(symbol_ref "TARGET_88100") (const_string "m88100") (symbol_ref "TARGET_88110") (const_string "m88110")] (const_string "m88000")))) (define_attr "memory" "fast,slow" (const (if_then_else (symbol_ref "TARGET_FAST_MEM") (const_string "fast") (const_string "slow")))) The routine generated for constant attributes has no parameters as it does not depend on any particular insn. RTL expressions used to define the value of a constant attribute may use the @{b}symbol_ref@{ub} form, but may not use either the @{b}match_operand@{ub} form or @{b}eq_attr@{ub} forms involving insn attributes. @EndNode @Node "Delay Slots" "gcc.guide/Delay Slots" @Next "Function Units" @Prev "Constant Attributes" @Toc "Insn Attributes" Delay Slot Scheduling --------------------- The insn attribute mechanism can be used to specify the requirements for delay slots, if any, on a target machine. An instruction is said to require a @{i}delay slot@{ui} if some instructions that are physically after the instruction are executed as if they were located before it. Classic examples are branch and call instructions, which often execute the following instruction before the branch or call is performed. On some machines, conditional branch instructions can optionally @{i}annul@{ui} instructions in the delay slot. This means that the instruction will not be executed for certain branch outcomes. Both instructions that annul if the branch is true and instructions that annul if the branch is false are supported. Delay slot scheduling differs from instruction scheduling in that determining whether an instruction needs a delay slot is dependent only on the type of instruction being generated, not on data flow between the instructions. See the next section for a discussion of data-dependent instruction scheduling. The requirement of an insn needing one or more delay slots is indicated via the @{b}define_delay@{ub} expression. It has the following form: (define_delay @{I}TEST@{ui} [@{I}DELAY-1@{ui} @{I}ANNUL-TRUE-1@{ui} @{I}ANNUL-FALSE-1@{ui} @{I}DELAY-2@{ui} @{I}ANNUL-TRUE-2@{ui} @{I}ANNUL-FALSE-2@{ui} ...]) @{I}TEST@{ui} is an attribute test that indicates whether this @{b}define_delay@{ub} applies to a particular insn. If so, the number of required delay slots is determined by the length of the vector specified as the second argument. An insn placed in delay slot @{I}N@{ui} must satisfy attribute test @{I}DELAY-N@{ui}. @{I}ANNUL-TRUE-N@{ui} is an attribute test that specifies which insns may be annulled if the branch is true. Similarly, @{I}ANNUL-FALSE-N@{ui} specifies which insns in the delay slot may be annulled if the branch is false. If annulling is not supported for that delay slot, @{b}(nil)@{ub} should be coded. For example, in the common case where branch and call insns require a single delay slot, which may contain any insn other than a branch or call, the following would be placed in the @{b}md@{ub} file: (define_delay (eq_attr "type" "branch,call") [(eq_attr "type" "!branch,call") (nil) (nil)]) Multiple @{b}define_delay@{ub} expressions may be specified. In this case, each such expression specifies different delay slot requirements and there must be no insn for which tests in two @{b}define_delay@{ub} expressions are both true. For example, if we have a machine that requires one delay slot for branches but two for calls, no delay slot can contain a branch or call insn, and any valid insn in the delay slot for the branch can be annulled if the branch is true, we might represent this as follows: (define_delay (eq_attr "type" "branch") [(eq_attr "type" "!branch,call") (eq_attr "type" "!branch,call") (nil)]) (define_delay (eq_attr "type" "call") [(eq_attr "type" "!branch,call") (nil) (nil) (eq_attr "type" "!branch,call") (nil) (nil)]) @EndNode @Node "Function Units" "gcc.guide/Function Units" @Prev "Delay Slots" @Toc "Insn Attributes" Specifying Function Units ------------------------- On most RISC machines, there are instructions whose results are not available for a specific number of cycles. Common cases are instructions that load data from memory. On many machines, a pipeline stall will result if the data is referenced too soon after the load instruction. In addition, many newer microprocessors have multiple function units, usually one for integer and one for floating point, and often will incur pipeline stalls when a result that is needed is not yet ready. The descriptions in this section allow the specification of how much time must elapse between the execution of an instruction and the time when its result is used. It also allows specification of when the execution of an instruction will delay execution of similar instructions due to function unit conflicts. For the purposes of the specifications in this section, a machine is divided into @{i}function units@{ui}, each of which execute a specific class of instructions in first-in-first-out order. Function units that accept one instruction each cycle and allow a result to be used in the succeeding instruction (usually via forwarding) need not be specified. Classic RISC microprocessors will normally have a single function unit, which we can call @{b}memory@{ub}. The newer "superscalar" processors will often have function units for floating point operations, usually at least a floating point adder and multiplier. Each usage of a function units by a class of insns is specified with a @{b}define_function_unit@{ub} expression, which looks like this: (define_function_unit @{I}NAME@{ui} @{I}MULTIPLICITY@{ui} @{I}SIMULTANEITY@{ui} @{I}TEST@{ui} @{I}READY-DELAY@{ui} @{I}ISSUE-DELAY@{ui} [@{I}CONFLICT-LIST@{ui}]) @{I}NAME@{ui} is a string giving the name of the function unit. @{I}MULTIPLICITY@{ui} is an integer specifying the number of identical units in the processor. If more than one unit is specified, they will be scheduled independently. Only truly independent units should be counted; a pipelined unit should be specified as a single unit. (The only common example of a machine that has multiple function units for a single instruction class that are truly independent and not pipelined are the two multiply and two increment units of the CDC 6600.) @{I}SIMULTANEITY@{ui} specifies the maximum number of insns that can be executing in each instance of the function unit simultaneously or zero if the unit is pipelined and has no limit. All @{b}define_function_unit@{ub} definitions referring to function unit @{I}NAME@{ui} must have the same name and values for @{I}MULTIPLICITY@{ui} and @{I}SIMULTANEITY@{ui}. @{I}TEST@{ui} is an attribute test that selects the insns we are describing in this definition. Note that an insn may use more than one function unit and a function unit may be specified in more than one @{b}define_function_unit@{ub}. @{I}READY-DELAY@{ui} is an integer that specifies the number of cycles after which the result of the instruction can be used without introducing any stalls. @{I}ISSUE-DELAY@{ui} is an integer that specifies the number of cycles after the instruction matching the @{I}TEST@{ui} expression begins using this unit until a subsequent instruction can begin. A cost of @{I}N@{ui} indicates an @{I}N-1@{ui} cycle delay. A subsequent instruction may also be delayed if an earlier instruction has a longer @{I}READY-DELAY@{ui} value. This blocking effect is computed using the @{I}SIMULTANEITY@{ui}, @{I}READY-DELAY@{ui}, @{I}ISSUE-DELAY@{ui}, and @{I}CONFLICT-LIST@{ui} terms. For a normal non-pipelined function unit, @{I}SIMULTANEITY@{ui} is one, the unit is taken to block for the @{I}READY-DELAY@{ui} cycles of the executing insn, and smaller values of @{I}ISSUE-DELAY@{ui} are ignored. @{I}CONFLICT-LIST@{ui} is an optional list giving detailed conflict costs for this unit. If specified, it is a list of condition test expressions to be applied to insns chosen to execute in @{I}NAME@{ui} following the particular insn matching @{I}TEST@{ui} that is already executing in @{I}NAME@{ui}. For each insn in the list, @{I}ISSUE-DELAY@{ui} specifies the conflict cost; for insns not in the list, the cost is zero. If not specified, @{I}CONFLICT-LIST@{ui} defaults to all instructions that use the function unit. Typical uses of this vector are where a floating point function unit can pipeline either single- or double-precision operations, but not both, or where a memory unit can pipeline loads, but not stores, etc. As an example, consider a classic RISC machine where the result of a load instruction is not available for two cycles (a single "delay" instruction is required) and where only one load instruction can be executed simultaneously. This would be specified as: (define_function_unit "memory" 1 1 (eq_attr "type" "load") 2 0) For the case of a floating point function unit that can pipeline either single or double precision, but not both, the following could be specified: (define_function_unit "fp" 1 0 (eq_attr "type" "sp_fp") 4 4 [(eq_attr "type" "dp_fp")]) (define_function_unit "fp" 1 0 (eq_attr "type" "dp_fp") 4 4 [(eq_attr "type" "sp_fp")]) @{i}Note:@{ui} The scheduler attempts to avoid function unit conflicts and uses all the specifications in the @{b}define_function_unit@{ub} expression. It has recently come to our attention that these specifications may not allow modeling of some of the newer "superscalar" processors that have insns using multiple pipelined units. These insns will cause a potential conflict for the second unit used during their execution and there is no way of representing that conflict. We welcome any examples of how function unit conflicts work in such processors and suggestions for their representation. @EndNode @Node "Target Macros" "gcc.guide/Target Macros" @Next "Config" @Prev "Machine Desc" @Toc "Main" Target Description Macros ************************* In addition to the file @{b}@{I}MACHINE@{ui}.md@{ub}, a machine description includes a C header file conventionally given the name @{b}@{I}MACHINE@{ui}.h@{ub}. This header file defines numerous macros that convey the information about the target machine that does not fit into the scheme of the @{b}.md@{ub} file. The file @{b}tm.h@{ub} should be a link to @{b}@{I}MACHINE@{ui}.h@{ub}. The header file @{b}config.h@{ub} includes @{b}tm.h@{ub} and most compiler source files include @{b}config.h@{ub}. @{" Driver " Link "Driver"} Controlling how the driver runs the compilation passes. @{" Run-time Target " Link "Run-time Target"} Defining @{b}-m@{ub} options like @{b}-m68000@{ub} and @{b}-m68020@{ub}. @{" Storage Layout " Link "Storage Layout"} Defining sizes and alignments of data. @{" Type Layout " Link "Type Layout"} Defining sizes and properties of basic user data types. @{" Registers " Link "Registers"} Naming and describing the hardware registers. @{" Register Classes " Link "Register Classes"} Defining the classes of hardware registers. @{" Stack and Calling " Link "Stack and Calling"} Defining which way the stack grows and by how much. @{" Varargs " Link "Varargs"} Defining the varargs macros. @{" Trampolines " Link "Trampolines"} Code set up at run time to enter a nested function. @{" Library Calls " Link "Library Calls"} Controlling how library routines are implicitly called. @{" Addressing Modes " Link "Addressing Modes"} Defining addressing modes valid for memory operands. @{" Condition Code " Link "Condition Code"} Defining how insns update the condition code. @{" Costs " Link "Costs"} Defining relative costs of different operations. @{" Sections " Link "Sections"} Dividing storage into text, data, and other sections. @{" PIC " Link "PIC"} Macros for position independent code. @{" Assembler Format " Link "Assembler Format"} Defining how to write insns and pseudo-ops to output. @{" Debugging Info " Link "Debugging Info"} Defining the format of debugging output. @{" Cross-compilation " Link "Cross-compilation"} Handling floating point for cross-compilers. @{" Misc " Link "Misc"} Everything else. @EndNode @Node "Driver" "gcc.guide/Driver" @Next "Run-time Target" @Toc "Target Macros" Controlling the Compilation Driver, @{b}gcc@{ub} ======================================= You can control the compilation driver. @{b}SWITCH_TAKES_ARG (@{I}CHAR@{ui})@{ub} A C expression which determines whether the option @{b} -@{I}CHAR@{ui}@{ub} takes arguments. The value should be the number of arguments that option takes-zero, for many options. By default, this macro is defined to handle the standard options properly. You need not define it unless you wish to add additional options which take arguments. @{b}WORD_SWITCH_TAKES_ARG (@{I}NAME@{ui})@{ub} A C expression which determines whether the option @{b} -@{I}NAME@{ui}@{ub} takes arguments. The value should be the number of arguments that option takes-zero, for many options. This macro rather than @{b}SWITCH_TAKES_ARG@{ub} is used for multi-character option names. By default, this macro is defined as @{b}DEFAULT_WORD_SWITCH_TAKES_ARG@{ub}, which handles the standard options properly. You need not define @{b}WORD_SWITCH_TAKES_ARG@{ub} unless you wish to add additional options which take arguments. Any redefinition should call @{b}DEFAULT_WORD_SWITCH_TAKES_ARG@{ub} and then check for additional options. @{b}SWITCHES_NEED_SPACES@{ub} A string-valued C expression which is nonempty if the linker needs a space between the @{b}-L@{ub} or @{b}-o@{ub} option and its argument. If this macro is not defined, the default value is 0. @{b}CPP_SPEC@{ub} A C string constant that tells the GNU CC driver program options to pass to CPP. It can also specify how to translate options you give to GNU CC into options for GNU CC to pass to the CPP. Do not define this macro if it does not need to do anything. @{b}NO_BUILTIN_SIZE_TYPE@{ub} If this macro is defined, the preprocessor will not define the builtin macro @{b}__SIZE_TYPE__@{ub}. The macro @{b}__SIZE_TYPE__@{ub} must then be defined by @{b}CPP_SPEC@{ub} instead. This should be defined if @{b}SIZE_TYPE@{ub} depends on target dependent flags which are not accessible to the preprocessor. Otherwise, it should not be defined. @{b}NO_BUILTIN_PTRDIFF_TYPE@{ub} If this macro is defined, the preprocessor will not define the builtin macro @{b}__PTRDIFF_TYPE__@{ub}. The macro @{b}__PTRDIFF_TYPE__@{ub} must then be defined by @{b}CPP_SPEC@{ub} instead. This should be defined if @{b}PTRDIFF_TYPE@{ub} depends on target dependent flags which are not accessible to the preprocessor. Otherwise, it should not be defined. @{b}SIGNED_CHAR_SPEC@{ub} A C string constant that tells the GNU CC driver program options to pass to CPP. By default, this macro is defined to pass the option @{b}-D__CHAR_UNSIGNED__@{ub} to CPP if @{b}char@{ub} will be treated as @{b}unsigned char@{ub} by @{b}cc1@{ub}. Do not define this macro unless you need to override the default definition. @{b}CC1_SPEC@{ub} A C string constant that tells the GNU CC driver program options to pass to @{b}cc1@{ub}. It can also specify how to translate options you give to GNU CC into options for GNU CC to pass to the @{b}cc1@{ub}. Do not define this macro if it does not need to do anything. @{b}CC1PLUS_SPEC@{ub} A C string constant that tells the GNU CC driver program options to pass to @{b}cc1plus@{ub}. It can also specify how to translate options you give to GNU CC into options for GNU CC to pass to the @{b}cc1plus@{ub}. Do not define this macro if it does not need to do anything. @{b}ASM_SPEC@{ub} A C string constant that tells the GNU CC driver program options to pass to the assembler. It can also specify how to translate options you give to GNU CC into options for GNU CC to pass to the assembler. See the file @{b}sun3.h@{ub} for an example of this. Do not define this macro if it does not need to do anything. @{b}ASM_FINAL_SPEC@{ub} A C string constant that tells the GNU CC driver program how to run any programs which cleanup after the normal assembler. Normally, this is not needed. See the file @{b}mips.h@{ub} for an example of this. Do not define this macro if it does not need to do anything. @{b}LINK_SPEC@{ub} A C string constant that tells the GNU CC driver program options to pass to the linker. It can also specify how to translate options you give to GNU CC into options for GNU CC to pass to the linker. Do not define this macro if it does not need to do anything. @{b}LIB_SPEC@{ub} Another C string constant used much like @{b}LINK_SPEC@{ub}. The difference between the two is that @{b}LIB_SPEC@{ub} is used at the end of the command given to the linker. If this macro is not defined, a default is provided that loads the standard C library from the usual place. See @{b}gcc.c@{ub}. @{b}LIBGCC_SPEC@{ub} Another C string constant that tells the GNU CC driver program how and when to place a reference to @{b}libgcc.a@{ub} into the linker command line. This constant is placed both before and after the value of @{b}LIB_SPEC@{ub}. If this macro is not defined, the GNU CC driver provides a default that passes the string @{b}-lgcc@{ub} to the linker unless the @{b}-shared@{ub} option is specified. @{b}STARTFILE_SPEC@{ub} Another C string constant used much like @{b}LINK_SPEC@{ub}. The difference between the two is that @{b}STARTFILE_SPEC@{ub} is used at the very beginning of the command given to the linker. If this macro is not defined, a default is provided that loads the standard C startup file from the usual place. See @{b}gcc.c@{ub}. @{b}ENDFILE_SPEC@{ub} Another C string constant used much like @{b}LINK_SPEC@{ub}. The difference between the two is that @{b}ENDFILE_SPEC@{ub} is used at the very end of the command given to the linker. Do not define this macro if it does not need to do anything. @{b}LINK_LIBGCC_SPECIAL@{ub} Define this macro if the driver program should find the library @{b}libgcc.a@{ub} itself and should not pass @{b}-L@{ub} options to the linker. If you do not define this macro, the driver program will pass the argument @{b}-lgcc@{ub} to tell the linker to do the search and will pass @{b}-L@{ub} options to it. @{b}LINK_LIBGCC_SPECIAL_1@{ub} Define this macro if the driver program should find the library @{b}libgcc.a@{ub}. If you do not define this macro, the driver program will pass the argument @{b}-lgcc@{ub} to tell the linker to do the search. This macro is similar to @{b}LINK_LIBGCC_SPECIAL@{ub}, except that it does not affect @{b}-L@{ub} options. @{b}MULTILIB_DEFAULTS@{ub} Define this macro as a C expression for the initializer of an array of string to tell the driver program which options are defaults for this target and thus do not need to be handled specially when using @{b}MULTILIB_OPTIONS@{ub}. Do not define this macro if @{b}MULTILIB_OPTIONS@{ub} is not defined in the target makefile fragment or if none of the options listed in @{b}MULTILIB_OPTIONS@{ub} are set by default. See @{"Target Fragment" Link "Target Fragment"}. @{b}RELATIVE_PREFIX_NOT_LINKDIR@{ub} Define this macro to tell @{b}gcc@{ub} that it should only translate a @{b}-B@{ub} prefix into a @{b}-L@{ub} linker option if the prefix indicates an absolute file name. @{b}STANDARD_EXEC_PREFIX@{ub} Define this macro as a C string constant if you wish to override the standard choice of @{b}/usr/local/lib/gcc-lib/@{ub} as the default prefix to try when searching for the executable files of the compiler. @{b}MD_EXEC_PREFIX@{ub} If defined, this macro is an additional prefix to try after @{b}STANDARD_EXEC_PREFIX@{ub}. @{b}MD_EXEC_PREFIX@{ub} is not searched when the @{b}-b@{ub} option is used, or the compiler is built as a cross compiler. @{b}STANDARD_STARTFILE_PREFIX@{ub} Define this macro as a C string constant if you wish to override the standard choice of @{b}/usr/local/lib/@{ub} as the default prefix to try when searching for startup files such as @{b}crt0.o@{ub}. @{b}MD_STARTFILE_PREFIX@{ub} If defined, this macro supplies an additional prefix to try after the standard prefixes. @{b}MD_EXEC_PREFIX@{ub} is not searched when the @{b}-b@{ub} option is used, or when the compiler is built as a cross compiler. @{b}MD_STARTFILE_PREFIX_1@{ub} If defined, this macro supplies yet another prefix to try after the standard prefixes. It is not searched when the @{b}-b@{ub} option is used, or when the compiler is built as a cross compiler. @{b}INIT_ENVIRONMENT@{ub} Define this macro as a C string constant if you with to set environment variables for programs called by the driver, such as the assembler and loader. The driver passes the value of this macro to @{b}putenv@{ub} to initialize the necessary environment variables. @{b}LOCAL_INCLUDE_DIR@{ub} Define this macro as a C string constant if you wish to override the standard choice of @{b}/usr/local/include@{ub} as the default prefix to try when searching for local header files. @{b}LOCAL_INCLUDE_DIR@{ub} comes before @{b}SYSTEM_INCLUDE_DIR@{ub} in the search order. Cross compilers do not use this macro and do not search either @{b}/usr/local/include@{ub} or its replacement. @{b}SYSTEM_INCLUDE_DIR@{ub} Define this macro as a C string constant if you wish to specify a system-specific directory to search for header files before the standard directory. @{b}SYSTEM_INCLUDE_DIR@{ub} comes before @{b}STANDARD_INCLUDE_DIR@{ub} in the search order. Cross compilers do not use this macro and do not search the directory specified. @{b}STANDARD_INCLUDE_DIR@{ub} Define this macro as a C string constant if you wish to override the standard choice of @{b}/usr/include@{ub} as the default prefix to try when searching for header files. Cross compilers do not use this macro and do not search either @{b}/usr/include@{ub} or its replacement. @{b}INCLUDE_DEFAULTS@{ub} Define this macro if you wish to override the entire default search path for include files. The default search path includes @{b}GCC_INCLUDE_DIR@{ub}, @{b}LOCAL_INCLUDE_DIR@{ub}, @{b}SYSTEM_INCLUDE_DIR@{ub}, @{b}GPLUSPLUS_INCLUDE_DIR@{ub}, and @{b}STANDARD_INCLUDE_DIR@{ub}. In addition, @{b}GPLUSPLUS_INCLUDE_DIR@{ub} and @{b}GCC_INCLUDE_DIR@{ub} are defined automatically by @{b}Makefile@{ub}, and specify private search areas for GCC. The directory @{b}GPLUSPLUS_INCLUDE_DIR@{ub} is used only for C++ programs. The definition should be an initializer for an array of structures. Each array element should have two elements: the directory name (a string constant) and a flag for C++-only directories. Mark the end of the array with a null element. For example, here is the definition used for VMS: #define INCLUDE_DEFAULTS \\ { \\ { "GNU_GXX_INCLUDE:", 1}, \\ { "GNU_CC_INCLUDE:", 0}, \\ { "SYS$SYSROOT:[SYSLIB.]", 0}, \\ { ".", 0}, \\ { 0, 0} \\ } Here is the order of prefixes tried for exec files: 1. Any prefixes specified by the user with @{b}-B@{ub}. 2. The environment variable @{b}GCC_EXEC_PREFIX@{ub}, if any. 3. The directories specified by the environment variable @{b}COMPILER_PATH@{ub}. 4. The macro @{b}STANDARD_EXEC_PREFIX@{ub}. 5. @{b}/usr/lib/gcc/@{ub}. 6. The macro @{b}MD_EXEC_PREFIX@{ub}, if any. Here is the order of prefixes tried for startfiles: 1. Any prefixes specified by the user with @{b}-B@{ub}. 2. The environment variable @{b}GCC_EXEC_PREFIX@{ub}, if any. 3. The directories specified by the environment variable @{b}LIBRARY_PATH@{ub} (native only, cross compilers do not use this). 4. The macro @{b}STANDARD_EXEC_PREFIX@{ub}. 5. @{b}/usr/lib/gcc/@{ub}. 6. The macro @{b}MD_EXEC_PREFIX@{ub}, if any. 7. The macro @{b}MD_STARTFILE_PREFIX@{ub}, if any. 8. The macro @{b}STANDARD_STARTFILE_PREFIX@{ub}. 9. @{b}/lib/@{ub}. 10. @{b}/usr/lib/@{ub}. @EndNode @Node "Run-time Target" "gcc.guide/Run-time Target" @Next "Storage Layout" @Prev "Driver" @Toc "Target Macros" Run-time Target Specification ============================= Here are run-time target specifications. @{b}CPP_PREDEFINES@{ub} Define this to be a string constant containing @{b}-D@{ub} options to define the predefined macros that identify this machine and system. These macros will be predefined unless the @{b}-ansi@{ub} option is specified. In addition, a parallel set of macros are predefined, whose names are made by appending @{b}__@{ub} at the beginning and at the end. These @{b}__@{ub} macros are permitted by the ANSI standard, so they are predefined regardless of whether @{b}-ansi@{ub} is specified. For example, on the Sun, one can use the following value: "-Dmc68000 -Dsun -Dunix" The result is to define the macros @{b}__mc68000__@{ub}, @{b}__sun__@{ub} and @{b}__unix__@{ub} unconditionally, and the macros @{b}mc68000@{ub}, @{b}sun@{ub} and @{b}unix@{ub} provided @{b}-ansi@{ub} is not specified. @{b}extern int target_flags;@{ub} This declaration should be present. @{b}TARGET_...@{ub} This series of macros is to allow compiler command arguments to enable or disable the use of optional features of the target machine. For example, one machine description serves both the 68000 and the 68020; a command argument tells the compiler whether it should use 68020-only instructions or not. This command argument works by means of a macro @{b}TARGET_68020@{ub} that tests a bit in @{b}target_flags@{ub}. Define a macro @{b}TARGET_@{I}FEATURENAME@{ui}@{ub} for each such option. Its definition should test a bit in @{b}target_flags@{ub}; for example: #define TARGET_68020 (target_flags & 1) One place where these macros are used is in the condition-expressions of instruction patterns. Note how @{b}TARGET_68020@{ub} appears frequently in the 68000 machine description file, @{b}m68k.md@{ub}. Another place they are used is in the definitions of the other macros in the @{b}@{I}MACHINE@{ui}.h@{ub} file. @{b}TARGET_SWITCHES@{ub} This macro defines names of command options to set and clear bits in @{b}target_flags@{ub}. Its definition is an initializer with a subgrouping for each command option. Each subgrouping contains a string constant, that defines the option name, and a number, which contains the bits to set in @{b}target_flags@{ub}. A negative number says to clear bits instead; the negative of the number is which bits to clear. The actual option name is made by appending @{b}-m@{ub} to the specified name. One of the subgroupings should have a null string. The number in this grouping is the default value for @{b}target_flags@{ub}. Any target options act starting with that value. Here is an example which defines @{b}-m68000@{ub} and @{b}-m68020@{ub} with opposite meanings, and picks the latter as the default: #define TARGET_SWITCHES \\ { { "68020", 1}, \\ { "68000", -1}, \\ { "", 1}} @{b}TARGET_OPTIONS@{ub} This macro is similar to @{b}TARGET_SWITCHES@{ub} but defines names of command options that have values. Its definition is an initializer with a subgrouping for each command option. Each subgrouping contains a string constant, that defines the fixed part of the option name, and the address of a variable. The variable, type @{b}char *@{ub}, is set to the variable part of the given option if the fixed part matches. The actual option name is made by appending @{b}-m@{ub} to the specified name. Here is an example which defines @{b}-mshort-data-@{I}NUMBER@{ui}@{ub}. If the given option is @{b}-mshort-data-512@{ub}, the variable @{b}m88k_short_data@{ub} will be set to the string @{b}"512"@{ub}. extern char *m88k_short_data; #define TARGET_OPTIONS \\ { { "short-data-", &m88k_short_data } } @{b}TARGET_VERSION@{ub} This macro is a C statement to print on @{b}stderr@{ub} a string describing the particular machine description choice. Every machine description should define @{b}TARGET_VERSION@{ub}. For example: #ifdef MOTOROLA #define TARGET_VERSION \\ fprintf (stderr, " (68k, Motorola syntax)"); #else #define TARGET_VERSION \\ fprintf (stderr, " (68k, MIT syntax)"); #endif @{b}OVERRIDE_OPTIONS@{ub} Sometimes certain combinations of command options do not make sense on a particular target machine. You can define a macro @{b}OVERRIDE_OPTIONS@{ub} to take account of this. This macro, if defined, is executed once just after all the command options have been parsed. Don't use this macro to turn on various extra optimizations for @{b}-O@{ub}. That is what @{b}OPTIMIZATION_OPTIONS@{ub} is for. @{b}OPTIMIZATION_OPTIONS (@{I}LEVEL@{ui})@{ub} Some machines may desire to change what optimizations are performed for various optimization levels. This macro, if defined, is executed once just after the optimization level is determined and before the remainder of the command options have been parsed. Values set in this macro are used as the default values for the other command line options. @{I}LEVEL@{ui} is the optimization level specified; 2 if @{b}-O2@{ub} is specified, 1 if @{b}-O@{ub} is specified, and 0 if neither is specified. You should not use this macro to change options that are not machine-specific. These should uniformly selected by the same optimization level on all supported machines. Use this macro to enable machbine-specific optimizations. @{i}Do not examine @{b}write_symbols@{ub} in this macro!@{ui} The debugging options are not supposed to alter the generated code. @{b}CAN_DEBUG_WITHOUT_FP@{ub} Define this macro if debugging can be performed even without a frame pointer. If this macro is defined, GNU CC will turn on the @{b}-fomit-frame-pointer@{ub} option whenever @{b}-O@{ub} is specified. @EndNode @Node "Storage Layout" "gcc.guide/Storage Layout" @Next "Type Layout" @Prev "Run-time Target" @Toc "Target Macros" Storage Layout ============== Note that the definitions of the macros in this table which are sizes or alignments measured in bits do not need to be constant. They can be C expressions that refer to static variables, such as the @{b}target_flags@{ub}. See @{"Run-time Target" Link "Run-time Target"}. @{b}BITS_BIG_ENDIAN@{ub} Define this macro to have the value 1 if the most significant bit in a byte has the lowest number; otherwise define it to have the value zero. This means that bit-field instructions count from the most significant bit. If the machine has no bit-field instructions, then this must still be defined, but it doesn't matter which value it is defined to. This macro need not be a constant. This macro does not affect the way structure fields are packed into bytes or words; that is controlled by @{b}BYTES_BIG_ENDIAN@{ub}. @{b}BYTES_BIG_ENDIAN@{ub} Define this macro to have the value 1 if the most significant byte in a word has the lowest number. This macro need not be a constant. @{b}WORDS_BIG_ENDIAN@{ub} Define this macro to have the value 1 if, in a multiword object, the most significant word has the lowest number. This applies to both memory locations and registers; GNU CC fundamentally assumes that the order of words in memory is the same as the order in registers. This macro need not be a constant. @{b}LIBGCC2_WORDS_BIG_ENDIAN@{ub} Define this macro if WORDS_BIG_ENDIAN is not constant. This must be a constant value with the same meaning as WORDS_BIG_ENDIAN, which will be used only when compiling libgcc2.c. Typically the value will be set based on preprocessor defines. @{b}FLOAT_WORDS_BIG_ENDIAN@{ub} Define this macro to have the value 1 if @{b}DFmode@{ub}, @{b}XFmode@{ub} or @{b}TFmode@{ub} floating point numbers are stored in memory with the word containing the sign bit at the lowest address; otherwise define it to have the value 0. This macro need not be a constant. You need not define this macro if the ordering is the same as for multi-word integers. @{b}BITS_PER_UNIT@{ub} Define this macro to be the number of bits in an addressable storage unit (byte); normally 8. @{b}BITS_PER_WORD@{ub} Number of bits in a word; normally 32. @{b}MAX_BITS_PER_WORD@{ub} Maximum number of bits in a word. If this is undefined, the default is @{b}BITS_PER_WORD@{ub}. Otherwise, it is the constant value that is the largest value that @{b}BITS_PER_WORD@{ub} can have at run-time. @{b}UNITS_PER_WORD@{ub} Number of storage units in a word; normally 4. @{b}MIN_UNITS_PER_WORD@{ub} Minimum number of units in a word. If this is undefined, the default is @{b}UNITS_PER_WORD@{ub}. Otherwise, it is the constant value that is the smallest value that @{b}UNITS_PER_WORD@{ub} can have at run-time. @{b}POINTER_SIZE@{ub} Width of a pointer, in bits. You must specify a value no wider than the width of @{b}Pmode@{ub}. If it is not equal to the width of @{b}Pmode@{ub}, you must define @{b}POINTERS_EXTEND_UNSIGNED@{ub}. @{b}POINTERS_EXTEND_UNSIGNED@{ub} A C expression whose value is nonzero if pointers that need to be extended from being @{b}POINTER_SIZE@{ub} bits wide to @{b}Pmode@{ub} are sign-extended and zero if they are zero-extended. You need not define this macro if the @{b}POINTER_SIZE@{ub} is equal to the width of @{b}Pmode@{ub}. @{b}PROMOTE_MODE (@{I}M@{ui}, @{I}UNSIGNEDP@{ui}, @{I}TYPE@{ui})@{ub} A macro to update @{I}M@{ui} and @{I}UNSIGNEDP@{ui} when an object whose type is @{I}TYPE@{ui} and which has the specified mode and signedness is to be stored in a register. This macro is only called when @{I}TYPE@{ui} is a scalar type. On most RISC machines, which only have operations that operate on a full register, define this macro to set @{I}M@{ui} to @{b}word_mode@{ub} if @{I}M@{ui} is an integer mode narrower than @{b}BITS_PER_WORD@{ub}. In most cases, only integer modes should be widened because wider-precision floating-point operations are usually more expensive than their narrower counterparts. For most machines, the macro definition does not change @{I}UNSIGNEDP@{ui}. However, some machines, have instructions that preferentially handle either signed or unsigned quantities of certain modes. For example, on the DEC Alpha, 32-bit loads from memory and 32-bit add instructions sign-extend the result to 64 bits. On such machines, set @{I}UNSIGNEDP@{ui} according to which kind of extension is more efficient. Do not define this macro if it would never modify @{I}M@{ui}. @{b}PROMOTE_FUNCTION_ARGS@{ub} Define this macro if the promotion described by @{b}PROMOTE_MODE@{ub} should also be done for outgoing function arguments. @{b}PROMOTE_FUNCTION_RETURN@{ub} Define this macro if the promotion described by @{b}PROMOTE_MODE@{ub} should also be done for the return value of functions. If this macro is defined, @{b}FUNCTION_VALUE@{ub} must perform the same promotions done by @{b}PROMOTE_MODE@{ub}. @{b}PROMOTE_FOR_CALL_ONLY@{ub} Define this macro if the promotion described by @{b}PROMOTE_MODE@{ub} should @{i}only@{ui} be performed for outgoing function arguments or function return values, as specified by @{b}PROMOTE_FUNCTION_ARGS@{ub} and @{b}PROMOTE_FUNCTION_RETURN@{ub}, respectively. @{b}PARM_BOUNDARY@{ub} Normal alignment required for function parameters on the stack, in bits. All stack parameters receive at least this much alignment regardless of data type. On most machines, this is the same as the size of an integer. @{b}STACK_BOUNDARY@{ub} Define this macro if you wish to preserve a certain alignment for the stack pointer. The definition is a C expression for the desired alignment (measured in bits). If @{b}PUSH_ROUNDING@{ub} is not defined, the stack will always be aligned to the specified boundary. If @{b}PUSH_ROUNDING@{ub} is defined and specifies a less strict alignment than @{b}STACK_BOUNDARY@{ub}, the stack may be momentarily unaligned while pushing arguments. @{b}FUNCTION_BOUNDARY@{ub} Alignment required for a function entry point, in bits. @{b}BIGGEST_ALIGNMENT@{ub} Biggest alignment that any data type can require on this machine, in bits. @{b}BIGGEST_FIELD_ALIGNMENT@{ub} Biggest alignment that any structure field can require on this machine, in bits. If defined, this overrides @{b}BIGGEST_ALIGNMENT@{ub} for structure fields only. @{b}MAX_OFILE_ALIGNMENT@{ub} Biggest alignment supported by the object file format of this machine. Use this macro to limit the alignment which can be specified using the @{b}__attribute__ ((aligned (@{I}N@{ui})))@{ub} construct. If not defined, the default value is @{b}BIGGEST_ALIGNMENT@{ub}. @{b}DATA_ALIGNMENT (@{I}TYPE@{ui}, @{I}BASIC-ALIGN@{ui})@{ub} If defined, a C expression to compute the alignment for a static variable. @{I}TYPE@{ui} is the data type, and @{I}BASIC-ALIGN@{ui} is the alignment that the object would ordinarily have. The value of this macro is used instead of that alignment to align the object. If this macro is not defined, then @{I}BASIC-ALIGN@{ui} is used. One use of this macro is to increase alignment of medium-size data to make it all fit in fewer cache lines. Another is to cause character arrays to be word-aligned so that @{b}strcpy@{ub} calls that copy constants to character arrays can be done inline. @{b}CONSTANT_ALIGNMENT (@{I}CONSTANT@{ui}, @{I}BASIC-ALIGN@{ui})@{ub} If defined, a C expression to compute the alignment given to a constant that is being placed in memory. @{I}CONSTANT@{ui} is the constant and @{I}BASIC-ALIGN@{ui} is the alignment that the object would ordinarily have. The value of this macro is used instead of that alignment to align the object. If this macro is not defined, then @{I}BASIC-ALIGN@{ui} is used. The typical use of this macro is to increase alignment for string constants to be word aligned so that @{b}strcpy@{ub} calls that copy constants can be done inline. @{b}EMPTY_FIELD_BOUNDARY@{ub} Alignment in bits to be given to a structure bit field that follows an empty field such as @{b}int : 0;@{ub}. Note that @{b}PCC_BITFIELD_TYPE_MATTERS@{ub} also affects the alignment that results from an empty field. @{b}STRUCTURE_SIZE_BOUNDARY@{ub} Number of bits which any structure or union's size must be a multiple of. Each structure or union's size is rounded up to a multiple of this. If you do not define this macro, the default is the same as @{b}BITS_PER_UNIT@{ub}. @{b}STRICT_ALIGNMENT@{ub} Define this macro to be the value 1 if instructions will fail to work if given data not on the nominal alignment. If instructions will merely go slower in that case, define this macro as 0. @{b}PCC_BITFIELD_TYPE_MATTERS@{ub} Define this if you wish to imitate the way many other C compilers handle alignment of bitfields and the structures that contain them. The behavior is that the type written for a bitfield (@{b}int@{ub}, @{b}short@{ub}, or other integer type) imposes an alignment for the entire structure, as if the structure really did contain an ordinary field of that type. In addition, the bitfield is placed within the structure so that it would fit within such a field, not crossing a boundary for it. Thus, on most machines, a bitfield whose type is written as @{b}int@{ub} would not cross a four-byte boundary, and would force four-byte alignment for the whole structure. (The alignment used may not be four bytes; it is controlled by the other alignment parameters.) If the macro is defined, its definition should be a C expression; a nonzero value for the expression enables this behavior. Note that if this macro is not defined, or its value is zero, some bitfields may cross more than one alignment boundary. The compiler can support such references if there are @{b}insv@{ub}, @{b}extv@{ub}, and @{b}extzv@{ub} insns that can directly reference memory. The other known way of making bitfields work is to define @{b}STRUCTURE_SIZE_BOUNDARY@{ub} as large as @{b}BIGGEST_ALIGNMENT@{ub}. Then every structure can be accessed with fullwords. Unless the machine has bitfield instructions or you define @{b}STRUCTURE_SIZE_BOUNDARY@{ub} that way, you must define @{b}PCC_BITFIELD_TYPE_MATTERS@{ub} to have a nonzero value. If your aim is to make GNU CC use the same conventions for laying out bitfields as are used by another compiler, here is how to investigate what the other compiler does. Compile and run this program: struct foo1 { char x; char :0; char y; }; struct foo2 { char x; int :0; char y; }; main () { printf ("Size of foo1 is %d\\n", sizeof (struct foo1)); printf ("Size of foo2 is %d\\n", sizeof (struct foo2)); exit (0); } If this prints 2 and 5, then the compiler's behavior is what you would get from @{b}PCC_BITFIELD_TYPE_MATTERS@{ub}. @{b}BITFIELD_NBYTES_LIMITED@{ub} Like PCC_BITFIELD_TYPE_MATTERS except that its effect is limited to aligning a bitfield within the structure. @{b}ROUND_TYPE_SIZE (@{I}STRUCT@{ui}, @{I}SIZE@{ui}, @{I}ALIGN@{ui})@{ub} Define this macro as an expression for the overall size of a structure (given by @{I}STRUCT@{ui} as a tree node) when the size computed from the fields is @{I}SIZE@{ui} and the alignment is @{I}ALIGN@{ui}. The default is to round @{I}SIZE@{ui} up to a multiple of @{I}ALIGN@{ui}. @{b}ROUND_TYPE_ALIGN (@{I}STRUCT@{ui}, @{I}COMPUTED@{ui}, @{I}SPECIFIED@{ui})@{ub} Define this macro as an expression for the alignment of a structure (given by @{I}STRUCT@{ui} as a tree node) if the alignment computed in the usual way is @{I}COMPUTED@{ui} and the alignment explicitly specified was @{I}SPECIFIED@{ui}. The default is to use @{I}SPECIFIED@{ui} if it is larger; otherwise, use the smaller of @{I}COMPUTED@{ui} and @{b}BIGGEST_ALIGNMENT@{ub} @{b}MAX_FIXED_MODE_SIZE@{ub} An integer expression for the size in bits of the largest integer machine mode that should actually be used. All integer machine modes of this size or smaller can be used for structures and unions with the appropriate sizes. If this macro is undefined, @{b}GET_MODE_BITSIZE (DImode)@{ub} is assumed. @{b}CHECK_FLOAT_VALUE (@{I}MODE@{ui}, @{I}VALUE@{ui}, @{I}OVERFLOW@{ui})@{ub} A C statement to validate the value @{I}VALUE@{ui} (of type @{b}double@{ub}) for mode @{I}MODE@{ui}. This means that you check whether @{I}VALUE@{ui} fits within the possible range of values for mode @{I}MODE@{ui} on this target machine. The mode @{I}MODE@{ui} is always a mode of class @{b}MODE_FLOAT@{ub}. @{I}OVERFLOW@{ui} is nonzero if the value is already known to be out of range. If @{I}VALUE@{ui} is not valid or if @{I}OVERFLOW@{ui} is nonzero, you should set @{I}OVERFLOW@{ui} to 1 and then assign some valid value to @{I}VALUE@{ui}. Allowing an invalid value to go through the compiler can produce incorrect assembler code which may even cause Unix assemblers to crash. This macro need not be defined if there is no work for it to do. @{b}TARGET_FLOAT_FORMAT@{ub} A code distinguishing the floating point format of the target machine. There are three defined values: @{b}IEEE_FLOAT_FORMAT@{ub} This code indicates IEEE floating point. It is the default; there is no need to define this macro when the format is IEEE. @{b}VAX_FLOAT_FORMAT@{ub} This code indicates the peculiar format used on the Vax. @{b}UNKNOWN_FLOAT_FORMAT@{ub} This code indicates any other format. The value of this macro is compared with @{b}HOST_FLOAT_FORMAT@{ub} (see @{"Config" Link "Config"}) to determine whether the target machine has the same format as the host machine. If any other formats are actually in use on supported machines, new codes should be defined for them. The ordering of the component words of floating point values stored in memory is controlled by @{b}FLOAT_WORDS_BIG_ENDIAN@{ub} for the target machine and @{b}HOST_FLOAT_WORDS_BIG_ENDIAN@{ub} for the host. @EndNode @Node "Type Layout" "gcc.guide/Type Layout" @Next "Registers" @Prev "Storage Layout" @Toc "Target Macros" Layout of Source Language Data Types ==================================== These macros define the sizes and other characteristics of the standard basic data types used in programs being compiled. Unlike the macros in the previous section, these apply to specific features of C and related languages, rather than to fundamental aspects of storage layout. @{b}INT_TYPE_SIZE@{ub} A C expression for the size in bits of the type @{b}int@{ub} on the target machine. If you don't define this, the default is one word. @{b}MAX_INT_TYPE_SIZE@{ub} Maximum number for the size in bits of the type @{b}int@{ub} on the target machine. If this is undefined, the default is @{b}INT_TYPE_SIZE@{ub}. Otherwise, it is the constant value that is the largest value that @{b}INT_TYPE_SIZE@{ub} can have at run-time. This is used in @{b}cpp@{ub}. @{b}SHORT_TYPE_SIZE@{ub} A C expression for the size in bits of the type @{b}short@{ub} on the target machine. If you don't define this, the default is half a word. (If this would be less than one storage unit, it is rounded up to one unit.) @{b}LONG_TYPE_SIZE@{ub} A C expression for the size in bits of the type @{b}long@{ub} on the target machine. If you don't define this, the default is one word. @{b}MAX_LONG_TYPE_SIZE@{ub} Maximum number for the size in bits of the type @{b}long@{ub} on the target machine. If this is undefined, the default is @{b}LONG_TYPE_SIZE@{ub}. Otherwise, it is the constant value that is the largest value that @{b}LONG_TYPE_SIZE@{ub} can have at run-time. This is used in @{b}cpp@{ub}. @{b}LONG_LONG_TYPE_SIZE@{ub} A C expression for the size in bits of the type @{b}long long@{ub} on the target machine. If you don't define this, the default is two words. If you want to support GNU Ada on your machine, the value of macro must be at least 64. @{b}CHAR_TYPE_SIZE@{ub} A C expression for the size in bits of the type @{b}char@{ub} on the target machine. If you don't define this, the default is one quarter of a word. (If this would be less than one storage unit, it is rounded up to one unit.) @{b}MAX_CHAR_TYPE_SIZE@{ub} Maximum number for the size in bits of the type @{b}char@{ub} on the target machine. If this is undefined, the default is @{b}CHAR_TYPE_SIZE@{ub}. Otherwise, it is the constant value that is the largest value that @{b}CHAR_TYPE_SIZE@{ub} can have at run-time. This is used in @{b}cpp@{ub}. @{b}FLOAT_TYPE_SIZE@{ub} A C expression for the size in bits of the type @{b}float@{ub} on the target machine. If you don't define this, the default is one word. @{b}DOUBLE_TYPE_SIZE@{ub} A C expression for the size in bits of the type @{b}double@{ub} on the target machine. If you don't define this, the default is two words. @{b}LONG_DOUBLE_TYPE_SIZE@{ub} A C expression for the size in bits of the type @{b}long double@{ub} on the target machine. If you don't define this, the default is two words. @{b}DEFAULT_SIGNED_CHAR@{ub} An expression whose value is 1 or 0, according to whether the type @{b}char@{ub} should be signed or unsigned by default. The user can always override this default with the options @{b}-fsigned-char@{ub} and @{b}-funsigned-char@{ub}. @{b}DEFAULT_SHORT_ENUMS@{ub} A C expression to determine whether to give an @{b}enum@{ub} type only as many bytes as it takes to represent the range of possible values of that type. A nonzero value means to do that; a zero value means all @{b}enum@{ub} types should be allocated like @{b}int@{ub}. If you don't define the macro, the default is 0. @{b}SIZE_TYPE@{ub} A C expression for a string describing the name of the data type to use for size values. The typedef name @{b}size_t@{ub} is defined using the contents of the string. The string can contain more than one keyword. If so, separate them with spaces, and write first any length keyword, then @{b}unsigned@{ub} if appropriate, and finally @{b}int@{ub}. The string must exactly match one of the data type names defined in the function @{b}init_decl_processing@{ub} in the file @{b}c-decl.c@{ub}. You may not omit @{b}int@{ub} or change the order--that would cause the compiler to crash on startup. If you don't define this macro, the default is @{b}"long unsigned int"@{ub}. @{b}PTRDIFF_TYPE@{ub} A C expression for a string describing the name of the data type to use for the result of subtracting two pointers. The typedef name @{b}ptrdiff_t@{ub} is defined using the contents of the string. See @{b}SIZE_TYPE@{ub} above for more information. If you don't define this macro, the default is @{b}"long int"@{ub}. @{b}WCHAR_TYPE@{ub} A C expression for a string describing the name of the data type to use for wide characters. The typedef name @{b}wchar_t@{ub} is defined using the contents of the string. See @{b}SIZE_TYPE@{ub} above for more information. If you don't define this macro, the default is @{b}"int"@{ub}. @{b}WCHAR_TYPE_SIZE@{ub} A C expression for the size in bits of the data type for wide characters. This is used in @{b}cpp@{ub}, which cannot make use of @{b}WCHAR_TYPE@{ub}. @{b}MAX_WCHAR_TYPE_SIZE@{ub} Maximum number for the size in bits of the data type for wide characters. If this is undefined, the default is @{b}WCHAR_TYPE_SIZE@{ub}. Otherwise, it is the constant value that is the largest value that @{b}WCHAR_TYPE_SIZE@{ub} can have at run-time. This is used in @{b}cpp@{ub}. @{b}OBJC_INT_SELECTORS@{ub} Define this macro if the type of Objective C selectors should be @{b}int@{ub}. If this macro is not defined, then selectors should have the type @{b}struct objc_selector *@{ub}. @{b}OBJC_SELECTORS_WITHOUT_LABELS@{ub} Define this macro if the compiler can group all the selectors together into a vector and use just one label at the beginning of the vector. Otherwise, the compiler must give each selector its own assembler label. On certain machines, it is important to have a separate label for each selector because this enables the linker to eliminate duplicate selectors. @{b}TARGET_BELL@{ub} A C constant expression for the integer value for escape sequence @{b}\\a@{ub}. @{b}TARGET_BS@{ub} @{b}TARGET_TAB@{ub} @{b}TARGET_NEWLINE@{ub} C constant expressions for the integer values for escape sequences @{b}\\b@{ub}, @{b}\\t@{ub} and @{b}\\n@{ub}. @{b}TARGET_VT@{ub} @{b}TARGET_FF@{ub} @{b}TARGET_CR@{ub} C constant expressions for the integer values for escape sequences @{b}\\v@{ub}, @{b}\\f@{ub} and @{b}\\r@{ub}. @EndNode @Node "Registers" "gcc.guide/Registers" @Next "Register Classes" @Prev "Type Layout" @Toc "Target Macros" Register Usage ============== This section explains how to describe what registers the target machine has, and how (in general) they can be used. The description of which registers a specific instruction can use is done with register classes; see @{"Register Classes" Link "Register Classes"}. For information on using registers to access a stack frame, see @{"Frame Registers" Link "Frame Registers"}. For passing values in registers, see @{"Register Arguments" Link "Register Arguments"}. For returning values in registers, see @{"Scalar Return" Link "Scalar Return"}. @{" Register Basics " Link "Register Basics"} Number and kinds of registers. @{" Allocation Order " Link "Allocation Order"} Order in which registers are allocated. @{" Values in Registers " Link "Values in Registers"} What kinds of values each reg can hold. @{" Leaf Functions " Link "Leaf Functions"} Renumbering registers for leaf functions. @{" Stack Registers " Link "Stack Registers"} Handling a register stack such as 80387. @{" Obsolete Register Macros " Link "Obsolete Register Macros"} Macros formerly used for the 80387. @EndNode @Node "Register Basics" "gcc.guide/Register Basics" @Next "Allocation Order" @Toc "Registers" Basic Characteristics of Registers ---------------------------------- Registers have various characteristics. @{b}FIRST_PSEUDO_REGISTER@{ub} Number of hardware registers known to the compiler. They receive numbers 0 through @{b}FIRST_PSEUDO_REGISTER-1@{ub}; thus, the first pseudo register's number really is assigned the number @{b}FIRST_PSEUDO_REGISTER@{ub}. @{b}FIXED_REGISTERS@{ub} An initializer that says which registers are used for fixed purposes all throughout the compiled code and are therefore not available for general allocation. These would include the stack pointer, the frame pointer (except on machines where that can be used as a general register when no frame pointer is needed), the program counter on machines where that is considered one of the addressable registers, and any other numbered register with a standard use. This information is expressed as a sequence of numbers, separated by commas and surrounded by braces. The @{I}N@{ui}th number is 1 if register @{I}N@{ui} is fixed, 0 otherwise. The table initialized from this macro, and the table initialized by the following one, may be overridden at run time either automatically, by the actions of the macro @{b}CONDITIONAL_REGISTER_USAGE@{ub}, or by the user with the command options @{b}-ffixed-@{I}REG@{ui}@{ub}, @{b}-fcall-used-@{I}REG@{ui}@{ub} and @{b}-fcall-saved-@{I}REG@{ui}@{ub}. @{b}CALL_USED_REGISTERS@{ub} Like @{b}FIXED_REGISTERS@{ub} but has 1 for each register that is clobbered (in general) by function calls as well as for fixed registers. This macro therefore identifies the registers that are not available for general allocation of values that must live across function calls. If a register has 0 in @{b}CALL_USED_REGISTERS@{ub}, the compiler automatically saves it on function entry and restores it on function exit, if the register is used within the function. @{b}CONDITIONAL_REGISTER_USAGE@{ub} Zero or more C statements that may conditionally modify two variables @{b}fixed_regs@{ub} and @{b}call_used_regs@{ub} (both of type @{b}char []@{ub}) after they have been initialized from the two preceding macros. This is necessary in case the fixed or call-clobbered registers depend on target flags. You need not define this macro if it has no work to do. If the usage of an entire class of registers depends on the target flags, you may indicate this to GCC by using this macro to modify @{b}fixed_regs@{ub} and @{b}call_used_regs@{ub} to 1 for each of the registers in the classes which should not be used by GCC. Also define the macro @{b}REG_CLASS_FROM_LETTER@{ub} to return @{b}NO_REGS@{ub} if it is called with a letter for a class that shouldn't be used. (However, if this class is not included in @{b}GENERAL_REGS@{ub} and all of the insn patterns whose constraints permit this class are controlled by target switches, then GCC will automatically avoid using these registers when the target switches are opposed to them.) @{b}NON_SAVING_SETJMP@{ub} If this macro is defined and has a nonzero value, it means that @{b}setjmp@{ub} and related functions fail to save the registers, or that @{b}longjmp@{ub} fails to restore them. To compensate, the compiler avoids putting variables in registers in functions that use @{b}setjmp@{ub}. @{b}INCOMING_REGNO (@{I}OUT@{ui})@{ub} Define this macro if the target machine has register windows. This C expression returns the register number as seen by the called function corresponding to the register number @{I}OUT@{ui} as seen by the calling function. Return @{I}OUT@{ui} if register number @{I}OUT@{ui} is not an outbound register. @{b}OUTGOING_REGNO (@{I}IN@{ui})@{ub} Define this macro if the target machine has register windows. This C expression returns the register number as seen by the calling function corresponding to the register number @{I}IN@{ui} as seen by the called function. Return @{I}IN@{ui} if register number @{I}IN@{ui} is not an inbound register. @EndNode @Node "Allocation Order" "gcc.guide/Allocation Order" @Next "Values in Registers" @Prev "Register Basics" @Toc "Registers" Order of Allocation of Registers -------------------------------- Registers are allocated in order. @{b}REG_ALLOC_ORDER@{ub} If defined, an initializer for a vector of integers, containing the numbers of hard registers in the order in which GNU CC should prefer to use them (from most preferred to least). If this macro is not defined, registers are used lowest numbered first (all else being equal). One use of this macro is on machines where the highest numbered registers must always be saved and the save-multiple-registers instruction supports only sequences of consecutive registers. On such machines, define @{b}REG_ALLOC_ORDER@{ub} to be an initializer that lists the highest numbered allocatable register first. @{b}ORDER_REGS_FOR_LOCAL_ALLOC@{ub} A C statement (sans semicolon) to choose the order in which to allocate hard registers for pseudo-registers local to a basic block. Store the desired register order in the array @{b}reg_alloc_order@{ub}. Element 0 should be the register to allocate first; element 1, the next register; and so on. The macro body should not assume anything about the contents of @{b}reg_alloc_order@{ub} before execution of the macro. On most machines, it is not necessary to define this macro. @EndNode @Node "Values in Registers" "gcc.guide/Values in Registers" @Next "Leaf Functions" @Prev "Allocation Order" @Toc "Registers" How Values Fit in Registers --------------------------- This section discusses the macros that describe which kinds of values (specifically, which machine modes) each register can hold, and how many consecutive registers are needed for a given mode. @{b}HARD_REGNO_NREGS (@{I}REGNO@{ui}, @{I}MODE@{ui})@{ub} A C expression for the number of consecutive hard registers, starting at register number @{I}REGNO@{ui}, required to hold a value of mode @{I}MODE@{ui}. On a machine where all registers are exactly one word, a suitable definition of this macro is #define HARD_REGNO_NREGS(REGNO, MODE) \\ ((GET_MODE_SIZE (MODE) + UNITS_PER_WORD - 1) \\ / UNITS_PER_WORD)) @{b}HARD_REGNO_MODE_OK (@{I}REGNO@{ui}, @{I}MODE@{ui})@{ub} A C expression that is nonzero if it is permissible to store a value of mode @{I}MODE@{ui} in hard register number @{I}REGNO@{ui} (or in several registers starting with that one). For a machine where all registers are equivalent, a suitable definition is #define HARD_REGNO_MODE_OK(REGNO, MODE) 1 It is not necessary for this macro to check for the numbers of fixed registers, because the allocation mechanism considers them to be always occupied. On some machines, double-precision values must be kept in even/odd register pairs. The way to implement that is to define this macro to reject odd register numbers for such modes. The minimum requirement for a mode to be OK in a register is that the @{b}mov@{I}MODE@{ui}@{ub} instruction pattern support moves between the register and any other hard register for which the mode is OK; and that moving a value into the register and back out not alter it. Since the same instruction used to move @{b}SImode@{ub} will work for all narrower integer modes, it is not necessary on any machine for @{b}HARD_REGNO_MODE_OK@{ub} to distinguish between these modes, provided you define patterns @{b}movhi@{ub}, etc., to take advantage of this. This is useful because of the interaction between @{b}HARD_REGNO_MODE_OK@{ub} and @{b}MODES_TIEABLE_P@{ub}; it is very desirable for all integer modes to be tieable. Many machines have special registers for floating point arithmetic. Often people assume that floating point machine modes are allowed only in floating point registers. This is not true. Any registers that can hold integers can safely @{i}hold@{ui} a floating point machine mode, whether or not floating arithmetic can be done on it in those registers. Integer move instructions can be used to move the values. On some machines, though, the converse is true: fixed-point machine modes may not go in floating registers. This is true if the floating registers normalize any value stored in them, because storing a non-floating value there would garble it. In this case, @{b}HARD_REGNO_MODE_OK@{ub} should reject fixed-point machine modes in floating registers. But if the floating registers do not automatically normalize, if you can store any bit pattern in one and retrieve it unchanged without a trap, then any machine mode may go in a floating register, so you can define this macro to say so. The primary significance of special floating registers is rather that they are the registers acceptable in floating point arithmetic instructions. However, this is of no concern to @{b}HARD_REGNO_MODE_OK@{ub}. You handle it by writing the proper constraints for those instructions. On some machines, the floating registers are especially slow to access, so that it is better to store a value in a stack frame than in such a register if floating point arithmetic is not being done. As long as the floating registers are not in class @{b}GENERAL_REGS@{ub}, they will not be used unless some pattern's constraint asks for one. @{b}MODES_TIEABLE_P (@{I}MODE1@{ui}, @{I}MODE2@{ui})@{ub} A C expression that is nonzero if it is desirable to choose register allocation so as to avoid move instructions between a value of mode @{I}MODE1@{ui} and a value of mode @{I}MODE2@{ui}. If @{b}HARD_REGNO_MODE_OK (@{I}R@{ui}, @{I}MODE1@{ui})@{ub} and @{b}HARD_REGNO_MODE_OK (@{I}R@{ui}, @{I}MODE2@{ui})@{ub} are ever different for any @{I}R@{ui}, then @{b}MODES_TIEABLE_P (@{I}MODE1@{ui}, @{I}MODE2@{ui})@{ub} must be zero. @EndNode @Node "Leaf Functions" "gcc.guide/Leaf Functions" @Next "Stack Registers" @Prev "Values in Registers" @Toc "Registers" Handling Leaf Functions ----------------------- On some machines, a leaf function (i.e., one which makes no calls) can run more efficiently if it does not make its own register window. Often this means it is required to receive its arguments in the registers where they are passed by the caller, instead of the registers where they would normally arrive. The special treatment for leaf functions generally applies only when other conditions are met; for example, often they may use only those registers for its own variables and temporaries. We use the term "leaf function" to mean a function that is suitable for this special handling, so that functions with no calls are not necessarily "leaf functions". GNU CC assigns register numbers before it knows whether the function is suitable for leaf function treatment. So it needs to renumber the registers in order to output a leaf function. The following macros accomplish this. @{b}LEAF_REGISTERS@{ub} A C initializer for a vector, indexed by hard register number, which contains 1 for a register that is allowable in a candidate for leaf function treatment. If leaf function treatment involves renumbering the registers, then the registers marked here should be the ones before renumbering--those that GNU CC would ordinarily allocate. The registers which will actually be used in the assembler code, after renumbering, should not be marked with 1 in this vector. Define this macro only if the target machine offers a way to optimize the treatment of leaf functions. @{b}LEAF_REG_REMAP (@{I}REGNO@{ui})@{ub} A C expression whose value is the register number to which @{I}REGNO@{ui} should be renumbered, when a function is treated as a leaf function. If @{I}REGNO@{ui} is a register number which should not appear in a leaf function before renumbering, then the expression should yield -1, which will cause the compiler to abort. Define this macro only if the target machine offers a way to optimize the treatment of leaf functions, and registers need to be renumbered to do this. Normally, @{b}FUNCTION_PROLOGUE@{ub} and @{b}FUNCTION_EPILOGUE@{ub} must treat leaf functions specially. It can test the C variable @{b}leaf_function@{ub} which is nonzero for leaf functions. (The variable @{b}leaf_function@{ub} is defined only if @{b}LEAF_REGISTERS@{ub} is defined.) @EndNode @Node "Stack Registers" "gcc.guide/Stack Registers" @Next "Obsolete Register Macros" @Prev "Leaf Functions" @Toc "Registers" Registers That Form a Stack --------------------------- There are special features to handle computers where some of the "registers" form a stack, as in the 80387 coprocessor for the 80386. Stack registers are normally written by pushing onto the stack, and are numbered relative to the top of the stack. Currently, GNU CC can only handle one group of stack-like registers, and they must be consecutively numbered. @{b}STACK_REGS@{ub} Define this if the machine has any stack-like registers. @{b}FIRST_STACK_REG@{ub} The number of the first stack-like register. This one is the top of the stack. @{b}LAST_STACK_REG@{ub} The number of the last stack-like register. This one is the bottom of the stack. @EndNode @Node "Obsolete Register Macros" "gcc.guide/Obsolete Register Macros" @Prev "Stack Registers" @Toc "Registers" Obsolete Macros for Controlling Register Usage ---------------------------------------------- These features do not work very well. They exist because they used to be required to generate correct code for the 80387 coprocessor of the 80386. They are no longer used by that machine description and may be removed in a later version of the compiler. Don't use them! @{b}OVERLAPPING_REGNO_P (@{I}REGNO@{ui})@{ub} If defined, this is a C expression whose value is nonzero if hard register number @{I}REGNO@{ui} is an overlapping register. This means a hard register which overlaps a hard register with a different number. (Such overlap is undesirable, but occasionally it allows a machine to be supported which otherwise could not be.) This macro must return nonzero for @{i}all@{ui} the registers which overlap each other. GNU CC can use an overlapping register only in certain limited ways. It can be used for allocation within a basic block, and may be spilled for reloading; that is all. If this macro is not defined, it means that none of the hard registers overlap each other. This is the usual situation. @{b}INSN_CLOBBERS_REGNO_P (@{I}INSN@{ui}, @{I}REGNO@{ui})@{ub} If defined, this is a C expression whose value should be nonzero if the insn @{I}INSN@{ui} has the effect of mysteriously clobbering the contents of hard register number @{I}REGNO@{ui}. By "mysterious" we mean that the insn's RTL expression doesn't describe such an effect. If this macro is not defined, it means that no insn clobbers registers mysteriously. This is the usual situation; all else being equal, it is best for the RTL expression to show all the activity. @{b}PRESERVE_DEATH_INFO_REGNO_P (@{I}REGNO@{ui})@{ub} If defined, this is a C expression whose value is nonzero if accurate @{b}REG_DEAD@{ub} notes are needed for hard register number @{I}REGNO@{ui} at the time of outputting the assembler code. When this is so, a few optimizations that take place after register allocation and could invalidate the death notes are not done when this register is involved. You would arrange to preserve death info for a register when some of the code in the machine description which is executed to write the assembler code looks at the death notes. This is necessary only when the actual hardware feature which GNU CC thinks of as a register is not actually a register of the usual sort. (It might, for example, be a hardware stack.) If this macro is not defined, it means that no death notes need to be preserved. This is the usual situation. @EndNode @Node "Register Classes" "gcc.guide/Register Classes" @Next "Stack and Calling" @Prev "Registers" @Toc "Target Macros" Register Classes ================ On many machines, the numbered registers are not all equivalent. For example, certain registers may not be allowed for indexed addressing; certain registers may not be allowed in some instructions. These machine restrictions are described to the compiler using @{i}register classes@{ui}. You define a number of register classes, giving each one a name and saying which of the registers belong to it. Then you can specify register classes that are allowed as operands to particular instruction patterns. In general, each register will belong to several classes. In fact, one class must be named @{b}ALL_REGS@{ub} and contain all the registers. Another class must be named @{b}NO_REGS@{ub} and contain no registers. Often the union of two classes will be another class; however, this is not required. One of the classes must be named @{b}GENERAL_REGS@{ub}. There is nothing terribly special about the name, but the operand constraint letters @{b}r@{ub} and @{b}g@{ub} specify this class. If @{b}GENERAL_REGS@{ub} is the same as @{b}ALL_REGS@{ub}, just define it as a macro which expands to @{b}ALL_REGS@{ub}. Order the classes so that if class @{I}X@{ui} is contained in class @{I}Y@{ui} then @{I}X@{ui} has a lower class number than @{I}Y@{ui}. The way classes other than @{b}GENERAL_REGS@{ub} are specified in operand constraints is through machine-dependent operand constraint letters. You can define such letters to correspond to various classes, then use them in operand constraints. You should define a class for the union of two classes whenever some instruction allows both classes. For example, if an instruction allows either a floating point (coprocessor) register or a general register for a certain operand, you should define a class @{b}FLOAT_OR_GENERAL_REGS@{ub} which includes both of them. Otherwise you will get suboptimal code. You must also specify certain redundant information about the register classes: for each class, which classes contain it and which ones are contained in it; for each pair of classes, the largest class contained in their union. When a value occupying several consecutive registers is expected in a certain class, all the registers used must belong to that class. Therefore, register classes cannot be used to enforce a requirement for a register pair to start with an even-numbered register. The way to specify this requirement is with @{b}HARD_REGNO_MODE_OK@{ub}. Register classes used for input-operands of bitwise-and or shift instructions have a special requirement: each such class must have, for each fixed-point machine mode, a subclass whose registers can transfer that mode to or from memory. For example, on some machines, the operations for single-byte values (@{b}QImode@{ub}) are limited to certain registers. When this is so, each register class that is used in a bitwise-and or shift instruction must have a subclass consisting of registers from which single-byte values can be loaded or stored. This is so that @{b}PREFERRED_RELOAD_CLASS@{ub} can always have a possible value to return. @{b}enum reg_class@{ub} An enumeral type that must be defined with all the register class names as enumeral values. @{b}NO_REGS@{ub} must be first. @{b}ALL_REGS@{ub} must be the last register class, followed by one more enumeral value, @{b}LIM_REG_CLASSES@{ub}, which is not a register class but rather tells how many classes there are. Each register class has a number, which is the value of casting the class name to type @{b}int@{ub}. The number serves as an index in many of the tables described below. @{b}N_REG_CLASSES@{ub} The number of distinct register classes, defined as follows: #define N_REG_CLASSES (int) LIM_REG_CLASSES @{b}REG_CLASS_NAMES@{ub} An initializer containing the names of the register classes as C string constants. These names are used in writing some of the debugging dumps. @{b}REG_CLASS_CONTENTS@{ub} An initializer containing the contents of the register classes, as integers which are bit masks. The @{I}N@{ui}th integer specifies the contents of class @{I}N@{ui}. The way the integer @{I}MASK@{ui} is interpreted is that register @{I}R@{ui} is in the class if @{b}@{I}MASK@{ui} & (1 << @{I}R@{ui})@{ub} is 1. When the machine has more than 32 registers, an integer does not suffice. Then the integers are replaced by sub-initializers, braced groupings containing several integers. Each sub-initializer must be suitable as an initializer for the type @{b}HARD_REG_SET@{ub} which is defined in @{b}hard-reg-set.h@{ub}. @{b}REGNO_REG_CLASS (@{I}REGNO@{ui})@{ub} A C expression whose value is a register class containing hard register @{I}REGNO@{ui}. In general there is more than one such class; choose a class which is @{i}minimal@{ui}, meaning that no smaller class also contains the register. @{b}BASE_REG_CLASS@{ub} A macro whose definition is the name of the class to which a valid base register must belong. A base register is one used in an address which is the register value plus a displacement. @{b}INDEX_REG_CLASS@{ub} A macro whose definition is the name of the class to which a valid index register must belong. An index register is one used in an address where its value is either multiplied by a scale factor or added to another register (as well as added to a displacement). @{b}REG_CLASS_FROM_LETTER (@{I}CHAR@{ui})@{ub} A C expression which defines the machine-dependent operand constraint letters for register classes. If @{I}CHAR@{ui} is such a letter, the value should be the register class corresponding to it. Otherwise, the value should be @{b}NO_REGS@{ub}. The register letter @{b}r@{ub}, corresponding to class @{b}GENERAL_REGS@{ub}, will not be passed to this macro; you do not need to handle it. @{b}REGNO_OK_FOR_BASE_P (@{I}NUM@{ui})@{ub} A C expression which is nonzero if register number @{I}NUM@{ui} is suitable for use as a base register in operand addresses. It may be either a suitable hard register or a pseudo register that has been allocated such a hard register. @{b}REGNO_OK_FOR_INDEX_P (@{I}NUM@{ui})@{ub} A C expression which is nonzero if register number @{I}NUM@{ui} is suitable for use as an index register in operand addresses. It may be either a suitable hard register or a pseudo register that has been allocated such a hard register. The difference between an index register and a base register is that the index register may be scaled. If an address involves the sum of two registers, neither one of them scaled, then either one may be labeled the "base" and the other the "index"; but whichever labeling is used must fit the machine's constraints of which registers may serve in each capacity. The compiler will try both labelings, looking for one that is valid, and will reload one or both registers only if neither labeling works. @{b}PREFERRED_RELOAD_CLASS (@{I}X@{ui}, @{I}CLASS@{ui})@{ub} A C expression that places additional restrictions on the register class to use when it is necessary to copy value @{I}X@{ui} into a register in class @{I}CLASS@{ui}. The value is a register class; perhaps @{I}CLASS@{ui}, or perhaps another, smaller class. On many machines, the following definition is safe: #define PREFERRED_RELOAD_CLASS(X,CLASS) CLASS Sometimes returning a more restrictive class makes better code. For example, on the 68000, when @{I}X@{ui} is an integer constant that is in range for a @{b}moveq@{ub} instruction, the value of this macro is always @{b}DATA_REGS@{ub} as long as @{I}CLASS@{ui} includes the data registers. Requiring a data register guarantees that a @{b}moveq@{ub} will be used. If @{I}X@{ui} is a @{b}const_double@{ub}, by returning @{b}NO_REGS@{ub} you can force @{I}X@{ui} into a memory constant. This is useful on certain machines where immediate floating values cannot be loaded into certain kinds of registers. @{b}PREFERRED_OUTPUT_RELOAD_CLASS (@{I}X@{ui}, @{I}CLASS@{ui})@{ub} Like @{b}PREFERRED_RELOAD_CLASS@{ub}, but for output reloads instead of input reloads. If you don't define this macro, the default is to use @{I}CLASS@{ui}, unchanged. @{b}LIMIT_RELOAD_CLASS (@{I}MODE@{ui}, @{I}CLASS@{ui})@{ub} A C expression that places additional restrictions on the register class to use when it is necessary to be able to hold a value of mode @{I}MODE@{ui} in a reload register for which class @{I}CLASS@{ui} would ordinarily be used. Unlike @{b}PREFERRED_RELOAD_CLASS@{ub}, this macro should be used when there are certain modes that simply can't go in certain reload classes. The value is a register class; perhaps @{I}CLASS@{ui}, or perhaps another, smaller class. Don't define this macro unless the target machine has limitations which require the macro to do something nontrivial. @{b}SECONDARY_RELOAD_CLASS (@{I}CLASS@{ui}, @{I}MODE@{ui}, @{I}X@{ui})@{ub} @{b}SECONDARY_INPUT_RELOAD_CLASS (@{I}CLASS@{ui}, @{I}MODE@{ui}, @{I}X@{ui})@{ub} @{b}SECONDARY_OUTPUT_RELOAD_CLASS (@{I}CLASS@{ui}, @{I}MODE@{ui}, @{I}X@{ui})@{ub} Many machines have some registers that cannot be copied directly to or from memory or even from other types of registers. An example is the @{b}MQ@{ub} register, which on most machines, can only be copied to or from general registers, but not memory. Some machines allow copying all registers to and from memory, but require a scratch register for stores to some memory locations (e.g., those with symbolic address on the RT, and those with certain symbolic address on the Sparc when compiling PIC). In some cases, both an intermediate and a scratch register are required. You should define these macros to indicate to the reload phase that it may need to allocate at least one register for a reload in addition to the register to contain the data. Specifically, if copying @{I}X@{ui} to a register @{I}CLASS@{ui} in @{I}MODE@{ui} requires an intermediate register, you should define @{b}SECONDARY_INPUT_RELOAD_CLASS@{ub} to return the largest register class all of whose registers can be used as intermediate registers or scratch registers. If copying a register @{I}CLASS@{ui} in @{I}MODE@{ui} to @{I}X@{ui} requires an intermediate or scratch register, @{b}SECONDARY_OUTPUT_RELOAD_CLASS@{ub} should be defined to return the largest register class required. If the requirements for input and output reloads are the same, the macro @{b}SECONDARY_RELOAD_CLASS@{ub} should be used instead of defining both macros identically. The values returned by these macros are often @{b}GENERAL_REGS@{ub}. Return @{b}NO_REGS@{ub} if no spare register is needed; i.e., if @{I}X@{ui} can be directly copied to or from a register of @{I}CLASS@{ui} in @{I}MODE@{ui} without requiring a scratch register. Do not define this macro if it would always return @{b}NO_REGS@{ub}. If a scratch register is required (either with or without an intermediate register), you should define patterns for @{b}reload_in@{I}M@{ui}@{ub} or @{b}reload_out@{I}M@{ui}@{ub}, as required (see @{"Standard Names" Link "Standard Names"}. These patterns, which will normally be implemented with a @{b}define_expand@{ub}, should be similar to the @{b}mov@{I}M@{ui}@{ub} patterns, except that operand 2 is the scratch register. Define constraints for the reload register and scratch register that contain a single register class. If the original reload register (whose class is @{I}CLASS@{ui}) can meet the constraint given in the pattern, the value returned by these macros is used for the class of the scratch register. Otherwise, two additional reload registers are required. Their classes are obtained from the constraints in the insn pattern. @{I}X@{ui} might be a pseudo-register or a @{b}subreg@{ub} of a pseudo-register, which could either be in a hard register or in memory. Use @{b}true_regnum@{ub} to find out; it will return -1 if the pseudo is in memory and the hard register number if it is in a register. These macros should not be used in the case where a particular class of registers can only be copied to memory and not to another class of registers. In that case, secondary reload registers are not needed and would not be helpful. Instead, a stack location must be used to perform the copy and the @{b}mov@{I}M@{ui}@{ub} pattern should use memory as a intermediate storage. This case often occurs between floating-point and general registers. @{b}SECONDARY_MEMORY_NEEDED (@{I}CLASS1@{ui}, @{I}CLASS2@{ui}, @{I}M@{ui})@{ub} Certain machines have the property that some registers cannot be copied to some other registers without using memory. Define this macro on those machines to be a C expression that is non-zero if objects of mode @{I}M@{ui} in registers of @{I}CLASS1@{ui} can only be copied to registers of class @{I}CLASS2@{ui} by storing a register of @{I}CLASS1@{ui} into memory and loading that memory location into a register of @{I}CLASS2@{ui}. Do not define this macro if its value would always be zero. @{b}SECONDARY_MEMORY_NEEDED_RTX (@{I}MODE@{ui})@{ub} Normally when @{b}SECONDARY_MEMORY_NEEDED@{ub} is defined, the compiler allocates a stack slot for a memory location needed for register copies. If this macro is defined, the compiler instead uses the memory location defined by this macro. Do not define this macro if you do not define @{b}SECONDARY_MEMORY_NEEDED@{ub}. @{b}SECONDARY_MEMORY_NEEDED_MODE (@{I}MODE@{ui})@{ub} When the compiler needs a secondary memory location to copy between two registers of mode @{I}MODE@{ui}, it normally allocates sufficient memory to hold a quantity of @{b}BITS_PER_WORD@{ub} bits and performs the store and load operations in a mode that many bits wide and whose class is the same as that of @{I}MODE@{ui}. This is right thing to do on most machines because it ensures that all bits of the register are copied and prevents accesses to the registers in a narrower mode, which some machines prohibit for floating-point registers. However, this default behavior is not correct on some machines, such as the DEC Alpha, that store short integers in floating-point registers differently than in integer registers. On those machines, the default widening will not work correctly and you must define this macro to suppress that widening in some cases. See the file @{b}alpha.h@{ub} for details. Do not define this macro if you do not define @{b}SECONDARY_MEMORY_NEEDED@{ub} or if widening @{I}MODE@{ui} to a mode that is @{b}BITS_PER_WORD@{ub} bits wide is correct for your machine. @{b}SMALL_REGISTER_CLASSES@{ub} Normally the compiler avoids choosing registers that have been explicitly mentioned in the rtl as spill registers (these registers are normally those used to pass parameters and return values). However, some machines have so few registers of certain classes that there would not be enough registers to use as spill registers if this were done. Define @{b}SMALL_REGISTER_CLASSES@{ub} on these machines. When it is defined, the compiler allows registers explicitly used in the rtl to be used as spill registers but avoids extending the lifetime of these registers. It is always safe to define this macro, but if you unnecessarily define it, you will reduce the amount of optimizations that can be performed in some cases. If you do not define this macro when it is required, the compiler will run out of spill registers and print a fatal error message. For most machines, you should not define this macro. @{b}CLASS_LIKELY_SPILLED_P (@{I}CLASS@{ui})@{ub} A C expression whose value is nonzero if pseudos that have been assigned to registers of class @{I}CLASS@{ui} would likely be spilled because registers of @{I}CLASS@{ui} are needed for spill registers. The default value of this macro returns 1 if @{I}CLASS@{ui} has exactly one register and zero otherwise. On most machines, this default should be used. Only define this macro to some other expression if pseudo allocated by @{b}local-alloc.c@{ub} end up in memory because their hard registers were needed for spill registers. If this macro returns nonzero for those classes, those pseudos will only be allocated by @{b}global.c@{ub}, which knows how to reallocate the pseudo to another register. If there would not be another register available for reallocation, you should not change the definition of this macro since the only effect of such a definition would be to slow down register allocation. @{b}CLASS_MAX_NREGS (@{I}CLASS@{ui}, @{I}MODE@{ui})@{ub} A C expression for the maximum number of consecutive registers of class @{I}CLASS@{ui} needed to hold a value of mode @{I}MODE@{ui}. This is closely related to the macro @{b}HARD_REGNO_NREGS@{ub}. In fact, the value of the macro @{b}CLASS_MAX_NREGS (@{I}CLASS@{ui}, @{I}MODE@{ui})@{ub} should be the maximum value of @{b}HARD_REGNO_NREGS (@{I}REGNO@{ui}, @{I}MODE@{ui})@{ub} for all @{I}REGNO@{ui} values in the class @{I}CLASS@{ui}. This macro helps control the handling of multiple-word values in the reload pass. @{b}CLASS_CANNOT_CHANGE_SIZE@{ub} If defined, a C expression for a class that contains registers which the compiler must always access in a mode that is the same size as the mode in which it loaded the register. For the example, loading 32-bit integer or floating-point objects into floating-point registers on the Alpha extends them to 64-bits. Therefore loading a 64-bit object and then storing it as a 32-bit object does not store the low-order 32-bits, as would be the case for a normal register. Therefore, @{b}alpha.h@{ub} defines this macro as @{b}FLOAT_REGS@{ub}. Three other special macros describe which operands fit which constraint letters. @{b}CONST_OK_FOR_LETTER_P (@{I}VALUE@{ui}, @{I}C@{ui})@{ub} A C expression that defines the machine-dependent operand constraint letters that specify particular ranges of integer values. If @{I}C@{ui} is one of those letters, the expression should check that @{I}VALUE@{ui}, an integer, is in the appropriate range and return 1 if so, 0 otherwise. If @{I}C@{ui} is not one of those letters, the value should be 0 regardless of @{I}VALUE@{ui}. @{b}CONST_DOUBLE_OK_FOR_LETTER_P (@{I}VALUE@{ui}, @{I}C@{ui})@{ub} A C expression that defines the machine-dependent operand constraint letters that specify particular ranges of @{b}const_double@{ub} values. If @{I}C@{ui} is one of those letters, the expression should check that @{I}VALUE@{ui}, an RTX of code @{b}const_double@{ub}, is in the appropriate range and return 1 if so, 0 otherwise. If @{I}C@{ui} is not one of those letters, the value should be 0 regardless of @{I}VALUE@{ui}. @{b}const_double@{ub} is used for all floating-point constants and for @{b}DImode@{ub} fixed-point constants. A given letter can accept either or both kinds of values. It can use @{b}GET_MODE@{ub} to distinguish between these kinds. @{b}EXTRA_CONSTRAINT (@{I}VALUE@{ui}, @{I}C@{ui})@{ub} A C expression that defines the optional machine-dependent constraint letters that can be used to segregate specific types of operands, usually memory references, for the target machine. Normally this macro will not be defined. If it is required for a particular target machine, it should return 1 if @{I}VALUE@{ui} corresponds to the operand type represented by the constraint letter @{I}C@{ui}. If @{I}C@{ui} is not defined as an extra constraint, the value returned should be 0 regardless of @{I}VALUE@{ui}. For example, on the ROMP, load instructions cannot have their output in r0 if the memory reference contains a symbolic address. Constraint letter @{b}Q@{ub} is defined as representing a memory address that does @{i}not@{ui} contain a symbolic address. An alternative is specified with a @{b}Q@{ub} constraint on the input and @{b}r@{ub} on the output. The next alternative specifies @{b}m@{ub} on the input and a register class that does not include r0 on the output. @EndNode @Node "Stack and Calling" "gcc.guide/Stack and Calling" @Next "Varargs" @Prev "Register Classes" @Toc "Target Macros" Stack Layout and Calling Conventions ==================================== This describes the stack layout and calling conventions. @{" Frame Layout " Link "Frame Layout"} @{" Frame Registers " Link "Frame Registers"} @{" Elimination " Link "Elimination"} @{" Stack Arguments " Link "Stack Arguments"} @{" Register Arguments " Link "Register Arguments"} @{" Scalar Return " Link "Scalar Return"} @{" Aggregate Return " Link "Aggregate Return"} @{" Caller Saves " Link "Caller Saves"} @{" Function Entry " Link "Function Entry"} @{" Profiling " Link "Profiling"} @EndNode @Node "Frame Layout" "gcc.guide/Frame Layout" @Next "Frame Registers" @Toc "Stack and Calling" Basic Stack Layout ------------------ Here is the basic stack layout. @{b}STACK_GROWS_DOWNWARD@{ub} Define this macro if pushing a word onto the stack moves the stack pointer to a smaller address. When we say, "define this macro if ...," it means that the compiler checks this macro only with @{b}#ifdef@{ub} so the precise definition used does not matter. @{b}FRAME_GROWS_DOWNWARD@{ub} Define this macro if the addresses of local variable slots are at negative offsets from the frame pointer. @{b}ARGS_GROW_DOWNWARD@{ub} Define this macro if successive arguments to a function occupy decreasing addresses on the stack. @{b}STARTING_FRAME_OFFSET@{ub} Offset from the frame pointer to the first local variable slot to be allocated. If @{b}FRAME_GROWS_DOWNWARD@{ub}, find the next slot's offset by subtracting the first slot's length from @{b}STARTING_FRAME_OFFSET@{ub}. Otherwise, it is found by adding the length of the first slot to the value @{b}STARTING_FRAME_OFFSET@{ub}. @{b}STACK_POINTER_OFFSET@{ub} Offset from the stack pointer register to the first location at which outgoing arguments are placed. If not specified, the default value of zero is used. This is the proper value for most machines. If @{b}ARGS_GROW_DOWNWARD@{ub}, this is the offset to the location above the first location at which outgoing arguments are placed. @{b}FIRST_PARM_OFFSET (@{I}FUNDECL@{ui})@{ub} Offset from the argument pointer register to the first argument's address. On some machines it may depend on the data type of the function. If @{b}ARGS_GROW_DOWNWARD@{ub}, this is the offset to the location above the first argument's address. @{b}STACK_DYNAMIC_OFFSET (@{I}FUNDECL@{ui})@{ub} Offset from the stack pointer register to an item dynamically allocated on the stack, e.g., by @{b}alloca@{ub}. The default value for this macro is @{b}STACK_POINTER_OFFSET@{ub} plus the length of the outgoing arguments. The default is correct for most machines. See @{b}function.c@{ub} for details. @{b}DYNAMIC_CHAIN_ADDRESS (@{I}FRAMEADDR@{ui})@{ub} A C expression whose value is RTL representing the address in a stack frame where the pointer to the caller's frame is stored. Assume that @{I}FRAMEADDR@{ui} is an RTL expression for the address of the stack frame itself. If you don't define this macro, the default is to return the value of @{I}FRAMEADDR@{ui}--that is, the stack frame address is also the address of the stack word that points to the previous frame. @{b}SETUP_FRAME_ADDRESSES ()@{ub} If defined, a C expression that produces the machine-specific code to setup the stack so that arbitrary frames can be accessed. For example, on the Sparc, we must flush all of the register windows to the stack before we can access arbitrary stack frames. This macro will seldom need to be defined. @{b}RETURN_ADDR_RTX (@{I}COUNT@{ui}, @{I}FRAMEADDR@{ui})@{ub} A C expression whose value is RTL representing the value of the return address for the frame @{I}COUNT@{ui} steps up from the current frame. @{I}FRAMEADDR@{ui} is the frame pointer of the @{I}COUNT@{ui} frame, or the frame pointer of the @{I}COUNT@{ui} - 1 frame if @{b}RETURN_ADDR_IN_PREVIOUS_FRAME@{ub} is defined. @{b}RETURN_ADDR_IN_PREVIOUS_FRAME@{ub} Define this if the return address of a particular stack frame is accessed from the frame pointer of the previous stack frame. @EndNode @Node "Frame Registers" "gcc.guide/Frame Registers" @Next "Elimination" @Prev "Frame Layout" @Toc "Stack and Calling" Registers That Address the Stack Frame -------------------------------------- This discusses registers that address the stack frame. @{b}STACK_POINTER_REGNUM@{ub} The register number of the stack pointer register, which must also be a fixed register according to @{b}FIXED_REGISTERS@{ub}. On most machines, the hardware determines which register this is. @{b}FRAME_POINTER_REGNUM@{ub} The register number of the frame pointer register, which is used to access automatic variables in the stack frame. On some machines, the hardware determines which register this is. On other machines, you can choose any register you wish for this purpose. @{b}HARD_FRAME_POINTER_REGNUM@{ub} On some machines the offset between the frame pointer and starting offset of the automatic variables is not known until after register allocation has been done (for example, because the saved registers are between these two locations). On those machines, define @{b}FRAME_POINTER_REGNUM@{ub} the number of a special, fixed register to be used internally until the offset is known, and define @{b}HARD_FRAME_POINTER_REGNUM@{ub} to be actual the hard register number used for the frame pointer. You should define this macro only in the very rare circumstances when it is not possible to calculate the offset between the frame pointer and the automatic variables until after register allocation has been completed. When this macro is defined, you must also indicate in your definition of @{b}ELIMINABLE_REGS@{ub} how to eliminate @{b}FRAME_POINTER_REGNUM@{ub} into either @{b}HARD_FRAME_POINTER_REGNUM@{ub} or @{b}STACK_POINTER_REGNUM@{ub}. Do not define this macro if it would be the same as @{b}FRAME_POINTER_REGNUM@{ub}. @{b}ARG_POINTER_REGNUM@{ub} The register number of the arg pointer register, which is used to access the function's argument list. On some machines, this is the same as the frame pointer register. On some machines, the hardware determines which register this is. On other machines, you can choose any register you wish for this purpose. If this is not the same register as the frame pointer register, then you must mark it as a fixed register according to @{b}FIXED_REGISTERS@{ub}, or arrange to be able to eliminate it (see @{"Elimination" Link "Elimination"}). @{b}STATIC_CHAIN_REGNUM@{ub} @{b}STATIC_CHAIN_INCOMING_REGNUM@{ub} Register numbers used for passing a function's static chain pointer. If register windows are used, the register number as seen by the called function is @{b}STATIC_CHAIN_INCOMING_REGNUM@{ub}, while the register number as seen by the calling function is @{b}STATIC_CHAIN_REGNUM@{ub}. If these registers are the same, @{b}STATIC_CHAIN_INCOMING_REGNUM@{ub} need not be defined. The static chain register need not be a fixed register. If the static chain is passed in memory, these macros should not be defined; instead, the next two macros should be defined. @{b}STATIC_CHAIN@{ub} @{b}STATIC_CHAIN_INCOMING@{ub} If the static chain is passed in memory, these macros provide rtx giving @{b}mem@{ub} expressions that denote where they are stored. @{b}STATIC_CHAIN@{ub} and @{b}STATIC_CHAIN_INCOMING@{ub} give the locations as seen by the calling and called functions, respectively. Often the former will be at an offset from the stack pointer and the latter at an offset from the frame pointer. The variables @{b}stack_pointer_rtx@{ub}, @{b}frame_pointer_rtx@{ub}, and @{b}arg_pointer_rtx@{ub} will have been initialized prior to the use of these macros and should be used to refer to those items. If the static chain is passed in a register, the two previous macros should be defined instead. @EndNode @Node "Elimination" "gcc.guide/Elimination" @Next "Stack Arguments" @Prev "Frame Registers" @Toc "Stack and Calling" Eliminating Frame Pointer and Arg Pointer ----------------------------------------- This is about eliminating the frame pointer and arg pointer. @{b}FRAME_POINTER_REQUIRED@{ub} A C expression which is nonzero if a function must have and use a frame pointer. This expression is evaluated in the reload pass. If its value is nonzero the function will have a frame pointer. The expression can in principle examine the current function and decide according to the facts, but on most machines the constant 0 or the constant 1 suffices. Use 0 when the machine allows code to be generated with no frame pointer, and doing so saves some time or space. Use 1 when there is no possible advantage to avoiding a frame pointer. In certain cases, the compiler does not know how to produce valid code without a frame pointer. The compiler recognizes those cases and automatically gives the function a frame pointer regardless of what @{b}FRAME_POINTER_REQUIRED@{ub} says. You don't need to worry about them. In a function that does not require a frame pointer, the frame pointer register can be allocated for ordinary usage, unless you mark it as a fixed register. See @{b}FIXED_REGISTERS@{ub} for more information. @{b}INITIAL_FRAME_POINTER_OFFSET (@{I}DEPTH-VAR@{ui})@{ub} A C statement to store in the variable @{I}DEPTH-VAR@{ui} the difference between the frame pointer and the stack pointer values immediately after the function prologue. The value would be computed from information such as the result of @{b}get_frame_size ()@{ub} and the tables of registers @{b}regs_ever_live@{ub} and @{b}call_used_regs@{ub}. If @{b}ELIMINABLE_REGS@{ub} is defined, this macro will be not be used and need not be defined. Otherwise, it must be defined even if @{b}FRAME_POINTER_REQUIRED@{ub} is defined to always be true; in that case, you may set @{I}DEPTH-VAR@{ui} to anything. @{b}ELIMINABLE_REGS@{ub} If defined, this macro specifies a table of register pairs used to eliminate unneeded registers that point into the stack frame. If it is not defined, the only elimination attempted by the compiler is to replace references to the frame pointer with references to the stack pointer. The definition of this macro is a list of structure initializations, each of which specifies an original and replacement register. On some machines, the position of the argument pointer is not known until the compilation is completed. In such a case, a separate hard register must be used for the argument pointer. This register can be eliminated by replacing it with either the frame pointer or the argument pointer, depending on whether or not the frame pointer has been eliminated. In this case, you might specify: #define ELIMINABLE_REGS \\ {{ARG_POINTER_REGNUM, STACK_POINTER_REGNUM}, \\ {ARG_POINTER_REGNUM, FRAME_POINTER_REGNUM}, \\ {FRAME_POINTER_REGNUM, STACK_POINTER_REGNUM}} Note that the elimination of the argument pointer with the stack pointer is specified first since that is the preferred elimination. @{b}CAN_ELIMINATE (@{I}FROM-REG@{ui}, @{I}TO-REG@{ui})@{ub} A C expression that returns non-zero if the compiler is allowed to try to replace register number @{I}FROM-REG@{ui} with register number @{I}TO-REG@{ui}. This macro need only be defined if @{b}ELIMINABLE_REGS@{ub} is defined, and will usually be the constant 1, since most of the cases preventing register elimination are things that the compiler already knows about. @{b}INITIAL_ELIMINATION_OFFSET (@{I}FROM-REG@{ui}, @{I}TO-REG@{ui}, @{I}OFFSET-VAR@{ui})@{ub} This macro is similar to @{b}INITIAL_FRAME_POINTER_OFFSET@{ub}. It specifies the initial difference between the specified pair of registers. This macro must be defined if @{b}ELIMINABLE_REGS@{ub} is defined. @{b}LONGJMP_RESTORE_FROM_STACK@{ub} Define this macro if the @{b}longjmp@{ub} function restores registers from the stack frames, rather than from those saved specifically by @{b}setjmp@{ub}. Certain quantities must not be kept in registers across a call to @{b}setjmp@{ub} on such machines. @EndNode @Node "Stack Arguments" "gcc.guide/Stack Arguments" @Next "Register Arguments" @Prev "Elimination" @Toc "Stack and Calling" Passing Function Arguments on the Stack --------------------------------------- The macros in this section control how arguments are passed on the stack. See the following section for other macros that control passing certain arguments in registers. @{b}PROMOTE_PROTOTYPES@{ub} Define this macro if an argument declared in a prototype as an integral type smaller than @{b}int@{ub} should actually be passed as an @{b}int@{ub}. In addition to avoiding errors in certain cases of mismatch, it also makes for better code on certain machines. @{b}PUSH_ROUNDING (@{I}NPUSHED@{ui})@{ub} A C expression that is the number of bytes actually pushed onto the stack when an instruction attempts to push @{I}NPUSHED@{ui} bytes. If the target machine does not have a push instruction, do not define this macro. That directs GNU CC to use an alternate strategy: to allocate the entire argument block and then store the arguments into it. On some machines, the definition #define PUSH_ROUNDING(BYTES) (BYTES) will suffice. But on other machines, instructions that appear to push one byte actually push two bytes in an attempt to maintain alignment. Then the definition should be #define PUSH_ROUNDING(BYTES) (((BYTES) + 1) & ~1) @{b}ACCUMULATE_OUTGOING_ARGS@{ub} If defined, the maximum amount of space required for outgoing arguments will be computed and placed into the variable @{b}current_function_outgoing_args_size@{ub}. No space will be pushed onto the stack for each call; instead, the function prologue should increase the stack frame size by this amount. Defining both @{b}PUSH_ROUNDING@{ub} and @{b}ACCUMULATE_OUTGOING_ARGS@{ub} is not proper. @{b}REG_PARM_STACK_SPACE (@{I}FNDECL@{ui})@{ub} Define this macro if functions should assume that stack space has been allocated for arguments even when their values are passed in registers. The value of this macro is the size, in bytes, of the area reserved for arguments passed in registers for the function represented by @{I}FNDECL@{ui}. This space can be allocated by the caller, or be a part of the machine-dependent stack frame: @{b}OUTGOING_REG_PARM_STACK_SPACE@{ub} says which. @{b}MAYBE_REG_PARM_STACK_SPACE@{ub} @{b}FINAL_REG_PARM_STACK_SPACE (@{I}CONST_SIZE@{ui}, @{I}VAR_SIZE@{ui})@{ub} Define these macros in addition to the one above if functions might allocate stack space for arguments even when their values are passed in registers. These should be used when the stack space allocated for arguments in registers is not a simple constant independent of the function declaration. The value of the first macro is the size, in bytes, of the area that we should initially assume would be reserved for arguments passed in registers. The value of the second macro is the actual size, in bytes, of the area that will be reserved for arguments passed in registers. This takes two arguments: an integer representing the number of bytes of fixed sized arguments on the stack, and a tree representing the number of bytes of variable sized arguments on the stack. When these macros are defined, @{b}REG_PARM_STACK_SPACE@{ub} will only be called for libcall functions, the current function, or for a function being called when it is known that such stack space must be allocated. In each case this value can be easily computed. When deciding whether a called function needs such stack space, and how much space to reserve, GNU CC uses these two macros instead of @{b}REG_PARM_STACK_SPACE@{ub}. @{b}OUTGOING_REG_PARM_STACK_SPACE@{ub} Define this if it is the responsibility of the caller to allocate the area reserved for arguments passed in registers. If @{b}ACCUMULATE_OUTGOING_ARGS@{ub} is defined, this macro controls whether the space for these arguments counts in the value of @{b}current_function_outgoing_args_size@{ub}. @{b}STACK_PARMS_IN_REG_PARM_AREA@{ub} Define this macro if @{b}REG_PARM_STACK_SPACE@{ub} is defined, but the stack parameters don't skip the area specified by it. Normally, when a parameter is not passed in registers, it is placed on the stack beyond the @{b}REG_PARM_STACK_SPACE@{ub} area. Defining this macro suppresses this behavior and causes the parameter to be passed on the stack in its natural location. @{b}RETURN_POPS_ARGS (@{I}FUNDECL@{ui}, @{I}FUNTYPE@{ui}, @{I}STACK-SIZE@{ui})@{ub} A C expression that should indicate the number of bytes of its own arguments that a function pops on returning, or 0 if the function pops no arguments and the caller must therefore pop them all after the function returns. @{I}FUNDECL@{ui} is a C variable whose value is a tree node that describes the function in question. Normally it is a node of type @{b}FUNCTION_DECL@{ub} that describes the declaration of the function. From this it is possible to obtain the DECL_MACHINE_ATTRIBUTES of the function. @{I}FUNTYPE@{ui} is a C variable whose value is a tree node that describes the function in question. Normally it is a node of type @{b}FUNCTION_TYPE@{ub} that describes the data type of the function. From this it is possible to obtain the data types of the value and arguments (if known). When a call to a library function is being considered, @{I}FUNTYPE@{ui} will contain an identifier node for the library function. Thus, if you need to distinguish among various library functions, you can do so by their names. Note that "library function" in this context means a function used to perform arithmetic, whose name is known specially in the compiler and was not mentioned in the C code being compiled. @{I}STACK-SIZE@{ui} is the number of bytes of arguments passed on the stack. If a variable number of bytes is passed, it is zero, and argument popping will always be the responsibility of the calling function. On the Vax, all functions always pop their arguments, so the definition of this macro is @{I}STACK-SIZE@{ui}. On the 68000, using the standard calling convention, no functions pop their arguments, so the value of the macro is always 0 in this case. But an alternative calling convention is available in which functions that take a fixed number of arguments pop them but other functions (such as @{b}printf@{ub}) pop nothing (the caller pops all). When this convention is in use, @{I}FUNTYPE@{ui} is examined to determine whether a function takes a fixed number of arguments. @EndNode @Node "Register Arguments" "gcc.guide/Register Arguments" @Next "Scalar Return" @Prev "Stack Arguments" @Toc "Stack and Calling" Passing Arguments in Registers ------------------------------ This section describes the macros which let you control how various types of arguments are passed in registers or how they are arranged in the stack. @{b}FUNCTION_ARG (@{I}CUM@{ui}, @{I}MODE@{ui}, @{I}TYPE@{ui}, @{I}NAMED@{ui})@{ub} A C expression that controls whether a function argument is passed in a register, and which register. The arguments are @{I}CUM@{ui}, which summarizes all the previous arguments; @{I}MODE@{ui}, the machine mode of the argument; @{I}TYPE@{ui}, the data type of the argument as a tree node or 0 if that is not known (which happens for C support library functions); and @{I}NAMED@{ui}, which is 1 for an ordinary argument and 0 for nameless arguments that correspond to @{b}...@{ub} in the called function's prototype. The value of the expression should either be a @{b}reg@{ub} RTX for the hard register in which to pass the argument, or zero to pass the argument on the stack. For machines like the Vax and 68000, where normally all arguments are pushed, zero suffices as a definition. The usual way to make the ANSI library @{b}stdarg.h@{ub} work on a machine where some arguments are usually passed in registers, is to cause nameless arguments to be passed on the stack instead. This is done by making @{b}FUNCTION_ARG@{ub} return 0 whenever @{I}NAMED@{ui} is 0. You may use the macro @{b}MUST_PASS_IN_STACK (@{I}MODE@{ui}, @{I}TYPE@{ui})@{ub} in the definition of this macro to determine if this argument is of a type that must be passed in the stack. If @{b}REG_PARM_STACK_SPACE@{ub} is not defined and @{b}FUNCTION_ARG@{ub} returns non-zero for such an argument, the compiler will abort. If @{b}REG_PARM_STACK_SPACE@{ub} is defined, the argument will be computed in the stack and then loaded into a register. @{b}FUNCTION_INCOMING_ARG (@{I}CUM@{ui}, @{I}MODE@{ui}, @{I}TYPE@{ui}, @{I}NAMED@{ui})@{ub} Define this macro if the target machine has "register windows", so that the register in which a function sees an arguments is not necessarily the same as the one in which the caller passed the argument. For such machines, @{b}FUNCTION_ARG@{ub} computes the register in which the caller passes the value, and @{b}FUNCTION_INCOMING_ARG@{ub} should be defined in a similar fashion to tell the function being called where the arguments will arrive. If @{b}FUNCTION_INCOMING_ARG@{ub} is not defined, @{b}FUNCTION_ARG@{ub} serves both purposes. @{b}FUNCTION_ARG_PARTIAL_NREGS (@{I}CUM@{ui}, @{I}MODE@{ui}, @{I}TYPE@{ui}, @{I}NAMED@{ui})@{ub} A C expression for the number of words, at the beginning of an argument, must be put in registers. The value must be zero for arguments that are passed entirely in registers or that are entirely pushed on the stack. On some machines, certain arguments must be passed partially in registers and partially in memory. On these machines, typically the first @{I}N@{ui} words of arguments are passed in registers, and the rest on the stack. If a multi-word argument (a @{b}double@{ub} or a structure) crosses that boundary, its first few words must be passed in registers and the rest must be pushed. This macro tells the compiler when this occurs, and how many of the words should go in registers. @{b}FUNCTION_ARG@{ub} for these arguments should return the first register to be used by the caller for this argument; likewise @{b}FUNCTION_INCOMING_ARG@{ub}, for the called function. @{b}FUNCTION_ARG_PASS_BY_REFERENCE (@{I}CUM@{ui}, @{I}MODE@{ui}, @{I}TYPE@{ui}, @{I}NAMED@{ui})@{ub} A C expression that indicates when an argument must be passed by reference. If nonzero for an argument, a copy of that argument is made in memory and a pointer to the argument is passed instead of the argument itself. The pointer is passed in whatever way is appropriate for passing a pointer to that type. On machines where @{b}REG_PARM_STACK_SPACE@{ub} is not defined, a suitable definition of this macro might be #define FUNCTION_ARG_PASS_BY_REFERENCE\\ (CUM, MODE, TYPE, NAMED) \\ MUST_PASS_IN_STACK (MODE, TYPE) @{b}FUNCTION_ARG_CALLEE_COPIES (@{I}CUM@{ui}, @{I}MODE@{ui}, @{I}TYPE@{ui}, @{I}NAMED@{ui})@{ub} If defined, a C expression that indicates when it is the called function's responsibility to make a copy of arguments passed by invisible reference. Normally, the caller makes a copy and passes the address of the copy to the routine being called. When FUNCTION_ARG_CALLEE_COPIES is defined and is nonzero, the caller does not make a copy. Instead, it passes a pointer to the "live" value. The called function must not modify this value. If it can be determined that the value won't be modified, it need not make a copy; otherwise a copy must be made. @{b}CUMULATIVE_ARGS@{ub} A C type for declaring a variable that is used as the first argument of @{b}FUNCTION_ARG@{ub} and other related values. For some target machines, the type @{b}int@{ub} suffices and can hold the number of bytes of argument so far. There is no need to record in @{b}CUMULATIVE_ARGS@{ub} anything about the arguments that have been passed on the stack. The compiler has other variables to keep track of that. For target machines on which all arguments are passed on the stack, there is no need to store anything in @{b}CUMULATIVE_ARGS@{ub}; however, the data structure must exist and should not be empty, so use @{b}int@{ub}. @{b}INIT_CUMULATIVE_ARGS (@{I}CUM@{ui}, @{I}FNTYPE@{ui}, @{I}LIBNAME@{ui})@{ub} A C statement (sans semicolon) for initializing the variable @{I}CUM@{ui} for the state at the beginning of the argument list. The variable has type @{b}CUMULATIVE_ARGS@{ub}. The value of @{I}FNTYPE@{ui} is the tree node for the data type of the function which will receive the args, or 0 if the args are to a compiler support library function. When processing a call to a compiler support library function, @{I}LIBNAME@{ui} identifies which one. It is a @{b}symbol_ref@{ub} rtx which contains the name of the function, as a string. @{I}LIBNAME@{ui} is 0 when an ordinary C function call is being processed. Thus, each time this macro is called, either @{I}LIBNAME@{ui} or @{I}FNTYPE@{ui} is nonzero, but never both of them at once. @{b}INIT_CUMULATIVE_INCOMING_ARGS (@{I}CUM@{ui}, @{I}FNTYPE@{ui}, @{I}LIBNAME@{ui})@{ub} Like @{b}INIT_CUMULATIVE_ARGS@{ub} but overrides it for the purposes of finding the arguments for the function being compiled. If this macro is undefined, @{b}INIT_CUMULATIVE_ARGS@{ub} is used instead. The value passed for @{I}LIBNAME@{ui} is always 0, since library routines with special calling conventions are never compiled with GNU CC. The argument @{I}LIBNAME@{ui} exists for symmetry with @{b}INIT_CUMULATIVE_ARGS@{ub}. @{b}FUNCTION_ARG_ADVANCE (@{I}CUM@{ui}, @{I}MODE@{ui}, @{I}TYPE@{ui}, @{I}NAMED@{ui})@{ub} A C statement (sans semicolon) to update the summarizer variable @{I}CUM@{ui} to advance past an argument in the argument list. The values @{I}MODE@{ui}, @{I}TYPE@{ui} and @{I}NAMED@{ui} describe that argument. Once this is done, the variable @{I}CUM@{ui} is suitable for analyzing the @{i}following@{ui} argument with @{b}FUNCTION_ARG@{ub}, etc. This macro need not do anything if the argument in question was passed on the stack. The compiler knows how to track the amount of stack space used for arguments without any special help. @{b}FUNCTION_ARG_PADDING (@{I}MODE@{ui}, @{I}TYPE@{ui})@{ub} If defined, a C expression which determines whether, and in which direction, to pad out an argument with extra space. The value should be of type @{b}enum direction@{ub}: either @{b}upward@{ub} to pad above the argument, @{b}downward@{ub} to pad below, or @{b}none@{ub} to inhibit padding. The @{i}amount@{ui} of padding is always just enough to reach the next multiple of @{b}FUNCTION_ARG_BOUNDARY@{ub}; this macro does not control it. This macro has a default definition which is right for most systems. For little-endian machines, the default is to pad upward. For big-endian machines, the default is to pad downward for an argument of constant size shorter than an @{b}int@{ub}, and upward otherwise. @{b}FUNCTION_ARG_BOUNDARY (@{I}MODE@{ui}, @{I}TYPE@{ui})@{ub} If defined, a C expression that gives the alignment boundary, in bits, of an argument with the specified mode and type. If it is not defined, @{b}PARM_BOUNDARY@{ub} is used for all arguments. @{b}FUNCTION_ARG_REGNO_P (@{I}REGNO@{ui})@{ub} A C expression that is nonzero if @{I}REGNO@{ui} is the number of a hard register in which function arguments are sometimes passed. This does @{i}not@{ui} include implicit arguments such as the static chain and the structure-value address. On many machines, no registers can be used for this purpose since all function arguments are pushed on the stack. @EndNode @Node "Scalar Return" "gcc.guide/Scalar Return" @Next "Aggregate Return" @Prev "Register Arguments" @Toc "Stack and Calling" How Scalar Function Values Are Returned --------------------------------------- This section discusses the macros that control returning scalars as values--values that can fit in registers. @{b}TRADITIONAL_RETURN_FLOAT@{ub} Define this macro if @{b}-traditional@{ub} should not cause functions declared to return @{b}float@{ub} to convert the value to @{b}double@{ub}. @{b}FUNCTION_VALUE (@{I}VALTYPE@{ui}, @{I}FUNC@{ui})@{ub} A C expression to create an RTX representing the place where a function returns a value of data type @{I}VALTYPE@{ui}. @{I}VALTYPE@{ui} is a tree node representing a data type. Write @{b}TYPE_MODE (@{I}VALTYPE@{ui})@{ub} to get the machine mode used to represent that type. On many machines, only the mode is relevant. (Actually, on most machines, scalar values are returned in the same place regardless of mode). If @{b}PROMOTE_FUNCTION_RETURN@{ub} is defined, you must apply the same promotion rules specified in @{b}PROMOTE_MODE@{ub} if @{I}VALTYPE@{ui} is a scalar type. If the precise function being called is known, @{I}FUNC@{ui} is a tree node (@{b}FUNCTION_DECL@{ub}) for it; otherwise, @{I}FUNC@{ui} is a null pointer. This makes it possible to use a different value-returning convention for specific functions when all their calls are known. @{b}FUNCTION_VALUE@{ub} is not used for return vales with aggregate data types, because these are returned in another way. See @{b}STRUCT_VALUE_REGNUM@{ub} and related macros, below. @{b}FUNCTION_OUTGOING_VALUE (@{I}VALTYPE@{ui}, @{I}FUNC@{ui})@{ub} Define this macro if the target machine has "register windows" so that the register in which a function returns its value is not the same as the one in which the caller sees the value. For such machines, @{b}FUNCTION_VALUE@{ub} computes the register in which the caller will see the value. @{b}FUNCTION_OUTGOING_VALUE@{ub} should be defined in a similar fashion to tell the function where to put the value. If @{b}FUNCTION_OUTGOING_VALUE@{ub} is not defined, @{b}FUNCTION_VALUE@{ub} serves both purposes. @{b}FUNCTION_OUTGOING_VALUE@{ub} is not used for return vales with aggregate data types, because these are returned in another way. See @{b}STRUCT_VALUE_REGNUM@{ub} and related macros, below. @{b}LIBCALL_VALUE (@{I}MODE@{ui})@{ub} A C expression to create an RTX representing the place where a library function returns a value of mode @{I}MODE@{ui}. If the precise function being called is known, @{I}FUNC@{ui} is a tree node (@{b}FUNCTION_DECL@{ub}) for it; otherwise, @{I}FUNC@{ui} is a null pointer. This makes it possible to use a different value-returning convention for specific functions when all their calls are known. Note that "library function" in this context means a compiler support routine, used to perform arithmetic, whose name is known specially by the compiler and was not mentioned in the C code being compiled. The definition of @{b}LIBRARY_VALUE@{ub} need not be concerned aggregate data types, because none of the library functions returns such types. @{b}FUNCTION_VALUE_REGNO_P (@{I}REGNO@{ui})@{ub} A C expression that is nonzero if @{I}REGNO@{ui} is the number of a hard register in which the values of called function may come back. A register whose use for returning values is limited to serving as the second of a pair (for a value of type @{b}double@{ub}, say) need not be recognized by this macro. So for most machines, this definition suffices: #define FUNCTION_VALUE_REGNO_P(N) ((N) == 0) If the machine has register windows, so that the caller and the called function use different registers for the return value, this macro should recognize only the caller's register numbers. @{b}APPLY_RESULT_SIZE@{ub} Define this macro if @{b}untyped_call@{ub} and @{b}untyped_return@{ub} need more space than is implied by @{b}FUNCTION_VALUE_REGNO_P@{ub} for saving and restoring an arbitrary return value. @EndNode @Node "Aggregate Return" "gcc.guide/Aggregate Return" @Next "Caller Saves" @Prev "Scalar Return" @Toc "Stack and Calling" How Large Values Are Returned ----------------------------- When a function value's mode is @{b}BLKmode@{ub} (and in some other cases), the value is not returned according to @{b}FUNCTION_VALUE@{ub} (see @{"Scalar Return" Link "Scalar Return"}). Instead, the caller passes the address of a block of memory in which the value should be stored. This address is called the @{i}structure value address@{ui}. This section describes how to control returning structure values in memory. @{b}RETURN_IN_MEMORY (@{I}TYPE@{ui})@{ub} A C expression which can inhibit the returning of certain function values in registers, based on the type of value. A nonzero value says to return the function value in memory, just as large structures are always returned. Here @{I}TYPE@{ui} will be a C expression of type @{b}tree@{ub}, representing the data type of the value. Note that values of mode @{b}BLKmode@{ub} must be explicitly handled by this macro. Also, the option @{b}-fpcc-struct-return@{ub} takes effect regardless of this macro. On most systems, it is possible to leave the macro undefined; this causes a default definition to be used, whose value is the constant 1 for @{b}BLKmode@{ub} values, and 0 otherwise. Do not use this macro to indicate that structures and unions should always be returned in memory. You should instead use @{b}DEFAULT_PCC_STRUCT_RETURN@{ub} to indicate this. @{b}DEFAULT_PCC_STRUCT_RETURN@{ub} Define this macro to be 1 if all structure and union return values must be in memory. Since this results in slower code, this should be defined only if needed for compatibility with other compilers or with an ABI. If you define this macro to be 0, then the conventions used for structure and union return values are decided by the @{b}RETURN_IN_MEMORY@{ub} macro. If not defined, this defaults to the value 1. @{b}STRUCT_VALUE_REGNUM@{ub} If the structure value address is passed in a register, then @{b}STRUCT_VALUE_REGNUM@{ub} should be the number of that register. @{b}STRUCT_VALUE@{ub} If the structure value address is not passed in a register, define @{b}STRUCT_VALUE@{ub} as an expression returning an RTX for the place where the address is passed. If it returns 0, the address is passed as an "invisible" first argument. @{b}STRUCT_VALUE_INCOMING_REGNUM@{ub} On some architectures the place where the structure value address is found by the called function is not the same place that the caller put it. This can be due to register windows, or it could be because the function prologue moves it to a different place. If the incoming location of the structure value address is in a register, define this macro as the register number. @{b}STRUCT_VALUE_INCOMING@{ub} If the incoming location is not a register, then you should define @{b}STRUCT_VALUE_INCOMING@{ub} as an expression for an RTX for where the called function should find the value. If it should find the value on the stack, define this to create a @{b}mem@{ub} which refers to the frame pointer. A definition of 0 means that the address is passed as an "invisible" first argument. @{b}PCC_STATIC_STRUCT_RETURN@{ub} Define this macro if the usual system convention on the target machine for returning structures and unions is for the called function to return the address of a static variable containing the value. Do not define this if the usual system convention is for the caller to pass an address to the subroutine. This macro has effect in @{b}-fpcc-struct-return@{ub} mode, but it does nothing when you use @{b}-freg-struct-return@{ub} mode. @EndNode @Node "Caller Saves" "gcc.guide/Caller Saves" @Next "Function Entry" @Prev "Aggregate Return" @Toc "Stack and Calling" Caller-Saves Register Allocation -------------------------------- If you enable it, GNU CC can save registers around function calls. This makes it possible to use call-clobbered registers to hold variables that must live across calls. @{b}DEFAULT_CALLER_SAVES@{ub} Define this macro if function calls on the target machine do not preserve any registers; in other words, if @{b}CALL_USED_REGISTERS@{ub} has 1 for all registers. This macro enables @{b}-fcaller-saves@{ub} by default. Eventually that option will be enabled by default on all machines and both the option and this macro will be eliminated. @{b}CALLER_SAVE_PROFITABLE (@{I}REFS@{ui}, @{I}CALLS@{ui})@{ub} A C expression to determine whether it is worthwhile to consider placing a pseudo-register in a call-clobbered hard register and saving and restoring it around each function call. The expression should be 1 when this is worth doing, and 0 otherwise. If you don't define this macro, a default is used which is good on most machines: @{b}4 * @{I}CALLS@{ui} < @{I}REFS@{ui}@{ub}. @EndNode @Node "Function Entry" "gcc.guide/Function Entry" @Next "Profiling" @Prev "Caller Saves" @Toc "Stack and Calling" Function Entry and Exit ----------------------- This section describes the macros that output function entry (@{i}prologue@{ui}) and exit (@{i}epilogue@{ui}) code. @{b}FUNCTION_PROLOGUE (@{I}FILE@{ui}, @{I}SIZE@{ui})@{ub} A C compound statement that outputs the assembler code for entry to a function. The prologue is responsible for setting up the stack frame, initializing the frame pointer register, saving registers that must be saved, and allocating @{I}SIZE@{ui} additional bytes of storage for the local variables. @{I}SIZE@{ui} is an integer. @{I}FILE@{ui} is a stdio stream to which the assembler code should be output. The label for the beginning of the function need not be output by this macro. That has already been done when the macro is run. To determine which registers to save, the macro can refer to the array @{b}regs_ever_live@{ub}: element @{I}R@{ui} is nonzero if hard register @{I}R@{ui} is used anywhere within the function. This implies the function prologue should save register @{I}R@{ui}, provided it is not one of the call-used registers. (@{b}FUNCTION_EPILOGUE@{ub} must likewise use @{b}regs_ever_live@{ub}.) On machines that have "register windows", the function entry code does not save on the stack the registers that are in the windows, even if they are supposed to be preserved by function calls; instead it takes appropriate steps to "push" the register stack, if any non-call-used registers are used in the function. On machines where functions may or may not have frame-pointers, the function entry code must vary accordingly; it must set up the frame pointer if one is wanted, and not otherwise. To determine whether a frame pointer is in wanted, the macro can refer to the variable @{b}frame_pointer_needed@{ub}. The variable's value will be 1 at run time in a function that needs a frame pointer. See @{"Elimination" Link "Elimination"}. The function entry code is responsible for allocating any stack space required for the function. This stack space consists of the regions listed below. In most cases, these regions are allocated in the order listed, with the last listed region closest to the top of the stack (the lowest address if @{b}STACK_GROWS_DOWNWARD@{ub} is defined, and the highest address if it is not defined). You can use a different order for a machine if doing so is more convenient or required for compatibility reasons. Except in cases where required by standard or by a debugger, there is no reason why the stack layout used by GCC need agree with that used by other compilers for a machine. @{b}*@{ub} A region of @{b}current_function_pretend_args_size@{ub} bytes of uninitialized space just underneath the first argument arriving on the stack. (This may not be at the very start of the allocated stack region if the calling sequence has pushed anything else since pushing the stack arguments. But usually, on such machines, nothing else has been pushed yet, because the function prologue itself does all the pushing.) This region is used on machines where an argument may be passed partly in registers and partly in memory, and, in some cases to support the features in @{b}varargs.h@{ub} and @{b}stdargs.h@{ub}. @{b}*@{ub} An area of memory used to save certain registers used by the function. The size of this area, which may also include space for such things as the return address and pointers to previous stack frames, is machine-specific and usually depends on which registers have been used in the function. Machines with register windows often do not require a save area. @{b}*@{ub} A region of at least @{I}SIZE@{ui} bytes, possibly rounded up to an allocation boundary, to contain the local variables of the function. On some machines, this region and the save area may occur in the opposite order, with the save area closer to the top of the stack. @{b}*@{ub} Optionally, when @{b}ACCUMULATE_OUTGOING_ARGS@{ub} is defined, a region of @{b}current_function_outgoing_args_size@{ub} bytes to be used for outgoing argument lists of the function. See @{"Stack Arguments" Link "Stack Arguments"}. Normally, it is necessary for the macros @{b}FUNCTION_PROLOGUE@{ub} and @{b}FUNCTION_EPILOGUE@{ub} to treat leaf functions specially. The C variable @{b}leaf_function@{ub} is nonzero for such a function. @{b}EXIT_IGNORE_STACK@{ub} Define this macro as a C expression that is nonzero if the return instruction or the function epilogue ignores the value of the stack pointer; in other words, if it is safe to delete an instruction to adjust the stack pointer before a return from the function. Note that this macro's value is relevant only for functions for which frame pointers are maintained. It is never safe to delete a final stack adjustment in a function that has no frame pointer, and the compiler knows this regardless of @{b}EXIT_IGNORE_STACK@{ub}. @{b}FUNCTION_EPILOGUE (@{I}FILE@{ui}, @{I}SIZE@{ui})@{ub} A C compound statement that outputs the assembler code for exit from a function. The epilogue is responsible for restoring the saved registers and stack pointer to their values when the function was called, and returning control to the caller. This macro takes the same arguments as the macro @{b}FUNCTION_PROLOGUE@{ub}, and the registers to restore are determined from @{b}regs_ever_live@{ub} and @{b}CALL_USED_REGISTERS@{ub} in the same way. On some machines, there is a single instruction that does all the work of returning from the function. On these machines, give that instruction the name @{b}return@{ub} and do not define the macro @{b}FUNCTION_EPILOGUE@{ub} at all. Do not define a pattern named @{b}return@{ub} if you want the @{b}FUNCTION_EPILOGUE@{ub} to be used. If you want the target switches to control whether return instructions or epilogues are used, define a @{b}return@{ub} pattern with a validity condition that tests the target switches appropriately. If the @{b}return@{ub} pattern's validity condition is false, epilogues will be used. On machines where functions may or may not have frame-pointers, the function exit code must vary accordingly. Sometimes the code for these two cases is completely different. To determine whether a frame pointer is wanted, the macro can refer to the variable @{b}frame_pointer_needed@{ub}. The variable's value will be 1 when compiling a function that needs a frame pointer. Normally, @{b}FUNCTION_PROLOGUE@{ub} and @{b}FUNCTION_EPILOGUE@{ub} must treat leaf functions specially. The C variable @{b}leaf_function@{ub} is nonzero for such a function. See @{"Leaf Functions" Link "Leaf Functions"}. On some machines, some functions pop their arguments on exit while others leave that for the caller to do. For example, the 68020 when given @{b}-mrtd@{ub} pops arguments in functions that take a fixed number of arguments. Your definition of the macro @{b}RETURN_POPS_ARGS@{ub} decides which functions pop their own arguments. @{b}FUNCTION_EPILOGUE@{ub} needs to know what was decided. The variable that is called @{b}current_function_pops_args@{ub} is the number of bytes of its arguments that a function should pop. See @{"Scalar Return" Link "Scalar Return"}. @{b}DELAY_SLOTS_FOR_EPILOGUE@{ub} Define this macro if the function epilogue contains delay slots to which instructions from the rest of the function can be "moved". The definition should be a C expression whose value is an integer representing the number of delay slots there. @{b}ELIGIBLE_FOR_EPILOGUE_DELAY (@{I}INSN@{ui}, @{I}N@{ui})@{ub} A C expression that returns 1 if @{I}INSN@{ui} can be placed in delay slot number @{I}N@{ui} of the epilogue. The argument @{I}N@{ui} is an integer which identifies the delay slot now being considered (since different slots may have different rules of eligibility). It is never negative and is always less than the number of epilogue delay slots (what @{b}DELAY_SLOTS_FOR_EPILOGUE@{ub} returns). If you reject a particular insn for a given delay slot, in principle, it may be reconsidered for a subsequent delay slot. Also, other insns may (at least in principle) be considered for the so far unfilled delay slot. The insns accepted to fill the epilogue delay slots are put in an RTL list made with @{b}insn_list@{ub} objects, stored in the variable @{b}current_function_epilogue_delay_list@{ub}. The insn for the first delay slot comes first in the list. Your definition of the macro @{b}FUNCTION_EPILOGUE@{ub} should fill the delay slots by outputting the insns in this list, usually by calling @{b}final_scan_insn@{ub}. You need not define this macro if you did not define @{b}DELAY_SLOTS_FOR_EPILOGUE@{ub}. @EndNode @Node "Profiling" "gcc.guide/Profiling" @Prev "Function Entry" @Toc "Stack and Calling" Generating Code for Profiling ----------------------------- These macros will help you generate code for profiling. @{b}FUNCTION_PROFILER (@{I}FILE@{ui}, @{I}LABELNO@{ui})@{ub} A C statement or compound statement to output to @{I}FILE@{ui} some assembler code to call the profiling subroutine @{b}mcount@{ub}. Before calling, the assembler code must load the address of a counter variable into a register where @{b}mcount@{ub} expects to find the address. The name of this variable is @{b}LP@{ub} followed by the number @{I}LABELNO@{ui}, so you would generate the name using @{b}LP%d@{ub} in a @{b}fprintf@{ub}. The details of how the address should be passed to @{b}mcount@{ub} are determined by your operating system environment, not by GNU CC. To figure them out, compile a small program for profiling using the system's installed C compiler and look at the assembler code that results. @{b}PROFILE_BEFORE_PROLOGUE@{ub} Define this macro if the code for function profiling should come before the function prologue. Normally, the profiling code comes after. @{b}FUNCTION_BLOCK_PROFILER (@{I}FILE@{ui}, @{I}LABELNO@{ui})@{ub} A C statement or compound statement to output to @{I}FILE@{ui} some assembler code to initialize basic-block profiling for the current object module. This code should call the subroutine @{b}__bb_init_func@{ub} once per object module, passing it as its sole argument the address of a block allocated in the object module. The name of the block is a local symbol made with this statement: ASM_GENERATE_INTERNAL_LABEL (@{I}BUFFER@{ui}, "LPBX", 0); Of course, since you are writing the definition of @{b}ASM_GENERATE_INTERNAL_LABEL@{ub} as well as that of this macro, you can take a short cut in the definition of this macro and use the name that you know will result. The first word of this block is a flag which will be nonzero if the object module has already been initialized. So test this word first, and do not call @{b}__bb_init_func@{ub} if the flag is nonzero. @{b}BLOCK_PROFILER (@{I}FILE@{ui}, @{I}BLOCKNO@{ui})@{ub} A C statement or compound statement to increment the count associated with the basic block number @{I}BLOCKNO@{ui}. Basic blocks are numbered separately from zero within each compilation. The count associated with block number @{I}BLOCKNO@{ui} is at index @{I}BLOCKNO@{ui} in a vector of words; the name of this array is a local symbol made with this statement: ASM_GENERATE_INTERNAL_LABEL (@{I}BUFFER@{ui}, "LPBX", 2); Of course, since you are writing the definition of @{b}ASM_GENERATE_INTERNAL_LABEL@{ub} as well as that of this macro, you can take a short cut in the definition of this macro and use the name that you know will result. @{b}BLOCK_PROFILER_CODE@{ub} A C function or functions which are needed in the library to support block profiling. @EndNode @Node "Varargs" "gcc.guide/Varargs" @Next "Trampolines" @Prev "Stack and Calling" @Toc "Target Macros" Implementing the Varargs Macros =============================== GNU CC comes with an implementation of @{b}varargs.h@{ub} and @{b}stdarg.h@{ub} that work without change on machines that pass arguments on the stack. Other machines require their own implementations of varargs, and the two machine independent header files must have conditionals to include it. ANSI @{b}stdarg.h@{ub} differs from traditional @{b}varargs.h@{ub} mainly in the calling convention for @{b}va_start@{ub}. The traditional implementation takes just one argument, which is the variable in which to store the argument pointer. The ANSI implementation of @{b}va_start@{ub} takes an additional second argument. The user is supposed to write the last named argument of the function here. However, @{b}va_start@{ub} should not use this argument. The way to find the end of the named arguments is with the built-in functions described below. @{b}__builtin_saveregs ()@{ub} Use this built-in function to save the argument registers in memory so that the varargs mechanism can access them. Both ANSI and traditional versions of @{b}va_start@{ub} must use @{b}__builtin_saveregs@{ub}, unless you use @{b}SETUP_INCOMING_VARARGS@{ub} (see below) instead. On some machines, @{b}__builtin_saveregs@{ub} is open-coded under the control of the macro @{b}EXPAND_BUILTIN_SAVEREGS@{ub}. On other machines, it calls a routine written in assembler language, found in @{b}libgcc2.c@{ub}. Code generated for the call to @{b}__builtin_saveregs@{ub} appears at the beginning of the function, as opposed to where the call to @{b}__builtin_saveregs@{ub} is written, regardless of what the code is. This is because the registers must be saved before the function starts to use them for its own purposes. @{b}__builtin_args_info (@{I}CATEGORY@{ui})@{ub} Use this built-in function to find the first anonymous arguments in registers. In general, a machine may have several categories of registers used for arguments, each for a particular category of data types. (For example, on some machines, floating-point registers are used for floating-point arguments while other arguments are passed in the general registers.) To make non-varargs functions use the proper calling convention, you have defined the @{b}CUMULATIVE_ARGS@{ub} data type to record how many registers in each category have been used so far @{b}__builtin_args_info@{ub} accesses the same data structure of type @{b}CUMULATIVE_ARGS@{ub} after the ordinary argument layout is finished with it, with @{I}CATEGORY@{ui} specifying which word to access. Thus, the value indicates the first unused register in a given category. Normally, you would use @{b}__builtin_args_info@{ub} in the implementation of @{b}va_start@{ub}, accessing each category just once and storing the value in the @{b}va_list@{ub} object. This is because @{b}va_list@{ub} will have to update the values, and there is no way to alter the values accessed by @{b}__builtin_args_info@{ub}. @{b}__builtin_next_arg (@{I}LASTARG@{ui})@{ub} This is the equivalent of @{b}__builtin_args_info@{ub}, for stack arguments. It returns the address of the first anonymous stack argument, as type @{b}void *@{ub}. If @{b}ARGS_GROW_DOWNWARD@{ub}, it returns the address of the location above the first anonymous stack argument. Use it in @{b}va_start@{ub} to initialize the pointer for fetching arguments from the stack. Also use it in @{b}va_start@{ub} to verify that the second parameter @{I}LASTARG@{ui} is the last named argument of the current function. @{b}__builtin_classify_type (@{I}OBJECT@{ui})@{ub} Since each machine has its own conventions for which data types are passed in which kind of register, your implementation of @{b}va_arg@{ub} has to embody these conventions. The easiest way to categorize the specified data type is to use @{b}__builtin_classify_type@{ub} together with @{b}sizeof@{ub} and @{b}__alignof__@{ub}. @{b}__builtin_classify_type@{ub} ignores the value of @{I}OBJECT@{ui}, considering only its data type. It returns an integer describing what kind of type that is--integer, floating, pointer, structure, and so on. The file @{b}typeclass.h@{ub} defines an enumeration that you can use to interpret the values of @{b}__builtin_classify_type@{ub}. These machine description macros help implement varargs: @{b}EXPAND_BUILTIN_SAVEREGS (@{I}ARGS@{ui})@{ub} If defined, is a C expression that produces the machine-specific code for a call to @{b}__builtin_saveregs@{ub}. This code will be moved to the very beginning of the function, before any parameter access are made. The return value of this function should be an RTX that contains the value to use as the return of @{b}__builtin_saveregs@{ub}. The argument @{I}ARGS@{ui} is a @{b}tree_list@{ub} containing the arguments that were passed to @{b}__builtin_saveregs@{ub}. If this macro is not defined, the compiler will output an ordinary call to the library function @{b}__builtin_saveregs@{ub}. @{b}SETUP_INCOMING_VARARGS (@{I}ARGS_SO_FAR@{ui}, @{I}MODE@{ui}, @{I}TYPE@{ui},@{ub} @{I}PRETEND_ARGS_SIZE@{ui}, @{I}SECOND_TIME@{ui}) This macro offers an alternative to using @{b}__builtin_saveregs@{ub} and defining the macro @{b}EXPAND_BUILTIN_SAVEREGS@{ub}. Use it to store the anonymous register arguments into the stack so that all the arguments appear to have been passed consecutively on the stack. Once this is done, you can use the standard implementation of varargs that works for machines that pass all their arguments on the stack. The argument @{I}ARGS_SO_FAR@{ui} is the @{b}CUMULATIVE_ARGS@{ub} data structure, containing the values that obtain after processing of the named arguments. The arguments @{I}MODE@{ui} and @{I}TYPE@{ui} describe the last named argument--its machine mode and its data type as a tree node. The macro implementation should do two things: first, push onto the stack all the argument registers @{i}not@{ui} used for the named arguments, and second, store the size of the data thus pushed into the @{b}int@{ub}-valued variable whose name is supplied as the argument @{I}PRETEND_ARGS_SIZE@{ui}. The value that you store here will serve as additional offset for setting up the stack frame. Because you must generate code to push the anonymous arguments at compile time without knowing their data types, @{b}SETUP_INCOMING_VARARGS@{ub} is only useful on machines that have just a single category of argument register and use it uniformly for all data types. If the argument @{I}SECOND_TIME@{ui} is nonzero, it means that the arguments of the function are being analyzed for the second time. This happens for an inline function, which is not actually compiled until the end of the source file. The macro @{b}SETUP_INCOMING_VARARGS@{ub} should not generate any instructions in this case. @{b}STRICT_ARGUMENT_NAMING@{ub} Define this macro if the location where a function argument is passed depends on whether or not it is a named argument. This macro controls how the @{I}NAMED@{ui} argument to @{b}FUNCTION_ARG@{ub} is set for varargs and stdarg functions. With this macro defined, the @{I}NAMED@{ui} argument is always true for named arguments, and false for unnamed arguments. If this is not defined, but @{b}SETUP_INCOMING_VARARGS@{ub} is defined, then all arguments are treated as named. Otherwise, all named arguments except the last are treated as named. @EndNode @Node "Trampolines" "gcc.guide/Trampolines" @Next "Library Calls" @Prev "Varargs" @Toc "Target Macros" Trampolines for Nested Functions ================================ A @{i}trampoline@{ui} is a small piece of code that is created at run time when the address of a nested function is taken. It normally resides on the stack, in the stack frame of the containing function. These macros tell GNU CC how to generate code to allocate and initialize a trampoline. The instructions in the trampoline must do two things: load a constant address into the static chain register, and jump to the real address of the nested function. On CISC machines such as the m68k, this requires two instructions, a move immediate and a jump. Then the two addresses exist in the trampoline as word-long immediate operands. On RISC machines, it is often necessary to load each address into a register in two parts. Then pieces of each address form separate immediate operands. The code generated to initialize the trampoline must store the variable parts--the static chain value and the function address--into the immediate operands of the instructions. On a CISC machine, this is simply a matter of copying each address to a memory reference at the proper offset from the start of the trampoline. On a RISC machine, it may be necessary to take out pieces of the address and store them separately. @{b}TRAMPOLINE_TEMPLATE (@{I}FILE@{ui})@{ub} A C statement to output, on the stream @{I}FILE@{ui}, assembler code for a block of data that contains the constant parts of a trampoline. This code should not include a label--the label is taken care of automatically. @{b}TRAMPOLINE_SECTION@{ub} The name of a subroutine to switch to the section in which the trampoline template is to be placed (see @{"Sections" Link "Sections"}). The default is a value of @{b}readonly_data_section@{ub}, which places the trampoline in the section containing read-only data. @{b}TRAMPOLINE_SIZE@{ub} A C expression for the size in bytes of the trampoline, as an integer. @{b}TRAMPOLINE_ALIGNMENT@{ub} Alignment required for trampolines, in bits. If you don't define this macro, the value of @{b}BIGGEST_ALIGNMENT@{ub} is used for aligning trampolines. @{b}INITIALIZE_TRAMPOLINE (@{I}ADDR@{ui}, @{I}FNADDR@{ui}, @{I}STATIC_CHAIN@{ui})@{ub} A C statement to initialize the variable parts of a trampoline. @{I}ADDR@{ui} is an RTX for the address of the trampoline; @{I}FNADDR@{ui} is an RTX for the address of the nested function; @{I}STATIC_CHAIN@{ui} is an RTX for the static chain value that should be passed to the function when it is called. @{b}ALLOCATE_TRAMPOLINE (@{I}FP@{ui})@{ub} A C expression to allocate run-time space for a trampoline. The expression value should be an RTX representing a memory reference to the space for the trampoline. If this macro is not defined, by default the trampoline is allocated as a stack slot. This default is right for most machines. The exceptions are machines where it is impossible to execute instructions in the stack area. On such machines, you may have to implement a separate stack, using this macro in conjunction with @{b}FUNCTION_PROLOGUE@{ub} and @{b}FUNCTION_EPILOGUE@{ub}. @{I}FP@{ui} points to a data structure, a @{b}struct function@{ub}, which describes the compilation status of the immediate containing function of the function which the trampoline is for. Normally (when @{b}ALLOCATE_TRAMPOLINE@{ub} is not defined), the stack slot for the trampoline is in the stack frame of this containing function. Other allocation strategies probably must do something analogous with this information. Implementing trampolines is difficult on many machines because they have separate instruction and data caches. Writing into a stack location fails to clear the memory in the instruction cache, so when the program jumps to that location, it executes the old contents. Here are two possible solutions. One is to clear the relevant parts of the instruction cache whenever a trampoline is set up. The other is to make all trampolines identical, by having them jump to a standard subroutine. The former technique makes trampoline execution faster; the latter makes initialization faster. To clear the instruction cache when a trampoline is initialized, define the following macros which describe the shape of the cache. @{b}INSN_CACHE_SIZE@{ub} The total size in bytes of the cache. @{b}INSN_CACHE_LINE_WIDTH@{ub} The length in bytes of each cache line. The cache is divided into cache lines which are disjoint slots, each holding a contiguous chunk of data fetched from memory. Each time data is brought into the cache, an entire line is read at once. The data loaded into a cache line is always aligned on a boundary equal to the line size. @{b}INSN_CACHE_DEPTH@{ub} The number of alternative cache lines that can hold any particular memory location. Alternatively, if the machine has system calls or instructions to clear the instruction cache directly, you can define the following macro. @{b}CLEAR_INSN_CACHE (@{I}BEG@{ui}, @{I}END@{ui})@{ub} If defined, expands to a C expression clearing the @{i}instruction cache@{ui} in the specified interval. If it is not defined, and the macro INSN_CACHE_SIZE is defined, some generic code is generated to clear the cache. The definition of this macro would typically be a series of @{b}asm@{ub} statements. Both @{I}BEG@{ui} and @{I}END@{ui} are both pointer expressions. To use a standard subroutine, define the following macro. In addition, you must make sure that the instructions in a trampoline fill an entire cache line with identical instructions, or else ensure that the beginning of the trampoline code is always aligned at the same point in its cache line. Look in @{b}m68k.h@{ub} as a guide. @{b}TRANSFER_FROM_TRAMPOLINE@{ub} Define this macro if trampolines need a special subroutine to do their work. The macro should expand to a series of @{b}asm@{ub} statements which will be compiled with GNU CC. They go in a library function named @{b}__transfer_from_trampoline@{ub}. If you need to avoid executing the ordinary prologue code of a compiled C function when you jump to the subroutine, you can do so by placing a special label of your own in the assembler code. Use one @{b}asm@{ub} statement to generate an assembler label, and another to make the label global. Then trampolines can use that label to jump directly to your special assembler code. @EndNode @Node "Library Calls" "gcc.guide/Library Calls" @Next "Addressing Modes" @Prev "Trampolines" @Toc "Target Macros" Implicit Calls to Library Routines ================================== Here is an explanation of implicit calls to library routines. @{b}MULSI3_LIBCALL@{ub} A C string constant giving the name of the function to call for multiplication of one signed full-word by another. If you do not define this macro, the default name is used, which is @{b}__mulsi3@{ub}, a function defined in @{b}libgcc.a@{ub}. @{b}DIVSI3_LIBCALL@{ub} A C string constant giving the name of the function to call for division of one signed full-word by another. If you do not define this macro, the default name is used, which is @{b}__divsi3@{ub}, a function defined in @{b}libgcc.a@{ub}. @{b}UDIVSI3_LIBCALL@{ub} A C string constant giving the name of the function to call for division of one unsigned full-word by another. If you do not define this macro, the default name is used, which is @{b}__udivsi3@{ub}, a function defined in @{b}libgcc.a@{ub}. @{b}MODSI3_LIBCALL@{ub} A C string constant giving the name of the function to call for the remainder in division of one signed full-word by another. If you do not define this macro, the default name is used, which is @{b}__modsi3@{ub}, a function defined in @{b}libgcc.a@{ub}. @{b}UMODSI3_LIBCALL@{ub} A C string constant giving the name of the function to call for the remainder in division of one unsigned full-word by another. If you do not define this macro, the default name is used, which is @{b}__umodsi3@{ub}, a function defined in @{b}libgcc.a@{ub}. @{b}MULDI3_LIBCALL@{ub} A C string constant giving the name of the function to call for multiplication of one signed double-word by another. If you do not define this macro, the default name is used, which is @{b}__muldi3@{ub}, a function defined in @{b}libgcc.a@{ub}. @{b}DIVDI3_LIBCALL@{ub} A C string constant giving the name of the function to call for division of one signed double-word by another. If you do not define this macro, the default name is used, which is @{b}__divdi3@{ub}, a function defined in @{b}libgcc.a@{ub}. @{b}UDIVDI3_LIBCALL@{ub} A C string constant giving the name of the function to call for division of one unsigned full-word by another. If you do not define this macro, the default name is used, which is @{b}__udivdi3@{ub}, a function defined in @{b}libgcc.a@{ub}. @{b}MODDI3_LIBCALL@{ub} A C string constant giving the name of the function to call for the remainder in division of one signed double-word by another. If you do not define this macro, the default name is used, which is @{b}__moddi3@{ub}, a function defined in @{b}libgcc.a@{ub}. @{b}UMODDI3_LIBCALL@{ub} A C string constant giving the name of the function to call for the remainder in division of one unsigned full-word by another. If you do not define this macro, the default name is used, which is @{b}__umoddi3@{ub}, a function defined in @{b}libgcc.a@{ub}. @{b}INIT_TARGET_OPTABS@{ub} Define this macro as a C statement that declares additional library routines renames existing ones. @{b}init_optabs@{ub} calls this macro after initializing all the normal library routines. @{b}TARGET_EDOM@{ub} The value of @{b}EDOM@{ub} on the target machine, as a C integer constant expression. If you don't define this macro, GNU CC does not attempt to deposit the value of @{b}EDOM@{ub} into @{b}errno@{ub} directly. Look in @{b}/usr/include/errno.h@{ub} to find the value of @{b}EDOM@{ub} on your system. If you do not define @{b}TARGET_EDOM@{ub}, then compiled code reports domain errors by calling the library function and letting it report the error. If mathematical functions on your system use @{b}matherr@{ub} when there is an error, then you should leave @{b}TARGET_EDOM@{ub} undefined so that @{b}matherr@{ub} is used normally. @{b}GEN_ERRNO_RTX@{ub} Define this macro as a C expression to create an rtl expression that refers to the global "variable" @{b}errno@{ub}. (On certain systems, @{b}errno@{ub} may not actually be a variable.) If you don't define this macro, a reasonable default is used. @{b}TARGET_MEM_FUNCTIONS@{ub} Define this macro if GNU CC should generate calls to the System V (and ANSI C) library functions @{b}memcpy@{ub} and @{b}memset@{ub} rather than the BSD functions @{b}bcopy@{ub} and @{b}bzero@{ub}. @{b}LIBGCC_NEEDS_DOUBLE@{ub} Define this macro if only @{b}float@{ub} arguments cannot be passed to library routines (so they must be converted to @{b}double@{ub}). This macro affects both how library calls are generated and how the library routines in @{b}libgcc1.c@{ub} accept their arguments. It is useful on machines where floating and fixed point arguments are passed differently, such as the i860. @{b}FLOAT_ARG_TYPE@{ub} Define this macro to override the type used by the library routines to pick up arguments of type @{b}float@{ub}. (By default, they use a union of @{b}float@{ub} and @{b}int@{ub}.) The obvious choice would be @{b}float@{ub}--but that won't work with traditional C compilers that expect all arguments declared as @{b}float@{ub} to arrive as @{b}double@{ub}. To avoid this conversion, the library routines ask for the value as some other type and then treat it as a @{b}float@{ub}. On some systems, no other type will work for this. For these systems, you must use @{b}LIBGCC_NEEDS_DOUBLE@{ub} instead, to force conversion of the values @{b}double@{ub} before they are passed. @{b}FLOATIFY (@{I}PASSED-VALUE@{ui})@{ub} Define this macro to override the way library routines redesignate a @{b}float@{ub} argument as a @{b}float@{ub} instead of the type it was passed as. The default is an expression which takes the @{b}float@{ub} field of the union. @{b}FLOAT_VALUE_TYPE@{ub} Define this macro to override the type used by the library routines to return values that ought to have type @{b}float@{ub}. (By default, they use @{b}int@{ub}.) The obvious choice would be @{b}float@{ub}--but that won't work with traditional C compilers gratuitously convert values declared as @{b}float@{ub} into @{b}double@{ub}. @{b}INTIFY (@{I}FLOAT-VALUE@{ui})@{ub} Define this macro to override the way the value of a @{b}float@{ub}-returning library routine should be packaged in order to return it. These functions are actually declared to return type @{b}FLOAT_VALUE_TYPE@{ub} (normally @{b}int@{ub}). These values can't be returned as type @{b}float@{ub} because traditional C compilers would gratuitously convert the value to a @{b}double@{ub}. A local variable named @{b}intify@{ub} is always available when the macro @{b}INTIFY@{ub} is used. It is a union of a @{b}float@{ub} field named @{b}f@{ub} and a field named @{b}i@{ub} whose type is @{b}FLOAT_VALUE_TYPE@{ub} or @{b}int@{ub}. If you don't define this macro, the default definition works by copying the value through that union. @{b}nongcc_SI_type@{ub} Define this macro as the name of the data type corresponding to @{b}SImode@{ub} in the system's own C compiler. You need not define this macro if that type is @{b}long int@{ub}, as it usually is. @{b}nongcc_word_type@{ub} Define this macro as the name of the data type corresponding to the word_mode in the system's own C compiler. You need not define this macro if that type is @{b}long int@{ub}, as it usually is. @{b}perform_...@{ub} Define these macros to supply explicit C statements to carry out various arithmetic operations on types @{b}float@{ub} and @{b}double@{ub} in the library routines in @{b}libgcc1.c@{ub}. See that file for a full list of these macros and their arguments. On most machines, you don't need to define any of these macros, because the C compiler that comes with the system takes care of doing them. @{b}NEXT_OBJC_RUNTIME@{ub} Define this macro to generate code for Objective C message sending using the calling convention of the NeXT system. This calling convention involves passing the object, the selector and the method arguments all at once to the method-lookup library function. The default calling convention passes just the object and the selector to the lookup function, which returns a pointer to the method. @EndNode @Node "Addressing Modes" "gcc.guide/Addressing Modes" @Next "Condition Code" @Prev "Library Calls" @Toc "Target Macros" Addressing Modes ================ This is about addressing modes. @{b}HAVE_POST_INCREMENT@{ub} Define this macro if the machine supports post-increment addressing. @{b}HAVE_PRE_INCREMENT@{ub} @{b}HAVE_POST_DECREMENT@{ub} @{b}HAVE_PRE_DECREMENT@{ub} Similar for other kinds of addressing. @{b}CONSTANT_ADDRESS_P (@{I}X@{ui})@{ub} A C expression that is 1 if the RTX @{I}X@{ui} is a constant which is a valid address. On most machines, this can be defined as @{b}CONSTANT_P (@{I}X@{ui})@{ub}, but a few machines are more restrictive in which constant addresses are supported. @{b}CONSTANT_P@{ub} accepts integer-values expressions whose values are not explicitly known, such as @{b}symbol_ref@{ub}, @{b}label_ref@{ub}, and @{b}high@{ub} expressions and @{b}const@{ub} arithmetic expressions, in addition to @{b}const_int@{ub} and @{b}const_double@{ub} expressions. @{b}MAX_REGS_PER_ADDRESS@{ub} A number, the maximum number of registers that can appear in a valid memory address. Note that it is up to you to specify a value equal to the maximum number that @{b}GO_IF_LEGITIMATE_ADDRESS@{ub} would ever accept. @{b}GO_IF_LEGITIMATE_ADDRESS (@{I}MODE@{ui}, @{I}X@{ui}, @{I}LABEL@{ui})@{ub} A C compound statement with a conditional @{b}goto @{I}LABEL@{ui};@{ub} executed if @{I}X@{ui} (an RTX) is a legitimate memory address on the target machine for a memory operand of mode @{I}MODE@{ui}. It usually pays to define several simpler macros to serve as subroutines for this one. Otherwise it may be too complicated to understand. This macro must exist in two variants: a strict variant and a non-strict one. The strict variant is used in the reload pass. It must be defined so that any pseudo-register that has not been allocated a hard register is considered a memory reference. In contexts where some kind of register is required, a pseudo-register with no hard register must be rejected. The non-strict variant is used in other passes. It must be defined to accept all pseudo-registers in every context where some kind of register is required. Compiler source files that want to use the strict variant of this macro define the macro @{b}REG_OK_STRICT@{ub}. You should use an @{b}#ifdef REG_OK_STRICT@{ub} conditional to define the strict variant in that case and the non-strict variant otherwise. Subroutines to check for acceptable registers for various purposes (one for base registers, one for index registers, and so on) are typically among the subroutines used to define @{b}GO_IF_LEGITIMATE_ADDRESS@{ub}. Then only these subroutine macros need have two variants; the higher levels of macros may be the same whether strict or not. Normally, constant addresses which are the sum of a @{b}symbol_ref@{ub} and an integer are stored inside a @{b}const@{ub} RTX to mark them as constant. Therefore, there is no need to recognize such sums specifically as legitimate addresses. Normally you would simply recognize any @{b}const@{ub} as legitimate. Usually @{b}PRINT_OPERAND_ADDRESS@{ub} is not prepared to handle constant sums that are not marked with @{b}const@{ub}. It assumes that a naked @{b}plus@{ub} indicates indexing. If so, then you @{i}must@{ui} reject such naked constant sums as illegitimate addresses, so that none of them will be given to @{b}PRINT_OPERAND_ADDRESS@{ub}. On some machines, whether a symbolic address is legitimate depends on the section that the address refers to. On these machines, define the macro @{b}ENCODE_SECTION_INFO@{ub} to store the information into the @{b}symbol_ref@{ub}, and then check for it here. When you see a @{b}const@{ub}, you will have to look inside it to find the @{b}symbol_ref@{ub} in order to determine the section. See @{"Assembler Format" Link "Assembler Format"}. The best way to modify the name string is by adding text to the beginning, with suitable punctuation to prevent any ambiguity. Allocate the new name in @{b}saveable_obstack@{ub}. You will have to modify @{b}ASM_OUTPUT_LABELREF@{ub} to remove and decode the added text and output the name accordingly, and define @{b}STRIP_NAME_ENCODING@{ub} to access the original name string. You can check the information stored here into the @{b}symbol_ref@{ub} in the definitions of the macros @{b}GO_IF_LEGITIMATE_ADDRESS@{ub} and @{b}PRINT_OPERAND_ADDRESS@{ub}. @{b}REG_OK_FOR_BASE_P (@{I}X@{ui})@{ub} A C expression that is nonzero if @{I}X@{ui} (assumed to be a @{b}reg@{ub} RTX) is valid for use as a base register. For hard registers, it should always accept those which the hardware permits and reject the others. Whether the macro accepts or rejects pseudo registers must be controlled by @{b}REG_OK_STRICT@{ub} as described above. This usually requires two variant definitions, of which @{b}REG_OK_STRICT@{ub} controls the one actually used. @{b}REG_OK_FOR_INDEX_P (@{I}X@{ui})@{ub} A C expression that is nonzero if @{I}X@{ui} (assumed to be a @{b}reg@{ub} RTX) is valid for use as an index register. The difference between an index register and a base register is that the index register may be scaled. If an address involves the sum of two registers, neither one of them scaled, then either one may be labeled the "base" and the other the "index"; but whichever labeling is used must fit the machine's constraints of which registers may serve in each capacity. The compiler will try both labelings, looking for one that is valid, and will reload one or both registers only if neither labeling works. @{b}LEGITIMIZE_ADDRESS (@{I}X@{ui}, @{I}OLDX@{ui}, @{I}MODE@{ui}, @{I}WIN@{ui})@{ub} A C compound statement that attempts to replace @{I}X@{ui} with a valid memory address for an operand of mode @{I}MODE@{ui}. @{I}WIN@{ui} will be a C statement label elsewhere in the code; the macro definition may use GO_IF_LEGITIMATE_ADDRESS (@{I}MODE@{ui}, @{I}X@{ui}, @{I}WIN@{ui}); to avoid further processing if the address has become legitimate. @{I}X@{ui} will always be the result of a call to @{b}break_out_memory_refs@{ub}, and @{I}OLDX@{ui} will be the operand that was given to that function to produce @{I}X@{ui}. The code generated by this macro should not alter the substructure of @{I}X@{ui}. If it transforms @{I}X@{ui} into a more legitimate form, it should assign @{I}X@{ui} (which will always be a C variable) a new value. It is not necessary for this macro to come up with a legitimate address. The compiler has standard ways of doing so in all cases. In fact, it is safe for this macro to do nothing. But often a machine-dependent strategy can generate better code. @{b}GO_IF_MODE_DEPENDENT_ADDRESS (@{I}ADDR@{ui}, @{I}LABEL@{ui})@{ub} A C statement or compound statement with a conditional @{b}goto @{I}LABEL@{ui};@{ub} executed if memory address @{I}X@{ui} (an RTX) can have different meanings depending on the machine mode of the memory reference it is used for or if the address is valid for some modes but not others. Autoincrement and autodecrement addresses typically have mode-dependent effects because the amount of the increment or decrement is the size of the operand being addressed. Some machines have other mode-dependent addresses. Many RISC machines have no mode-dependent addresses. You may assume that @{I}ADDR@{ui} is a valid address for the machine. @{b}LEGITIMATE_CONSTANT_P (@{I}X@{ui})@{ub} A C expression that is nonzero if @{I}X@{ui} is a legitimate constant for an immediate operand on the target machine. You can assume that @{I}X@{ui} satisfies @{b}CONSTANT_P@{ub}, so you need not check this. In fact, @{b}1@{ub} is a suitable definition for this macro on machines where anything @{b}CONSTANT_P@{ub} is valid. @EndNode @Node "Condition Code" "gcc.guide/Condition Code" @Next "Costs" @Prev "Addressing Modes" @Toc "Target Macros" Condition Code Status ===================== This describes the condition code status. The file @{b}conditions.h@{ub} defines a variable @{b}cc_status@{ub} to describe how the condition code was computed (in case the interpretation of the condition code depends on the instruction that it was set by). This variable contains the RTL expressions on which the condition code is currently based, and several standard flags. Sometimes additional machine-specific flags must be defined in the machine description header file. It can also add additional machine-specific information by defining @{b}CC_STATUS_MDEP@{ub}. @{b}CC_STATUS_MDEP@{ub} C code for a data type which is used for declaring the @{b}mdep@{ub} component of @{b}cc_status@{ub}. It defaults to @{b}int@{ub}. This macro is not used on machines that do not use @{b}cc0@{ub}. @{b}CC_STATUS_MDEP_INIT@{ub} A C expression to initialize the @{b}mdep@{ub} field to "empty". The default definition does nothing, since most machines don't use the field anyway. If you want to use the field, you should probably define this macro to initialize it. This macro is not used on machines that do not use @{b}cc0@{ub}. @{b}NOTICE_UPDATE_CC (@{I}EXP@{ui}, @{I}INSN@{ui})@{ub} A C compound statement to set the components of @{b}cc_status@{ub} appropriately for an insn @{I}INSN@{ui} whose body is @{I}EXP@{ui}. It is this macro's responsibility to recognize insns that set the condition code as a byproduct of other activity as well as those that explicitly set @{b}(cc0)@{ub}. This macro is not used on machines that do not use @{b}cc0@{ub}. If there are insns that do not set the condition code but do alter other machine registers, this macro must check to see whether they invalidate the expressions that the condition code is recorded as reflecting. For example, on the 68000, insns that store in address registers do not set the condition code, which means that usually @{b}NOTICE_UPDATE_CC@{ub} can leave @{b}cc_status@{ub} unaltered for such insns. But suppose that the previous insn set the condition code based on location @{b}a4@(102)@{ub} and the current insn stores a new value in @{b}a4@{ub}. Although the condition code is not changed by this, it will no longer be true that it reflects the contents of @{b}a4@(102)@{ub}. Therefore, @{b}NOTICE_UPDATE_CC@{ub} must alter @{b}cc_status@{ub} in this case to say that nothing is known about the condition code value. The definition of @{b}NOTICE_UPDATE_CC@{ub} must be prepared to deal with the results of peephole optimization: insns whose patterns are @{b}parallel@{ub} RTXs containing various @{b}reg@{ub}, @{b}mem@{ub} or constants which are just the operands. The RTL structure of these insns is not sufficient to indicate what the insns actually do. What @{b}NOTICE_UPDATE_CC@{ub} should do when it sees one is just to run @{b}CC_STATUS_INIT@{ub}. A possible definition of @{b}NOTICE_UPDATE_CC@{ub} is to call a function that looks at an attribute (see @{"Insn Attributes" Link "Insn Attributes"}) named, for example, @{b}cc@{ub}. This avoids having detailed information about patterns in two places, the @{b}md@{ub} file and in @{b}NOTICE_UPDATE_CC@{ub}. @{b}EXTRA_CC_MODES@{ub} A list of names to be used for additional modes for condition code values in registers (see @{"Jump Patterns" Link "Jump Patterns"}). These names are added to @{b}enum machine_mode@{ub} and all have class @{b}MODE_CC@{ub}. By convention, they should start with @{b}CC@{ub} and end with @{b}mode@{ub}. You should only define this macro if your machine does not use @{b}cc0@{ub} and only if additional modes are required. @{b}EXTRA_CC_NAMES@{ub} A list of C strings giving the names for the modes listed in @{b}EXTRA_CC_MODES@{ub}. For example, the Sparc defines this macro and @{b}EXTRA_CC_MODES@{ub} as #define EXTRA_CC_MODES CC_NOOVmode, CCFPmode, CCFPEmode #define EXTRA_CC_NAMES "CC_NOOV", "CCFP", "CCFPE" This macro is not required if @{b}EXTRA_CC_MODES@{ub} is not defined. @{b}SELECT_CC_MODE (@{I}OP@{ui}, @{I}X@{ui}, @{I}Y@{ui})@{ub} Returns a mode from class @{b}MODE_CC@{ub} to be used when comparison operation code @{I}OP@{ui} is applied to rtx @{I}X@{ui} and @{I}Y@{ui}. For example, on the Sparc, @{b}SELECT_CC_MODE@{ub} is defined as (see see @{"Jump Patterns" Link "Jump Patterns"} for a description of the reason for this definition) #define SELECT_CC_MODE(OP,X,Y) \\ (GET_MODE_CLASS (GET_MODE (X)) == MODE_FLOAT \\ ? ((OP == EQ || OP == NE) ? CCFPmode : CCFPEmode) \\ : ((GET_CODE (X) == PLUS || GET_CODE (X) == MINUS \\ || GET_CODE (X) == NEG) \\ ? CC_NOOVmode : CCmode)) You need not define this macro if @{b}EXTRA_CC_MODES@{ub} is not defined. @{b}CANONICALIZE_COMPARISON (@{I}CODE@{ui}, @{I}OP0@{ui}, @{I}OP1@{ui})@{ub} One some machines not all possible comparisons are defined, but you can convert an invalid comparison into a valid one. For example, the Alpha does not have a @{b}GT@{ub} comparison, but you can use an @{b}LT@{ub} comparison instead and swap the order of the operands. On such machines, define this macro to be a C statement to do any required conversions. @{I}CODE@{ui} is the initial comparison code and @{I}OP0@{ui} and @{I}OP1@{ui} are the left and right operands of the comparison, respectively. You should modify @{I}CODE@{ui}, @{I}OP0@{ui}, and @{I}OP1@{ui} as required. GNU CC will not assume that the comparison resulting from this macro is valid but will see if the resulting insn matches a pattern in the @{b}md@{ub} file. You need not define this macro if it would never change the comparison code or operands. @{b}REVERSIBLE_CC_MODE (@{I}MODE@{ui})@{ub} A C expression whose value is one if it is always safe to reverse a comparison whose mode is @{I}MODE@{ui}. If @{b}SELECT_CC_MODE@{ub} can ever return @{I}MODE@{ui} for a floating-point inequality comparison, then @{b}REVERSIBLE_CC_MODE (@{I}MODE@{ui})@{ub} must be zero. You need not define this macro if it would always returns zero or if the floating-point format is anything other than @{b}IEEE_FLOAT_FORMAT@{ub}. For example, here is the definition used on the Sparc, where floating-point inequality comparisons are always given @{b}CCFPEmode@{ub}: #define REVERSIBLE_CC_MODE(MODE) ((MODE) != CCFPEmode) @EndNode @Node "Costs" "gcc.guide/Costs" @Next "Sections" @Prev "Condition Code" @Toc "Target Macros" Describing Relative Costs of Operations ======================================= These macros let you describe the relative speed of various operations on the target machine. @{b}CONST_COSTS (@{I}X@{ui}, @{I}CODE@{ui}, @{I}OUTER_CODE@{ui})@{ub} A part of a C @{b}switch@{ub} statement that describes the relative costs of constant RTL expressions. It must contain @{b}case@{ub} labels for expression codes @{b}const_int@{ub}, @{b}const@{ub}, @{b}symbol_ref@{ub}, @{b}label_ref@{ub} and @{b}const_double@{ub}. Each case must ultimately reach a @{b}return@{ub} statement to return the relative cost of the use of that kind of constant value in an expression. The cost may depend on the precise value of the constant, which is available for examination in @{I}X@{ui}, and the rtx code of the expression in which it is contained, found in @{I}OUTER_CODE@{ui}. @{I}CODE@{ui} is the expression code--redundant, since it can be obtained with @{b}GET_CODE (@{I}X@{ui})@{ub}. @{b}RTX_COSTS (@{I}X@{ui}, @{I}CODE@{ui}, @{I}OUTER_CODE@{ui})@{ub} Like @{b}CONST_COSTS@{ub} but applies to nonconstant RTL expressions. This can be used, for example, to indicate how costly a multiply instruction is. In writing this macro, you can use the construct @{b}COSTS_N_INSNS (@{I}N@{ui})@{ub} to specify a cost equal to @{I}N@{ui} fast instructions. @{I}OUTER_CODE@{ui} is the code of the expression in which @{I}X@{ui} is contained. This macro is optional; do not define it if the default cost assumptions are adequate for the target machine. @{b}ADDRESS_COST (@{I}ADDRESS@{ui})@{ub} An expression giving the cost of an addressing mode that contains @{I}ADDRESS@{ui}. If not defined, the cost is computed from the @{I}ADDRESS@{ui} expression and the @{b}CONST_COSTS@{ub} values. For most CISC machines, the default cost is a good approximation of the true cost of the addressing mode. However, on RISC machines, all instructions normally have the same length and execution time. Hence all addresses will have equal costs. In cases where more than one form of an address is known, the form with the lowest cost will be used. If multiple forms have the same, lowest, cost, the one that is the most complex will be used. For example, suppose an address that is equal to the sum of a register and a constant is used twice in the same basic block. When this macro is not defined, the address will be computed in a register and memory references will be indirect through that register. On machines where the cost of the addressing mode containing the sum is no higher than that of a simple indirect reference, this will produce an additional instruction and possibly require an additional register. Proper specification of this macro eliminates this overhead for such machines. Similar use of this macro is made in strength reduction of loops. @{I}ADDRESS@{ui} need not be valid as an address. In such a case, the cost is not relevant and can be any value; invalid addresses need not be assigned a different cost. On machines where an address involving more than one register is as cheap as an address computation involving only one register, defining @{b}ADDRESS_COST@{ub} to reflect this can cause two registers to be live over a region of code where only one would have been if @{b}ADDRESS_COST@{ub} were not defined in that manner. This effect should be considered in the definition of this macro. Equivalent costs should probably only be given to addresses with different numbers of registers on machines with lots of registers. This macro will normally either not be defined or be defined as a constant. @{b}REGISTER_MOVE_COST (@{I}FROM@{ui}, @{I}TO@{ui})@{ub} A C expression for the cost of moving data from a register in class @{I}FROM@{ui} to one in class @{I}TO@{ui}. The classes are expressed using the enumeration values such as @{b}GENERAL_REGS@{ub}. A value of 4 is the default; other values are interpreted relative to that. It is not required that the cost always equal 2 when @{I}FROM@{ui} is the same as @{I}TO@{ui}; on some machines it is expensive to move between registers if they are not general registers. If reload sees an insn consisting of a single @{b}set@{ub} between two hard registers, and if @{b}REGISTER_MOVE_COST@{ub} applied to their classes returns a value of 2, reload does not check to ensure that the constraints of the insn are met. Setting a cost of other than 2 will allow reload to verify that the constraints are met. You should do this if the @{b}mov@{I}M@{ui}@{ub} pattern's constraints do not allow such copying. @{b}MEMORY_MOVE_COST (@{I}M@{ui})@{ub} A C expression for the cost of moving data of mode @{I}M@{ui} between a register and memory. A value of 2 is the default; this cost is relative to those in @{b}REGISTER_MOVE_COST@{ub}. If moving between registers and memory is more expensive than between two registers, you should define this macro to express the relative cost. @{b}BRANCH_COST@{ub} A C expression for the cost of a branch instruction. A value of 1 is the default; other values are interpreted relative to that. Here are additional macros which do not specify precise relative costs, but only that certain actions are more expensive than GNU CC would ordinarily expect. @{b}SLOW_BYTE_ACCESS@{ub} Define this macro as a C expression which is nonzero if accessing less than a word of memory (i.e. a @{b}char@{ub} or a @{b}short@{ub}) is no faster than accessing a word of memory, i.e., if such access require more than one instruction or if there is no difference in cost between byte and (aligned) word loads. When this macro is not defined, the compiler will access a field by finding the smallest containing object; when it is defined, a fullword load will be used if alignment permits. Unless bytes accesses are faster than word accesses, using word accesses is preferable since it may eliminate subsequent memory access if subsequent accesses occur to other fields in the same word of the structure, but to different bytes. @{b}SLOW_ZERO_EXTEND@{ub} Define this macro if zero-extension (of a @{b}char@{ub} or @{b}short@{ub} to an @{b}int@{ub}) can be done faster if the destination is a register that is known to be zero. If you define this macro, you must have instruction patterns that recognize RTL structures like this: (set (strict_low_part (subreg:QI (reg:SI ...) 0)) ...) and likewise for @{b}HImode@{ub}. @{b}SLOW_UNALIGNED_ACCESS@{ub} Define this macro to be the value 1 if unaligned accesses have a cost many times greater than aligned accesses, for example if they are emulated in a trap handler. When this macro is non-zero, the compiler will act as if @{b}STRICT_ALIGNMENT@{ub} were non-zero when generating code for block moves. This can cause significantly more instructions to be produced. Therefore, do not set this macro non-zero if unaligned accesses only add a cycle or two to the time for a memory access. If the value of this macro is always zero, it need not be defined. @{b}DONT_REDUCE_ADDR@{ub} Define this macro to inhibit strength reduction of memory addresses. (On some machines, such strength reduction seems to do harm rather than good.) @{b}MOVE_RATIO@{ub} The number of scalar move insns which should be generated instead of a string move insn or a library call. Increasing the value will always make code faster, but eventually incurs high cost in increased code size. If you don't define this, a reasonable default is used. @{b}NO_FUNCTION_CSE@{ub} Define this macro if it is as good or better to call a constant function address than to call an address kept in a register. @{b}NO_RECURSIVE_FUNCTION_CSE@{ub} Define this macro if it is as good or better for a function to call itself with an explicit address than to call an address kept in a register. @{b}ADJUST_COST (@{I}INSN@{ui}, @{I}LINK@{ui}, @{I}DEP_INSN@{ui}, @{I}COST@{ui})@{ub} A C statement (sans semicolon) to update the integer variable @{I}COST@{ui} based on the relationship between @{I}INSN@{ui} that is dependent on @{I}DEP_INSN@{ui} through the dependence @{I}LINK@{ui}. The default is to make no adjustment to @{I}COST@{ui}. This can be used for example to specify to the scheduler that an output- or anti-dependence does not incur the same cost as a data-dependence. @EndNode @Node "Sections" "gcc.guide/Sections" @Next "PIC" @Prev "Costs" @Toc "Target Macros" Dividing the Output into Sections (Texts, Data, ...) ==================================================== An object file is divided into sections containing different types of data. In the most common case, there are three sections: the @{i}text section@{ui}, which holds instructions and read-only data; the @{i}data section@{ui}, which holds initialized writable data; and the @{i}bss section@{ui}, which holds uninitialized data. Some systems have other kinds of sections. The compiler must tell the assembler when to switch sections. These macros control what commands to output to tell the assembler this. You can also define additional sections. @{b}TEXT_SECTION_ASM_OP@{ub} A C expression whose value is a string containing the assembler operation that should precede instructions and read-only data. Normally @{b}".text"@{ub} is right. @{b}DATA_SECTION_ASM_OP@{ub} A C expression whose value is a string containing the assembler operation to identify the following data as writable initialized data. Normally @{b}".data"@{ub} is right. @{b}SHARED_SECTION_ASM_OP@{ub} if defined, a C expression whose value is a string containing the assembler operation to identify the following data as shared data. If not defined, @{b}DATA_SECTION_ASM_OP@{ub} will be used. @{b}INIT_SECTION_ASM_OP@{ub} if defined, a C expression whose value is a string containing the assembler operation to identify the following data as initialization code. If not defined, GNU CC will assume such a section does not exist. @{b}EXTRA_SECTIONS@{ub} A list of names for sections other than the standard two, which are @{b}in_text@{ub} and @{b}in_data@{ub}. You need not define this macro on a system with no other sections (that GCC needs to use). @{b}EXTRA_SECTION_FUNCTIONS@{ub} One or more functions to be defined in @{b}varasm.c@{ub}. These functions should do jobs analogous to those of @{b}text_section@{ub} and @{b}data_section@{ub}, for your additional sections. Do not define this macro if you do not define @{b}EXTRA_SECTIONS@{ub}. @{b}READONLY_DATA_SECTION@{ub} On most machines, read-only variables, constants, and jump tables are placed in the text section. If this is not the case on your machine, this macro should be defined to be the name of a function (either @{b}data_section@{ub} or a function defined in @{b}EXTRA_SECTIONS@{ub}) that switches to the section to be used for read-only items. If these items should be placed in the text section, this macro should not be defined. @{b}SELECT_SECTION (@{I}EXP@{ui}, @{I}RELOC@{ui})@{ub} A C statement or statements to switch to the appropriate section for output of @{I}EXP@{ui}. You can assume that @{I}EXP@{ui} is either a @{b}VAR_DECL@{ub} node or a constant of some sort. @{I}RELOC@{ui} indicates whether the initial value of @{I}EXP@{ui} requires link-time relocations. Select the section by calling @{b}text_section@{ub} or one of the alternatives for other sections. Do not define this macro if you put all read-only variables and constants in the read-only data section (usually the text section). @{b}SELECT_RTX_SECTION (@{I}MODE@{ui}, @{I}RTX@{ui})@{ub} A C statement or statements to switch to the appropriate section for output of @{I}RTX@{ui} in mode @{I}MODE@{ui}. You can assume that @{I}RTX@{ui} is some kind of constant in RTL. The argument @{I}MODE@{ui} is redundant except in the case of a @{b}const_int@{ub} rtx. Select the section by calling @{b}text_section@{ub} or one of the alternatives for other sections. Do not define this macro if you put all constants in the read-only data section. @{b}JUMP_TABLES_IN_TEXT_SECTION@{ub} Define this macro if jump tables (for @{b}tablejump@{ub} insns) should be output in the text section, along with the assembler instructions. Otherwise, the readonly data section is used. This macro is irrelevant if there is no separate readonly data section. @{b}ENCODE_SECTION_INFO (@{I}DECL@{ui})@{ub} Define this macro if references to a symbol must be treated differently depending on something about the variable or function named by the symbol (such as what section it is in). The macro definition, if any, is executed immediately after the rtl for @{I}DECL@{ui} has been created and stored in @{b}DECL_RTL (@{I}DECL@{ui})@{ub}. The value of the rtl will be a @{b}mem@{ub} whose address is a @{b}symbol_ref@{ub}. The usual thing for this macro to do is to record a flag in the @{b}symbol_ref@{ub} (such as @{b}SYMBOL_REF_FLAG@{ub}) or to store a modified name string in the @{b}symbol_ref@{ub} (if one bit is not enough information). @{b}STRIP_NAME_ENCODING (@{I}VAR@{ui}, @{I}SYM_NAME@{ui})@{ub} Decode @{I}SYM_NAME@{ui} and store the real name part in @{I}VAR@{ui}, sans the characters that encode section info. Define this macro if @{b}ENCODE_SECTION_INFO@{ub} alters the symbol's name string. @EndNode @Node "PIC" "gcc.guide/PIC" @Next "Assembler Format" @Prev "Sections" @Toc "Target Macros" Position Independent Code ========================= This section describes macros that help implement generation of position independent code. Simply defining these macros is not enough to generate valid PIC; you must also add support to the macros @{b}GO_IF_LEGITIMATE_ADDRESS@{ub} and @{b}PRINT_OPERAND_ADDRESS@{ub}, as well as @{b}LEGITIMIZE_ADDRESS@{ub}. You must modify the definition of @{b}movsi@{ub} to do something appropriate when the source operand contains a symbolic address. You may also need to alter the handling of switch statements so that they use relative addresses. @{b}PIC_OFFSET_TABLE_REGNUM@{ub} The register number of the register used to address a table of static data addresses in memory. In some cases this register is defined by a processor's "application binary interface" (ABI). When this macro is defined, RTL is generated for this register once, as with the stack pointer and frame pointer registers. If this macro is not defined, it is up to the machine-dependent files to allocate such a register (if necessary). @{b}PIC_OFFSET_TABLE_REG_CALL_CLOBBERED@{ub} Define this macro if the register defined by @{b}PIC_OFFSET_TABLE_REGNUM@{ub} is clobbered by calls. Do not define this macro if @{b}PPIC_OFFSET_TABLE_REGNUM@{ub} is not defined. @{b}FINALIZE_PIC@{ub} By generating position-independent code, when two different programs (A and B) share a common library (libC.a), the text of the library can be shared whether or not the library is linked at the same address for both programs. In some of these environments, position-independent code requires not only the use of different addressing modes, but also special code to enable the use of these addressing modes. The @{b}FINALIZE_PIC@{ub} macro serves as a hook to emit these special codes once the function is being compiled into assembly code, but not before. (It is not done before, because in the case of compiling an inline function, it would lead to multiple PIC prologues being included in functions which used inline functions and were compiled to assembly language.) @{b}LEGITIMATE_PIC_OPERAND_P (@{I}X@{ui})@{ub} A C expression that is nonzero if @{I}X@{ui} is a legitimate immediate operand on the target machine when generating position independent code. You can assume that @{I}X@{ui} satisfies @{b}CONSTANT_P@{ub}, so you need not check this. You can also assume @{I}FLAG_PIC@{ui} is true, so you need not check it either. You need not define this macro if all constants (including @{b}SYMBOL_REF@{ub}) can be immediate operands when generating position independent code. @EndNode @Node "Assembler Format" "gcc.guide/Assembler Format" @Next "Debugging Info" @Prev "PIC" @Toc "Target Macros" Defining the Output Assembler Language ====================================== This section describes macros whose principal purpose is to describe how to write instructions in assembler language-rather than what the instructions do. @{" File Framework " Link "File Framework"} Structural information for the assembler file. @{" Data Output " Link "Data Output"} Output of constants (numbers, strings, addresses). @{" Uninitialized Data " Link "Uninitialized Data"} Output of uninitialized variables. @{" Label Output " Link "Label Output"} Output and generation of labels. @{" Initialization " Link "Initialization"} General principles of initialization and termination routines. @{" Macros for Initialization " Link "Macros for Initialization"} Specific macros that control the handling of initialization and termination routines. @{" Instruction Output " Link "Instruction Output"} Output of actual instructions. @{" Dispatch Tables " Link "Dispatch Tables"} Output of jump tables. @{" Alignment Output " Link "Alignment Output"} Pseudo ops for alignment and skipping data. @EndNode @Node "File Framework" "gcc.guide/File Framework" @Next "Data Output" @Toc "Assembler Format" The Overall Framework of an Assembler File ------------------------------------------ This describes the overall framework of an assembler file. @{b}ASM_FILE_START (@{I}STREAM@{ui})@{ub} A C expression which outputs to the stdio stream @{I}STREAM@{ui} some appropriate text to go at the start of an assembler file. Normally this macro is defined to output a line containing @{b}#NO_APP@{ub}, which is a comment that has no effect on most assemblers but tells the GNU assembler that it can save time by not checking for certain assembler constructs. On systems that use SDB, it is necessary to output certain commands; see @{b}attasm.h@{ub}. @{b}ASM_FILE_END (@{I}STREAM@{ui})@{ub} A C expression which outputs to the stdio stream @{I}STREAM@{ui} some appropriate text to go at the end of an assembler file. If this macro is not defined, the default is to output nothing special at the end of the file. Most systems don't require any definition. On systems that use SDB, it is necessary to output certain commands; see @{b}attasm.h@{ub}. @{b}ASM_IDENTIFY_GCC (@{I}FILE@{ui})@{ub} A C statement to output assembler commands which will identify the object file as having been compiled with GNU CC (or another GNU compiler). If you don't define this macro, the string @{b}gcc_compiled.:@{ub} is output. This string is calculated to define a symbol which, on BSD systems, will never be defined for any other reason. GDB checks for the presence of this symbol when reading the symbol table of an executable. On non-BSD systems, you must arrange communication with GDB in some other fashion. If GDB is not used on your system, you can define this macro with an empty body. @{b}ASM_COMMENT_START@{ub} A C string constant describing how to begin a comment in the target assembler language. The compiler assumes that the comment will end at the end of the line. @{b}ASM_APP_ON@{ub} A C string constant for text to be output before each @{b}asm@{ub} statement or group of consecutive ones. Normally this is @{b}"#APP"@{ub}, which is a comment that has no effect on most assemblers but tells the GNU assembler that it must check the lines that follow for all valid assembler constructs. @{b}ASM_APP_OFF@{ub} A C string constant for text to be output after each @{b}asm@{ub} statement or group of consecutive ones. Normally this is @{b}"#NO_APP"@{ub}, which tells the GNU assembler to resume making the time-saving assumptions that are valid for ordinary compiler output. @{b}ASM_OUTPUT_SOURCE_FILENAME (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} A C statement to output COFF information or DWARF debugging information which indicates that filename @{I}NAME@{ui} is the current source file to the stdio stream @{I}STREAM@{ui}. This macro need not be defined if the standard form of output for the file format in use is appropriate. @{b}ASM_OUTPUT_SOURCE_LINE (@{I}STREAM@{ui}, @{I}LINE@{ui})@{ub} A C statement to output DBX or SDB debugging information before code for line number @{I}LINE@{ui} of the current source file to the stdio stream @{I}STREAM@{ui}. This macro need not be defined if the standard form of debugging information for the debugger in use is appropriate. @{b}ASM_OUTPUT_IDENT (@{I}STREAM@{ui}, @{I}STRING@{ui})@{ub} A C statement to output something to the assembler file to handle a @{b}#ident@{ub} directive containing the text @{I}STRING@{ui}. If this macro is not defined, nothing is output for a @{b}#ident@{ub} directive. @{b}ASM_OUTPUT_SECTION_NAME (@{I}STREAM@{ui}, @{I}DECL@{ui}, @{I}NAME@{ui})@{ub} A C statement to output something to the assembler file to switch to section @{I}NAME@{ui} for object @{I}DECL@{ui} which is either a @{b}FUNCTION_DECL@{ub}, a @{b}VAR_DECL@{ub} or @{b}NULL_TREE@{ub}. Some target formats do not support arbitrary sections. Do not define this macro in such cases. At present this macro is only used to support section attributes. When this macro is undefined, section attributes are disabled. @{b}OBJC_PROLOGUE@{ub} A C statement to output any assembler statements which are required to precede any Objective C object definitions or message sending. The statement is executed only when compiling an Objective C program. @EndNode @Node "Data Output" "gcc.guide/Data Output" @Next "Uninitialized Data" @Prev "File Framework" @Toc "Assembler Format" Output of Data -------------- This describes data output. @{b}ASM_OUTPUT_LONG_DOUBLE (@{I}STREAM@{ui}, @{I}VALUE@{ui})@{ub} @{b}ASM_OUTPUT_DOUBLE (@{I}STREAM@{ui}, @{I}VALUE@{ui})@{ub} @{b}ASM_OUTPUT_FLOAT (@{I}STREAM@{ui}, @{I}VALUE@{ui})@{ub} @{b}ASM_OUTPUT_THREE_QUARTER_FLOAT (@{I}STREAM@{ui}, @{I}VALUE@{ui})@{ub} @{b}ASM_OUTPUT_SHORT_FLOAT (@{I}STREAM@{ui}, @{I}VALUE@{ui})@{ub} @{b}ASM_OUTPUT_BYTE_FLOAT (@{I}STREAM@{ui}, @{I}VALUE@{ui})@{ub} A C statement to output to the stdio stream @{I}STREAM@{ui} an assembler instruction to assemble a floating-point constant of @{b}TFmode@{ub}, @{b}DFmode@{ub}, @{b}SFmode@{ub}, @{b}TQFmode@{ub}, @{b}HFmode@{ub}, or @{b}QFmode@{ub}, respectively, whose value is @{I}VALUE@{ui}. @{I}VALUE@{ui} will be a C expression of type @{b}REAL_VALUE_TYPE@{ub}. Macros such as @{b}REAL_VALUE_TO_TARGET_DOUBLE@{ub} are useful for writing these definitions. @{b}ASM_OUTPUT_QUADRUPLE_INT (@{I}STREAM@{ui}, @{I}EXP@{ui})@{ub} @{b}ASM_OUTPUT_DOUBLE_INT (@{I}STREAM@{ui}, @{I}EXP@{ui})@{ub} @{b}ASM_OUTPUT_INT (@{I}STREAM@{ui}, @{I}EXP@{ui})@{ub} @{b}ASM_OUTPUT_SHORT (@{I}STREAM@{ui}, @{I}EXP@{ui})@{ub} @{b}ASM_OUTPUT_CHAR (@{I}STREAM@{ui}, @{I}EXP@{ui})@{ub} A C statement to output to the stdio stream @{I}STREAM@{ui} an assembler instruction to assemble an integer of 16, 8, 4, 2 or 1 bytes, respectively, whose value is @{I}VALUE@{ui}. The argument @{I}EXP@{ui} will be an RTL expression which represents a constant value. Use @{b}output_addr_const (@{I}STREAM@{ui}, @{I}EXP@{ui})@{ub} to output this value as an assembler expression. For sizes larger than @{b}UNITS_PER_WORD@{ub}, if the action of a macro would be identical to repeatedly calling the macro corresponding to a size of @{b}UNITS_PER_WORD@{ub}, once for each word, you need not define the macro. @{b}ASM_OUTPUT_BYTE (@{I}STREAM@{ui}, @{I}VALUE@{ui})@{ub} A C statement to output to the stdio stream @{I}STREAM@{ui} an assembler instruction to assemble a single byte containing the number @{I}VALUE@{ui}. @{b}ASM_BYTE_OP@{ub} A C string constant giving the pseudo-op to use for a sequence of single-byte constants. If this macro is not defined, the default is @{b}"byte"@{ub}. @{b}ASM_OUTPUT_ASCII (@{I}STREAM@{ui}, @{I}PTR@{ui}, @{I}LEN@{ui})@{ub} A C statement to output to the stdio stream @{I}STREAM@{ui} an assembler instruction to assemble a string constant containing the @{I}LEN@{ui} bytes at @{I}PTR@{ui}. @{I}PTR@{ui} will be a C expression of type @{b}char *@{ub} and @{I}LEN@{ui} a C expression of type @{b}int@{ub}. If the assembler has a @{b}.ascii@{ub} pseudo-op as found in the Berkeley Unix assembler, do not define the macro @{b}ASM_OUTPUT_ASCII@{ub}. @{b}ASM_OUTPUT_POOL_PROLOGUE (@{I}FILE@{ui} @{I}FUNNAME@{ui} @{I}FUNDECL@{ui} @{I}SIZE@{ui})@{ub} A C statement to output assembler commands to define the start of the constant pool for a function. @{I}FUNNAME@{ui} is a string giving the name of the function. Should the return type of the function be required, it can be obtained via @{I}FUNDECL@{ui}. @{I}SIZE@{ui} is the size, in bytes, of the constant pool that will be written immediately after this call. If no constant-pool prefix is required, the usual case, this macro need not be defined. @{b}ASM_OUTPUT_SPECIAL_POOL_ENTRY (@{I}FILE@{ui}, @{I}X@{ui}, @{I}MODE@{ui}, @{I}ALIGN@{ui}, @{I}LABELNO@{ui}, @{I}JUMPTO@{ui})@{ub} A C statement (with or without semicolon) to output a constant in the constant pool, if it needs special treatment. (This macro need not do anything for RTL expressions that can be output normally.) The argument @{I}FILE@{ui} is the standard I/O stream to output the assembler code on. @{I}X@{ui} is the RTL expression for the constant to output, and @{I}MODE@{ui} is the machine mode (in case @{I}X@{ui} is a @{b}const_int@{ub}). @{I}ALIGN@{ui} is the required alignment for the value @{I}X@{ui}; you should output an assembler directive to force this much alignment. The argument @{I}LABELNO@{ui} is a number to use in an internal label for the address of this pool entry. The definition of this macro is responsible for outputting the label definition at the proper place. Here is how to do this: ASM_OUTPUT_INTERNAL_LABEL (@{I}FILE@{ui}, "LC", @{I}LABELNO@{ui}); When you output a pool entry specially, you should end with a @{b}goto@{ub} to the label @{I}JUMPTO@{ui}. This will prevent the same pool entry from being output a second time in the usual manner. You need not define this macro if it would do nothing. @{b}IS_ASM_LOGICAL_LINE_SEPARATOR (@{I}C@{ui})@{ub} Define this macro as a C expression which is nonzero if @{I}C@{ui} is used as a logical line separator by the assembler. If you do not define this macro, the default is that only the character @{b};@{ub} is treated as a logical line separator. @{b}ASM_OPEN_PAREN@{ub} @{b}ASM_CLOSE_PAREN@{ub} These macros are defined as C string constant, describing the syntax in the assembler for grouping arithmetic expressions. The following definitions are correct for most assemblers: #define ASM_OPEN_PAREN "(" #define ASM_CLOSE_PAREN ")" These macros are provided by @{b}real.h@{ub} for writing the definitions of @{b}ASM_OUTPUT_DOUBLE@{ub} and the like: @{b}REAL_VALUE_TO_TARGET_SINGLE (@{I}X@{ui}, @{I}L@{ui})@{ub} @{b}REAL_VALUE_TO_TARGET_DOUBLE (@{I}X@{ui}, @{I}L@{ui})@{ub} @{b}REAL_VALUE_TO_TARGET_LONG_DOUBLE (@{I}X@{ui}, @{I}L@{ui})@{ub} These translate @{I}X@{ui}, of type @{b}REAL_VALUE_TYPE@{ub}, to the target's floating point representation, and store its bit pattern in the array of @{b}long int@{ub} whose address is @{I}L@{ui}. The number of elements in the output array is determined by the size of the desired target floating point data type: 32 bits of it go in each @{b}long int@{ub} array element. Each array element holds 32 bits of the result, even if @{b}long int@{ub} is wider than 32 bits on the host machine. The array element values are designed so that you can print them out using @{b}fprintf@{ub} in the order they should appear in the target machine's memory. @{b}REAL_VALUE_TO_DECIMAL (@{I}X@{ui}, @{I}FORMAT@{ui}, @{I}STRING@{ui})@{ub} This macro converts @{I}X@{ui}, of type @{b}REAL_VALUE_TYPE@{ub}, to a decimal number and stores it as a string into @{I}STRING@{ui}. You must pass, as @{I}STRING@{ui}, the address of a long enough block of space to hold the result. The argument @{I}FORMAT@{ui} is a @{b}printf@{ub}-specification that serves as a suggestion for how to format the output string. @EndNode @Node "Uninitialized Data" "gcc.guide/Uninitialized Data" @Next "Label Output" @Prev "Data Output" @Toc "Assembler Format" Output of Uninitialized Variables --------------------------------- Each of the macros in this section is used to do the whole job of outputting a single uninitialized variable. @{b}ASM_OUTPUT_COMMON (@{I}STREAM@{ui}, @{I}NAME@{ui}, @{I}SIZE@{ui}, @{I}ROUNDED@{ui})@{ub} A C statement (sans semicolon) to output to the stdio stream @{I}STREAM@{ui} the assembler definition of a common-label named @{I}NAME@{ui} whose size is @{I}SIZE@{ui} bytes. The variable @{I}ROUNDED@{ui} is the size rounded up to whatever alignment the caller wants. Use the expression @{b}assemble_name (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} to output the name itself; before and after that, output the additional assembler syntax for defining the name, and a newline. This macro controls how the assembler definitions of uninitialized global variables are output. @{b}ASM_OUTPUT_ALIGNED_COMMON (@{I}STREAM@{ui}, @{I}NAME@{ui}, @{I}SIZE@{ui}, @{I}ALIGNMENT@{ui})@{ub} Like @{b}ASM_OUTPUT_COMMON@{ub} except takes the required alignment as a separate, explicit argument. If you define this macro, it is used in place of @{b}ASM_OUTPUT_COMMON@{ub}, and gives you more flexibility in handling the required alignment of the variable. The alignment is specified as the number of bits. @{b}ASM_OUTPUT_SHARED_COMMON (@{I}STREAM@{ui}, @{I}NAME@{ui}, @{I}SIZE@{ui}, @{I}ROUNDED@{ui})@{ub} If defined, it is similar to @{b}ASM_OUTPUT_COMMON@{ub}, except that it is used when @{I}NAME@{ui} is shared. If not defined, @{b}ASM_OUTPUT_COMMON@{ub} will be used. @{b}ASM_OUTPUT_LOCAL (@{I}STREAM@{ui}, @{I}NAME@{ui}, @{I}SIZE@{ui}, @{I}ROUNDED@{ui})@{ub} A C statement (sans semicolon) to output to the stdio stream @{I}STREAM@{ui} the assembler definition of a local-common-label named @{I}NAME@{ui} whose size is @{I}SIZE@{ui} bytes. The variable @{I}ROUNDED@{ui} is the size rounded up to whatever alignment the caller wants. Use the expression @{b}assemble_name (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} to output the name itself; before and after that, output the additional assembler syntax for defining the name, and a newline. This macro controls how the assembler definitions of uninitialized static variables are output. @{b}ASM_OUTPUT_ALIGNED_LOCAL (@{I}STREAM@{ui}, @{I}NAME@{ui}, @{I}SIZE@{ui}, @{I}ALIGNMENT@{ui})@{ub} Like @{b}ASM_OUTPUT_LOCAL@{ub} except takes the required alignment as a separate, explicit argument. If you define this macro, it is used in place of @{b}ASM_OUTPUT_LOCAL@{ub}, and gives you more flexibility in handling the required alignment of the variable. The alignment is specified as the number of bits. @{b}ASM_OUTPUT_SHARED_LOCAL (@{I}STREAM@{ui}, @{I}NAME@{ui}, @{I}SIZE@{ui}, @{I}ROUNDED@{ui})@{ub} If defined, it is similar to @{b}ASM_OUTPUT_LOCAL@{ub}, except that it is used when @{I}NAME@{ui} is shared. If not defined, @{b}ASM_OUTPUT_LOCAL@{ub} will be used. @EndNode @Node "Label Output" "gcc.guide/Label Output" @Next "Initialization" @Prev "Uninitialized Data" @Toc "Assembler Format" Output and Generation of Labels ------------------------------- This is about outputting labels. @{b}ASM_OUTPUT_LABEL (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} A C statement (sans semicolon) to output to the stdio stream @{I}STREAM@{ui} the assembler definition of a label named @{I}NAME@{ui}. Use the expression @{b}assemble_name (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} to output the name itself; before and after that, output the additional assembler syntax for defining the name, and a newline. @{b}ASM_DECLARE_FUNCTION_NAME (@{I}STREAM@{ui}, @{I}NAME@{ui}, @{I}DECL@{ui})@{ub} A C statement (sans semicolon) to output to the stdio stream @{I}STREAM@{ui} any text necessary for declaring the name @{I}NAME@{ui} of a function which is being defined. This macro is responsible for outputting the label definition (perhaps using @{b}ASM_OUTPUT_LABEL@{ub}). The argument @{I}DECL@{ui} is the @{b}FUNCTION_DECL@{ub} tree node representing the function. If this macro is not defined, then the function name is defined in the usual manner as a label (by means of @{b}ASM_OUTPUT_LABEL@{ub}). @{b}ASM_DECLARE_FUNCTION_SIZE (@{I}STREAM@{ui}, @{I}NAME@{ui}, @{I}DECL@{ui})@{ub} A C statement (sans semicolon) to output to the stdio stream @{I}STREAM@{ui} any text necessary for declaring the size of a function which is being defined. The argument @{I}NAME@{ui} is the name of the function. The argument @{I}DECL@{ui} is the @{b}FUNCTION_DECL@{ub} tree node representing the function. If this macro is not defined, then the function size is not defined. @{b}ASM_DECLARE_OBJECT_NAME (@{I}STREAM@{ui}, @{I}NAME@{ui}, @{I}DECL@{ui})@{ub} A C statement (sans semicolon) to output to the stdio stream @{I}STREAM@{ui} any text necessary for declaring the name @{I}NAME@{ui} of an initialized variable which is being defined. This macro must output the label definition (perhaps using @{b}ASM_OUTPUT_LABEL@{ub}). The argument @{I}DECL@{ui} is the @{b}VAR_DECL@{ub} tree node representing the variable. If this macro is not defined, then the variable name is defined in the usual manner as a label (by means of @{b}ASM_OUTPUT_LABEL@{ub}). @{b}ASM_FINISH_DECLARE_OBJECT (@{I}STREAM@{ui}, @{I}DECL@{ui}, @{I}TOPLEVEL@{ui}, @{I}ATEND@{ui})@{ub} A C statement (sans semicolon) to finish up declaring a variable name once the compiler has processed its initializer fully and thus has had a chance to determine the size of an array when controlled by an initializer. This is used on systems where it's necessary to declare something about the size of the object. If you don't define this macro, that is equivalent to defining it to do nothing. @{b}ASM_GLOBALIZE_LABEL (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} A C statement (sans semicolon) to output to the stdio stream @{I}STREAM@{ui} some commands that will make the label @{I}NAME@{ui} global; that is, available for reference from other files. Use the expression @{b}assemble_name (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} to output the name itself; before and after that, output the additional assembler syntax for making that name global, and a newline. @{b}ASM_WEAKEN_LABEL@{ub} A C statement (sans semicolon) to output to the stdio stream @{I}STREAM@{ui} some commands that will make the label @{I}NAME@{ui} weak; that is, available for reference from other files but only used if no other definition is available. Use the expression @{b}assemble_name (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} to output the name itself; before and after that, output the additional assembler syntax for making that name weak, and a newline. If you don't define this macro, GNU CC will not support weak symbols and you should not define the @{b}SUPPORTS_WEAK@{ub} macro. @{b}SUPPORTS_WEAK@{ub} A C expression which evaluates to true if the target supports weak symbols. If you don't define this macro, @{b}defaults.h@{ub} provides a default definition. If @{b}ASM_WEAKEN_LABEL@{ub} is defined, the default definition is @{b}1@{ub}; otherwise, it is @{b}0@{ub}. Define this macro if you want to control weak symbol support with a compiler flag such as @{b}-melf@{ub}. @{b}ASM_OUTPUT_EXTERNAL (@{I}STREAM@{ui}, @{I}DECL@{ui}, @{I}NAME@{ui})@{ub} A C statement (sans semicolon) to output to the stdio stream @{I}STREAM@{ui} any text necessary for declaring the name of an external symbol named @{I}NAME@{ui} which is referenced in this compilation but not defined. The value of @{I}DECL@{ui} is the tree node for the declaration. This macro need not be defined if it does not need to output anything. The GNU assembler and most Unix assemblers don't require anything. @{b}ASM_OUTPUT_EXTERNAL_LIBCALL (@{I}STREAM@{ui}, @{I}SYMREF@{ui})@{ub} A C statement (sans semicolon) to output on @{I}STREAM@{ui} an assembler pseudo-op to declare a library function name external. The name of the library function is given by @{I}SYMREF@{ui}, which has type @{b}rtx@{ub} and is a @{b}symbol_ref@{ub}. This macro need not be defined if it does not need to output anything. The GNU assembler and most Unix assemblers don't require anything. @{b}ASM_OUTPUT_LABELREF (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} A C statement (sans semicolon) to output to the stdio stream @{I}STREAM@{ui} a reference in assembler syntax to a label named @{I}NAME@{ui}. This should add @{b}_@{ub} to the front of the name, if that is customary on your operating system, as it is in most Berkeley Unix systems. This macro is used in @{b}assemble_name@{ub}. @{b}ASM_OUTPUT_INTERNAL_LABEL (@{I}STREAM@{ui}, @{I}PREFIX@{ui}, @{I}NUM@{ui})@{ub} A C statement to output to the stdio stream @{I}STREAM@{ui} a label whose name is made from the string @{I}PREFIX@{ui} and the number @{I}NUM@{ui}. It is absolutely essential that these labels be distinct from the labels used for user-level functions and variables. Otherwise, certain programs will have name conflicts with internal labels. It is desirable to exclude internal labels from the symbol table of the object file. Most assemblers have a naming convention for labels that should be excluded; on many systems, the letter @{b}L@{ub} at the beginning of a label has this effect. You should find out what convention your system uses, and follow it. The usual definition of this macro is as follows: fprintf (@{I}STREAM@{ui}, "L%s%d:\\n", @{I}PREFIX@{ui}, @{I}NUM@{ui}) @{b}ASM_GENERATE_INTERNAL_LABEL (@{I}STRING@{ui}, @{I}PREFIX@{ui}, @{I}NUM@{ui})@{ub} A C statement to store into the string @{I}STRING@{ui} a label whose name is made from the string @{I}PREFIX@{ui} and the number @{I}NUM@{ui}. This string, when output subsequently by @{b}assemble_name@{ub}, should produce the output that @{b}ASM_OUTPUT_INTERNAL_LABEL@{ub} would produce with the same @{I}PREFIX@{ui} and @{I}NUM@{ui}. If the string begins with @{b}*@{ub}, then @{b}assemble_name@{ub} will output the rest of the string unchanged. It is often convenient for @{b}ASM_GENERATE_INTERNAL_LABEL@{ub} to use @{b}*@{ub} in this way. If the string doesn't start with @{b}*@{ub}, then @{b}ASM_OUTPUT_LABELREF@{ub} gets to output the string, and may change it. (Of course, @{b}ASM_OUTPUT_LABELREF@{ub} is also part of your machine description, so you should know what it does on your machine.) @{b}ASM_FORMAT_PRIVATE_NAME (@{I}OUTVAR@{ui}, @{I}NAME@{ui}, @{I}NUMBER@{ui})@{ub} A C expression to assign to @{I}OUTVAR@{ui} (which is a variable of type @{b}char *@{ub}) a newly allocated string made from the string @{I}NAME@{ui} and the number @{I}NUMBER@{ui}, with some suitable punctuation added. Use @{b}alloca@{ub} to get space for the string. The string will be used as an argument to @{b}ASM_OUTPUT_LABELREF@{ub} to produce an assembler label for an internal static variable whose name is @{I}NAME@{ui}. Therefore, the string must be such as to result in valid assembler code. The argument @{I}NUMBER@{ui} is different each time this macro is executed; it prevents conflicts between similarly-named internal static variables in different scopes. Ideally this string should not be a valid C identifier, to prevent any conflict with the user's own symbols. Most assemblers allow periods or percent signs in assembler symbols; putting at least one of these between the name and the number will suffice. @{b}ASM_OUTPUT_DEF (@{I}STREAM@{ui}, @{I}NAME@{ui}, @{I}VALUE@{ui})@{ub} A C statement to output to the stdio stream @{I}STREAM@{ui} assembler code which defines (equates) the symbol @{I}NAME@{ui} to have the value @{I}VALUE@{ui}. If SET_ASM_OP is defined, a default definition is provided which is correct for most systems. @{b}OBJC_GEN_METHOD_LABEL (@{I}BUF@{ui}, @{I}IS_INST@{ui}, @{I}CLASS_NAME@{ui}, @{I}CAT_NAME@{ui}, @{I}SEL_NAME@{ui})@{ub} Define this macro to override the default assembler names used for Objective C methods. The default name is a unique method number followed by the name of the class (e.g. @{b}_1_Foo@{ub}). For methods in categories, the name of the category is also included in the assembler name (e.g. @{b}_1_Foo_Bar@{ub}). These names are safe on most systems, but make debugging difficult since the method's selector is not present in the name. Therefore, particular systems define other ways of computing names. @{I}BUF@{ui} is an expression of type @{b}char *@{ub} which gives you a buffer in which to store the name; its length is as long as @{I}CLASS_NAME@{ui}, @{I}CAT_NAME@{ui} and @{I}SEL_NAME@{ui} put together, plus 50 characters extra. The argument @{I}IS_INST@{ui} specifies whether the method is an instance method or a class method; @{I}CLASS_NAME@{ui} is the name of the class; @{I}CAT_NAME@{ui} is the name of the category (or NULL if the method is not in a category); and @{I}SEL_NAME@{ui} is the name of the selector. On systems where the assembler can handle quoted names, you can use this macro to provide more human-readable names. @EndNode @Node "Initialization" "gcc.guide/Initialization" @Next "Macros for Initialization" @Prev "Label Output" @Toc "Assembler Format" How Initialization Functions Are Handled ---------------------------------------- The compiled code for certain languages includes @{i}constructors@{ui} (also called @{i}initialization routines@{ui})--functions to initialize data in the program when the program is started. These functions need to be called before the program is "started"--that is to say, before @{b}main@{ub} is called. Compiling some languages generates @{i}destructors@{ui} (also called @{i}termination routines@{ui}) that should be called when the program terminates. To make the initialization and termination functions work, the compiler must output something in the assembler code to cause those functions to be called at the appropriate time. When you port the compiler to a new system, you need to specify how to do this. There are two major ways that GCC currently supports the execution of initialization and termination functions. Each way has two variants. Much of the structure is common to all four variations. The linker must build two lists of these functions--a list of initialization functions, called @{b}__CTOR_LIST__@{ub}, and a list of termination functions, called @{b}__DTOR_LIST__@{ub}. Each list always begins with an ignored function pointer (which may hold 0, -1, or a count of the function pointers after it, depending on the environment). This is followed by a series of zero or more function pointers to constructors (or destructors), followed by a function pointer containing zero. Depending on the operating system and its executable file format, either @{b}crtstuff.c@{ub} or @{b}libgcc2.c@{ub} traverses these lists at startup time and exit time. Constructors are called in reverse order of the list; destructors in forward order. The best way to handle static constructors works only for object file formats which provide arbitrarily-named sections. A section is set aside for a list of constructors, and another for a list of destructors. Traditionally these are called @{b}.ctors@{ub} and @{b}.dtors@{ub}. Each object file that defines an initialization function also puts a word in the constructor section to point to that function. The linker accumulates all these words into one contiguous @{b}.ctors@{ub} section. Termination functions are handled similarly. To use this method, you need appropriate definitions of the macros @{b}ASM_OUTPUT_CONSTRUCTOR@{ub} and @{b}ASM_OUTPUT_DESTRUCTOR@{ub}. Usually you can get them by including @{b}svr4.h@{ub}. When arbitrary sections are available, there are two variants, depending upon how the code in @{b}crtstuff.c@{ub} is called. On systems that support an @{i}init@{ui} section which is executed at program startup, parts of @{b}crtstuff.c@{ub} are compiled into that section. The program is linked by the @{b}gcc@{ub} driver like this: ld -o @{I}OUTPUT_FILE@{ui} crtbegin.o ... crtend.o -lgcc The head of a function (@{b}__do_global_ctors@{ub}) appears in the init section of @{b}crtbegin.o@{ub}; the remainder of the function appears in the init section of @{b}crtend.o@{ub}. The linker will pull these two parts of the section together, making a whole function. If any of the user's object files linked into the middle of it contribute code, then that code will be executed as part of the body of @{b}__do_global_ctors@{ub}. To use this variant, you must define the @{b}INIT_SECTION_ASM_OP@{ub} macro properly. If no init section is available, do not define @{b}INIT_SECTION_ASM_OP@{ub}. Then @{b}__do_global_ctors@{ub} is built into the text section like all other functions, and resides in @{b}libgcc.a@{ub}. When GCC compiles any function called @{b}main@{ub}, it inserts a procedure call to @{b}__main@{ub} as the first executable code after the function prologue. The @{b}__main@{ub} function, also defined in @{b}libgcc2.c@{ub}, simply calls @{b}__do_global_ctors@{ub}. In file formats that don't support arbitrary sections, there are again two variants. In the simplest variant, the GNU linker (GNU @{b}ld@{ub}) and an `a.out' format must be used. In this case, @{b}ASM_OUTPUT_CONSTRUCTOR@{ub} is defined to produce a @{b}.stabs@{ub} entry of type @{b}N_SETT@{ub}, referencing the name @{b}__CTOR_LIST__@{ub}, and with the address of the void function containing the initialization code as its value. The GNU linker recognizes this as a request to add the value to a "set"; the values are accumulated, and are eventually placed in the executable as a vector in the format described above, with a leading (ignored) count and a trailing zero element. @{b}ASM_OUTPUT_DESTRUCTOR@{ub} is handled similarly. Since no init section is available, the absence of @{b}INIT_SECTION_ASM_OP@{ub} causes the compilation of @{b}main@{ub} to call @{b}__main@{ub} as above, starting the initialization process. The last variant uses neither arbitrary sections nor the GNU linker. This is preferable when you want to do dynamic linking and when using file formats which the GNU linker does not support, such as `ECOFF'. In this case, @{b}ASM_OUTPUT_CONSTRUCTOR@{ub} does not produce an @{b}N_SETT@{ub} symbol; initialization and termination functions are recognized simply by their names. This requires an extra program in the linkage step, called @{b}collect2@{ub}. This program pretends to be the linker, for use with GNU CC; it does its job by running the ordinary linker, but also arranges to include the vectors of initialization and termination functions. These functions are called via @{b}__main@{ub} as described above. Choosing among these configuration options has been simplified by a set of operating-system-dependent files in the @{b}config@{ub} subdirectory. These files define all of the relevant parameters. Usually it is sufficient to include one into your specific machine-dependent configuration file. These files are: @{b}aoutos.h@{ub} For operating systems using the `a.out' format. @{b}next.h@{ub} For operating systems using the `MachO' format. @{b}svr3.h@{ub} For System V Release 3 and similar systems using `COFF' format. @{b}svr4.h@{ub} For System V Release 4 and similar systems using `ELF' format. @{b}vms.h@{ub} For the VMS operating system. The following section describes the specific macros that control and customize the handling of initialization and termination functions. @EndNode @Node "Macros for Initialization" "gcc.guide/Macros for Initialization" @Next "Instruction Output" @Prev "Initialization" @Toc "Assembler Format" Macros Controlling Initialization Routines ------------------------------------------ Here are the macros that control how the compiler handles initialization and termination functions: @{b}INIT_SECTION_ASM_OP@{ub} If defined, a C string constant for the assembler operation to identify the following data as initialization code. If not defined, GNU CC will assume such a section does not exist. When you are using special sections for initialization and termination functions, this macro also controls how @{b}crtstuff.c@{ub} and @{b}libgcc2.c@{ub} arrange to run the initialization functions. @{b}HAS_INIT_SECTION@{ub} If defined, @{b}main@{ub} will not call @{b}__main@{ub} as described above. This macro should be defined for systems that control the contents of the init section on a symbol-by-symbol basis, such as OSF/1, and should not be defined explicitly for systems that support @{b}INIT_SECTION_ASM_OP@{ub}. @{b}LD_INIT_SWITCH@{ub} If defined, a C string constant for a switch that tells the linker that the following symbol is an initialization routine. @{b}LD_FINI_SWITCH@{ub} If defined, a C string constant for a switch that tells the linker that the following symbol is a finalization routine. @{b}INVOKE__main@{ub} If defined, @{b}main@{ub} will call @{b}__main@{ub} despite the presence of @{b}INIT_SECTION_ASM_OP@{ub}. This macro should be defined for systems where the init section is not actually run automatically, but is still useful for collecting the lists of constructors and destructors. @{b}ASM_OUTPUT_CONSTRUCTOR (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} Define this macro as a C statement to output on the stream @{I}STREAM@{ui} the assembler code to arrange to call the function named @{I}NAME@{ui} at initialization time. Assume that @{I}NAME@{ui} is the name of a C function generated automatically by the compiler. This function takes no arguments. Use the function @{b}assemble_name@{ub} to output the name @{I}NAME@{ui}; this performs any system-specific syntactic transformations such as adding an underscore. If you don't define this macro, nothing special is output to arrange to call the function. This is correct when the function will be called in some other manner--for example, by means of the @{b}collect2@{ub} program, which looks through the symbol table to find these functions by their names. @{b}ASM_OUTPUT_DESTRUCTOR (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} This is like @{b}ASM_OUTPUT_CONSTRUCTOR@{ub} but used for termination functions rather than initialization functions. If your system uses @{b}collect2@{ub} as the means of processing constructors, then that program normally uses @{b}nm@{ub} to scan an object file for constructor functions to be called. On certain kinds of systems, you can define these macros to make @{b}collect2@{ub} work faster (and, in some cases, make it work at all): @{b}OBJECT_FORMAT_COFF@{ub} Define this macro if the system uses COFF (Common Object File Format) object files, so that @{b}collect2@{ub} can assume this format and scan object files directly for dynamic constructor/destructor functions. @{b}OBJECT_FORMAT_ROSE@{ub} Define this macro if the system uses ROSE format object files, so that @{b}collect2@{ub} can assume this format and scan object files directly for dynamic constructor/destructor functions. These macros are effective only in a native compiler; @{b}collect2@{ub} as part of a cross compiler always uses @{b}nm@{ub} for the target machine. @{b}REAL_NM_FILE_NAME@{ub} Define this macro as a C string constant containing the file name to use to execute @{b}nm@{ub}. The default is to search the path normally for @{b}nm@{ub}. If your system supports shared libraries and has a program to list the dynamic dependencies of a given library or executable, you can define these macros to enable support for running initialization and termination functions in shared libraries: @{b}LDD_SUFFIX@{ub} Define this macro to a C string constant containing the name of the program which lists dynamic dependencies, like @{b}"ldd"@{ub} under SunOS 4. @{b}PARSE_LDD_OUTPUT (@{I}PTR@{ui})@{ub} Define this macro to be C code that extracts filenames from the output of the program denoted by @{b}LDD_SUFFIX@{ub}. @{I}PTR@{ui} is a variable of type @{b}char *@{ub} that points to the beginning of a line of output from @{b}LDD_SUFFIX@{ub}. If the line lists a dynamic dependency, the code must advance @{I}PTR@{ui} to the beginning of the filename on that line. Otherwise, it must set @{I}PTR@{ui} to @{b}NULL@{ub}. @EndNode @Node "Instruction Output" "gcc.guide/Instruction Output" @Next "Dispatch Tables" @Prev "Macros for Initialization" @Toc "Assembler Format" Output of Assembler Instructions -------------------------------- This describes assembler instruction output. @{b}REGISTER_NAMES@{ub} A C initializer containing the assembler's names for the machine registers, each one as a C string constant. This is what translates register numbers in the compiler into assembler language. @{b}ADDITIONAL_REGISTER_NAMES@{ub} If defined, a C initializer for an array of structures containing a name and a register number. This macro defines additional names for hard registers, thus allowing the @{b}asm@{ub} option in declarations to refer to registers using alternate names. @{b}ASM_OUTPUT_OPCODE (@{I}STREAM@{ui}, @{I}PTR@{ui})@{ub} Define this macro if you are using an unusual assembler that requires different names for the machine instructions. The definition is a C statement or statements which output an assembler instruction opcode to the stdio stream @{I}STREAM@{ui}. The macro-operand @{I}PTR@{ui} is a variable of type @{b}char *@{ub} which points to the opcode name in its "internal" form--the form that is written in the machine description. The definition should output the opcode name to @{I}STREAM@{ui}, performing any translation you desire, and increment the variable @{I}PTR@{ui} to point at the end of the opcode so that it will not be output twice. In fact, your macro definition may process less than the entire opcode name, or more than the opcode name; but if you want to process text that includes @{b}%@{ub}-sequences to substitute operands, you must take care of the substitution yourself. Just be sure to increment @{I}PTR@{ui} over whatever text should not be output normally. If you need to look at the operand values, they can be found as the elements of @{b}recog_operand@{ub}. If the macro definition does nothing, the instruction is output in the usual way. @{b}FINAL_PRESCAN_INSN (@{I}INSN@{ui}, @{I}OPVEC@{ui}, @{I}NOPERANDS@{ui})@{ub} If defined, a C statement to be executed just prior to the output of assembler code for @{I}INSN@{ui}, to modify the extracted operands so they will be output differently. Here the argument @{I}OPVEC@{ui} is the vector containing the operands extracted from @{I}INSN@{ui}, and @{I}NOPERANDS@{ui} is the number of elements of the vector which contain meaningful data for this insn. The contents of this vector are what will be used to convert the insn template into assembler code, so you can change the assembler output by changing the contents of the vector. This macro is useful when various assembler syntaxes share a single file of instruction patterns; by defining this macro differently, you can cause a large class of instructions to be output differently (such as with rearranged operands). Naturally, variations in assembler syntax affecting individual insn patterns ought to be handled by writing conditional output routines in those patterns. If this macro is not defined, it is equivalent to a null statement. @{b}PRINT_OPERAND (@{I}STREAM@{ui}, @{I}X@{ui}, @{I}CODE@{ui})@{ub} A C compound statement to output to stdio stream @{I}STREAM@{ui} the assembler syntax for an instruction operand @{I}X@{ui}. @{I}X@{ui} is an RTL expression. @{I}CODE@{ui} is a value that can be used to specify one of several ways of printing the operand. It is used when identical operands must be printed differently depending on the context. @{I}CODE@{ui} comes from the @{b}%@{ub} specification that was used to request printing of the operand. If the specification was just @{b} %@{I}DIGIT@{ui}@{ub} then @{I}CODE@{ui} is 0; if the specification was @{b} %@{I}LTR@{ui} @{I}DIGIT@{ui}@{ub} then @{I}CODE@{ui} is the ASCII code for @{I}LTR@{ui}. If @{I}X@{ui} is a register, this macro should print the register's name. The names can be found in an array @{b}reg_names@{ub} whose type is @{b}char *[]@{ub}. @{b}reg_names@{ub} is initialized from @{b}REGISTER_NAMES@{ub}. When the machine description has a specification @{b} %@{I}PUNCT@{ui}@{ub} (a @{b}%@{ub} followed by a punctuation character), this macro is called with a null pointer for @{I}X@{ui} and the punctuation character for @{I}CODE@{ui}. @{b}PRINT_OPERAND_PUNCT_VALID_P (@{I}CODE@{ui})@{ub} A C expression which evaluates to true if @{I}CODE@{ui} is a valid punctuation character for use in the @{b}PRINT_OPERAND@{ub} macro. If @{b}PRINT_OPERAND_PUNCT_VALID_P@{ub} is not defined, it means that no punctuation characters (except for the standard one, @{b}%@{ub}) are used in this way. @{b}PRINT_OPERAND_ADDRESS (@{I}STREAM@{ui}, @{I}X@{ui})@{ub} A C compound statement to output to stdio stream @{I}STREAM@{ui} the assembler syntax for an instruction operand that is a memory reference whose address is @{I}X@{ui}. @{I}X@{ui} is an RTL expression. On some machines, the syntax for a symbolic address depends on the section that the address refers to. On these machines, define the macro @{b}ENCODE_SECTION_INFO@{ub} to store the information into the @{b}symbol_ref@{ub}, and then check for it here. See @{"Assembler Format" Link "Assembler Format"}. @{b}DBR_OUTPUT_SEQEND(@{I}FILE@{ui})@{ub} A C statement, to be executed after all slot-filler instructions have been output. If necessary, call @{b}dbr_sequence_length@{ub} to determine the number of slots filled in a sequence (zero if not currently outputting a sequence), to decide how many no-ops to output, or whatever. Don't define this macro if it has nothing to do, but it is helpful in reading assembly output if the extent of the delay sequence is made explicit (e.g. with white space). Note that output routines for instructions with delay slots must be prepared to deal with not being output as part of a sequence (i.e. when the scheduling pass is not run, or when no slot fillers could be found.) The variable @{b}final_sequence@{ub} is null when not processing a sequence, otherwise it contains the @{b}sequence@{ub} rtx being output. @{b}REGISTER_PREFIX@{ub} @{b}LOCAL_LABEL_PREFIX@{ub} @{b}USER_LABEL_PREFIX@{ub} @{b}IMMEDIATE_PREFIX@{ub} If defined, C string expressions to be used for the @{b}%R@{ub}, @{b}%L@{ub}, @{b}%U@{ub}, and @{b}%I@{ub} options of @{b}asm_fprintf@{ub} (see @{b}final.c@{ub}). These are useful when a single @{b}md@{ub} file must support multiple assembler formats. In that case, the various @{b}tm.h@{ub} files can define these macros differently. @{b}ASSEMBLER_DIALECT@{ub} If your target supports multiple dialects of assembler language (such as different opcodes), define this macro as a C expression that gives the numeric index of the assembler language dialect to use, with zero as the first variant. If this macro is defined, you may use @{b}{option0|option1|option2...}@{ub} constructs in the output templates of patterns (see @{"Output Template" Link "Output Template"}) or in the first argument of @{b}asm_fprintf@{ub}. This construct outputs @{b}option0@{ub}, @{b}option1@{ub} or @{b}option2@{ub}, etc., if the value of @{b}ASSEMBLER_DIALECT@{ub} is zero, one or two, etc. Any special characters within these strings retain their usual meaning. If you do not define this macro, the characters @{b}{@{ub}, @{b}|@{ub} and @{b}}@{ub} do not have any special meaning when used in templates or operands to @{b}asm_fprintf@{ub}. Define the macros @{b}REGISTER_PREFIX@{ub}, @{b}LOCAL_LABEL_PREFIX@{ub}, @{b}USER_LABEL_PREFIX@{ub} and @{b}IMMEDIATE_PREFIX@{ub} if you can express the variations in assemble language syntax with that mechanism. Define @{b}ASSEMBLER_DIALECT@{ub} and use the @{b}{option0|option1}@{ub} syntax if the syntax variant are larger and involve such things as different opcodes or operand order. @{b}ASM_OUTPUT_REG_PUSH (@{I}STREAM@{ui}, @{I}REGNO@{ui})@{ub} A C expression to output to @{I}STREAM@{ui} some assembler code which will push hard register number @{I}REGNO@{ui} onto the stack. The code need not be optimal, since this macro is used only when profiling. @{b}ASM_OUTPUT_REG_POP (@{I}STREAM@{ui}, @{I}REGNO@{ui})@{ub} A C expression to output to @{I}STREAM@{ui} some assembler code which will pop hard register number @{I}REGNO@{ui} off of the stack. The code need not be optimal, since this macro is used only when profiling. @EndNode @Node "Dispatch Tables" "gcc.guide/Dispatch Tables" @Next "Alignment Output" @Prev "Instruction Output" @Toc "Assembler Format" Output of Dispatch Tables ------------------------- This concerns dispatch tables. @{b}ASM_OUTPUT_ADDR_DIFF_ELT (@{I}STREAM@{ui}, @{I}VALUE@{ui}, @{I}REL@{ui})@{ub} This macro should be provided on machines where the addresses in a dispatch table are relative to the table's own address. The definition should be a C statement to output to the stdio stream @{I}STREAM@{ui} an assembler pseudo-instruction to generate a difference between two labels. @{I}VALUE@{ui} and @{I}REL@{ui} are the numbers of two internal labels. The definitions of these labels are output using @{b}ASM_OUTPUT_INTERNAL_LABEL@{ub}, and they must be printed in the same way here. For example, fprintf (@{I}STREAM@{ui}, "\\t.word L%d-L%d\\n", @{I}VALUE@{ui}, @{I}REL@{ui}) @{b}ASM_OUTPUT_ADDR_VEC_ELT (@{I}STREAM@{ui}, @{I}VALUE@{ui})@{ub} This macro should be provided on machines where the addresses in a dispatch table are absolute. The definition should be a C statement to output to the stdio stream @{I}STREAM@{ui} an assembler pseudo-instruction to generate a reference to a label. @{I}VALUE@{ui} is the number of an internal label whose definition is output using @{b}ASM_OUTPUT_INTERNAL_LABEL@{ub}. For example, fprintf (@{I}STREAM@{ui}, "\\t.word L%d\\n", @{I}VALUE@{ui}) @{b}ASM_OUTPUT_CASE_LABEL (@{I}STREAM@{ui}, @{I}PREFIX@{ui}, @{I}NUM@{ui}, @{I}TABLE@{ui})@{ub} Define this if the label before a jump-table needs to be output specially. The first three arguments are the same as for @{b}ASM_OUTPUT_INTERNAL_LABEL@{ub}; the fourth argument is the jump-table which follows (a @{b}jump_insn@{ub} containing an @{b}addr_vec@{ub} or @{b}addr_diff_vec@{ub}). This feature is used on system V to output a @{b}swbeg@{ub} statement for the table. If this macro is not defined, these labels are output with @{b}ASM_OUTPUT_INTERNAL_LABEL@{ub}. @{b}ASM_OUTPUT_CASE_END (@{I}STREAM@{ui}, @{I}NUM@{ui}, @{I}TABLE@{ui})@{ub} Define this if something special must be output at the end of a jump-table. The definition should be a C statement to be executed after the assembler code for the table is written. It should write the appropriate code to stdio stream @{I}STREAM@{ui}. The argument @{I}TABLE@{ui} is the jump-table insn, and @{I}NUM@{ui} is the label-number of the preceding label. If this macro is not defined, nothing special is output at the end of the jump-table. @EndNode @Node "Alignment Output" "gcc.guide/Alignment Output" @Prev "Dispatch Tables" @Toc "Assembler Format" Assembler Commands for Alignment -------------------------------- This describes commands for alignment. @{b}ASM_OUTPUT_ALIGN_CODE (@{I}FILE@{ui})@{ub} A C expression to output text to align the location counter in the way that is desirable at a point in the code that is reached only by jumping. This macro need not be defined if you don't want any special alignment to be done at such a time. Most machine descriptions do not currently define the macro. @{b}ASM_OUTPUT_LOOP_ALIGN (@{I}FILE@{ui})@{ub} A C expression to output text to align the location counter in the way that is desirable at the beginning of a loop. This macro need not be defined if you don't want any special alignment to be done at such a time. Most machine descriptions do not currently define the macro. @{b}ASM_OUTPUT_SKIP (@{I}STREAM@{ui}, @{I}NBYTES@{ui})@{ub} A C statement to output to the stdio stream @{I}STREAM@{ui} an assembler instruction to advance the location counter by @{I}NBYTES@{ui} bytes. Those bytes should be zero when loaded. @{I}NBYTES@{ui} will be a C expression of type @{b}int@{ub}. @{b}ASM_NO_SKIP_IN_TEXT@{ub} Define this macro if @{b}ASM_OUTPUT_SKIP@{ub} should not be used in the text section because it fails put zeros in the bytes that are skipped. This is true on many Unix systems, where the pseudo-op to skip bytes produces no-op instructions rather than zeros when used in the text section. @{b}ASM_OUTPUT_ALIGN (@{I}STREAM@{ui}, @{I}POWER@{ui})@{ub} A C statement to output to the stdio stream @{I}STREAM@{ui} an assembler command to advance the location counter to a multiple of 2 to the @{I}POWER@{ui} bytes. @{I}POWER@{ui} will be a C expression of type @{b}int@{ub}. @EndNode @Node "Debugging Info" "gcc.guide/Debugging Info" @Next "Cross-compilation" @Prev "Assembler Format" @Toc "Target Macros" Controlling Debugging Information Format ======================================== This describes how to specify debugging information. @{" All Debuggers " Link "All Debuggers"} Macros that affect all debugging formats uniformly. @{" DBX Options " Link "DBX Options"} Macros enabling specific options in DBX format. @{" DBX Hooks " Link "DBX Hooks"} Hook macros for varying DBX format. @{" File Names and DBX " Link "File Names and DBX"} Macros controlling output of file names in DBX format. @{" SDB and DWARF " Link "SDB and DWARF"} Macros for SDB (COFF) and DWARF formats. @EndNode @Node "All Debuggers" "gcc.guide/All Debuggers" @Next "DBX Options" @Toc "Debugging Info" Macros Affecting All Debugging Formats -------------------------------------- These macros affect all debugging formats. @{b}DBX_REGISTER_NUMBER (@{I}REGNO@{ui})@{ub} A C expression that returns the DBX register number for the compiler register number @{I}REGNO@{ui}. In simple cases, the value of this expression may be @{I}REGNO@{ui} itself. But sometimes there are some registers that the compiler knows about and DBX does not, or vice versa. In such cases, some register may need to have one number in the compiler and another for DBX. If two registers have consecutive numbers inside GNU CC, and they can be used as a pair to hold a multiword value, then they @{i}must@{ui} have consecutive numbers after renumbering with @{b}DBX_REGISTER_NUMBER@{ub}. Otherwise, debuggers will be unable to access such a pair, because they expect register pairs to be consecutive in their own numbering scheme. If you find yourself defining @{b}DBX_REGISTER_NUMBER@{ub} in way that does not preserve register pairs, then what you must do instead is redefine the actual register numbering scheme. @{b}DEBUGGER_AUTO_OFFSET (@{I}X@{ui})@{ub} A C expression that returns the integer offset value for an automatic variable having address @{I}X@{ui} (an RTL expression). The default computation assumes that @{I}X@{ui} is based on the frame-pointer and gives the offset from the frame-pointer. This is required for targets that produce debugging output for DBX or COFF-style debugging output for SDB and allow the frame-pointer to be eliminated when the @{b}-g@{ub} options is used. @{b}DEBUGGER_ARG_OFFSET (@{I}OFFSET@{ui}, @{I}X@{ui})@{ub} A C expression that returns the integer offset value for an argument having address @{I}X@{ui} (an RTL expression). The nominal offset is @{I}OFFSET@{ui}. @{b}PREFERRED_DEBUGGING_TYPE@{ub} A C expression that returns the type of debugging output GNU CC produces when the user specifies @{b}-g@{ub} or @{b}-ggdb@{ub}. Define this if you have arranged for GNU CC to support more than one format of debugging output. Currently, the allowable values are @{b}DBX_DEBUG@{ub}, @{b}SDB_DEBUG@{ub}, @{b}DWARF_DEBUG@{ub}, and @{b}XCOFF_DEBUG@{ub}. The value of this macro only affects the default debugging output; the user can always get a specific type of output by using @{b}-gstabs@{ub}, @{b}-gcoff@{ub}, @{b}-gdwarf@{ub}, or @{b}-gxcoff@{ub}. @EndNode @Node "DBX Options" "gcc.guide/DBX Options" @Next "DBX Hooks" @Prev "All Debuggers" @Toc "Debugging Info" Specific Options for DBX Output ------------------------------- These are specific options for DBX output. @{b}DBX_DEBUGGING_INFO@{ub} Define this macro if GNU CC should produce debugging output for DBX in response to the @{b}-g@{ub} option. @{b}XCOFF_DEBUGGING_INFO@{ub} Define this macro if GNU CC should produce XCOFF format debugging output in response to the @{b}-g@{ub} option. This is a variant of DBX format. @{b}DEFAULT_GDB_EXTENSIONS@{ub} Define this macro to control whether GNU CC should by default generate GDB's extended version of DBX debugging information (assuming DBX-format debugging information is enabled at all). If you don't define the macro, the default is 1: always generate the extended information if there is any occasion to. @{b}DEBUG_SYMS_TEXT@{ub} Define this macro if all @{b}.stabs@{ub} commands should be output while in the text section. @{b}ASM_STABS_OP@{ub} A C string constant naming the assembler pseudo op to use instead of @{b}.stabs@{ub} to define an ordinary debugging symbol. If you don't define this macro, @{b}.stabs@{ub} is used. This macro applies only to DBX debugging information format. @{b}ASM_STABD_OP@{ub} A C string constant naming the assembler pseudo op to use instead of @{b}.stabd@{ub} to define a debugging symbol whose value is the current location. If you don't define this macro, @{b}.stabd@{ub} is used. This macro applies only to DBX debugging information format. @{b}ASM_STABN_OP@{ub} A C string constant naming the assembler pseudo op to use instead of @{b}.stabn@{ub} to define a debugging symbol with no name. If you don't define this macro, @{b}.stabn@{ub} is used. This macro applies only to DBX debugging information format. @{b}DBX_NO_XREFS@{ub} Define this macro if DBX on your system does not support the construct @{b}xs@{I}TAGNAME@{ui}@{ub}. On some systems, this construct is used to describe a forward reference to a structure named @{I}TAGNAME@{ui}. On other systems, this construct is not supported at all. @{b}DBX_CONTIN_LENGTH@{ub} A symbol name in DBX-format debugging information is normally continued (split into two separate @{b}.stabs@{ub} directives) when it exceeds a certain length (by default, 80 characters). On some operating systems, DBX requires this splitting; on others, splitting must not be done. You can inhibit splitting by defining this macro with the value zero. You can override the default splitting-length by defining this macro as an expression for the length you desire. @{b}DBX_CONTIN_CHAR@{ub} Normally continuation is indicated by adding a @{b}\\ @{ub} character to the end of a @{b}.stabs@{ub} string when a continuation follows. To use a different character instead, define this macro as a character constant for the character you want to use. Do not define this macro if backslash is correct for your system. @{b}DBX_STATIC_STAB_DATA_SECTION@{ub} Define this macro if it is necessary to go to the data section before outputting the @{b}.stabs@{ub} pseudo-op for a non-global static variable. @{b}DBX_TYPE_DECL_STABS_CODE@{ub} The value to use in the "code" field of the @{b}.stabs@{ub} directive for a typedef. The default is @{b}N_LSYM@{ub}. @{b}DBX_STATIC_CONST_VAR_CODE@{ub} The value to use in the "code" field of the @{b}.stabs@{ub} directive for a static variable located in the text section. DBX format does not provide any "right" way to do this. The default is @{b}N_FUN@{ub}. @{b}DBX_REGPARM_STABS_CODE@{ub} The value to use in the "code" field of the @{b}.stabs@{ub} directive for a parameter passed in registers. DBX format does not provide any "right" way to do this. The default is @{b}N_RSYM@{ub}. @{b}DBX_REGPARM_STABS_LETTER@{ub} The letter to use in DBX symbol data to identify a symbol as a parameter passed in registers. DBX format does not customarily provide any way to do this. The default is @{b}'P'@{ub}. @{b}DBX_MEMPARM_STABS_LETTER@{ub} The letter to use in DBX symbol data to identify a symbol as a stack parameter. The default is @{b}'p'@{ub}. @{b}DBX_FUNCTION_FIRST@{ub} Define this macro if the DBX information for a function and its arguments should precede the assembler code for the function. Normally, in DBX format, the debugging information entirely follows the assembler code. @{b}DBX_LBRAC_FIRST@{ub} Define this macro if the @{b}N_LBRAC@{ub} symbol for a block should precede the debugging information for variables and functions defined in that block. Normally, in DBX format, the @{b}N_LBRAC@{ub} symbol comes first. @{b}DBX_BLOCKS_FUNCTION_RELATIVE@{ub} Define this macro if the value of a symbol describing the scope of a block (@{b}N_LBRAC@{ub} or @{b}N_RBRAC@{ub}) should be relative to the start of the enclosing function. Normally, GNU C uses an absolute address. @EndNode @Node "DBX Hooks" "gcc.guide/DBX Hooks" @Next "File Names and DBX" @Prev "DBX Options" @Toc "Debugging Info" Open-Ended Hooks for DBX Format ------------------------------- These are hooks for DBX format. @{b}DBX_OUTPUT_LBRAC (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} Define this macro to say how to output to @{I}STREAM@{ui} the debugging information for the start of a scope level for variable names. The argument @{I}NAME@{ui} is the name of an assembler symbol (for use with @{b}assemble_name@{ub}) whose value is the address where the scope begins. @{b}DBX_OUTPUT_RBRAC (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} Like @{b}DBX_OUTPUT_LBRAC@{ub}, but for the end of a scope level. @{b}DBX_OUTPUT_ENUM (@{I}STREAM@{ui}, @{I}TYPE@{ui})@{ub} Define this macro if the target machine requires special handling to output an enumeration type. The definition should be a C statement (sans semicolon) to output the appropriate information to @{I}STREAM@{ui} for the type @{I}TYPE@{ui}. @{b}DBX_OUTPUT_FUNCTION_END (@{I}STREAM@{ui}, @{I}FUNCTION@{ui})@{ub} Define this macro if the target machine requires special output at the end of the debugging information for a function. The definition should be a C statement (sans semicolon) to output the appropriate information to @{I}STREAM@{ui}. @{I}FUNCTION@{ui} is the @{b}FUNCTION_DECL@{ub} node for the function. @{b}DBX_OUTPUT_STANDARD_TYPES (@{I}SYMS@{ui})@{ub} Define this macro if you need to control the order of output of the standard data types at the beginning of compilation. The argument @{I}SYMS@{ui} is a @{b}tree@{ub} which is a chain of all the predefined global symbols, including names of data types. Normally, DBX output starts with definitions of the types for integers and characters, followed by all the other predefined types of the particular language in no particular order. On some machines, it is necessary to output different particular types first. To do this, define @{b}DBX_OUTPUT_STANDARD_TYPES@{ub} to output those symbols in the necessary order. Any predefined types that you don't explicitly output will be output afterward in no particular order. Be careful not to define this macro so that it works only for C. There are no global variables to access most of the built-in types, because another language may have another set of types. The way to output a particular type is to look through @{I}SYMS@{ui} to see if you can find it. Here is an example: { tree decl; for (decl = syms; decl; decl = TREE_CHAIN (decl)) if (!strcmp (IDENTIFIER_POINTER (DECL_NAME (decl)), "long int")) dbxout_symbol (decl); ... } This does nothing if the expected type does not exist. See the function @{b}init_decl_processing@{ub} in @{b}c-decl.c@{ub} to find the names to use for all the built-in C types. Here is another way of finding a particular type: { tree decl; for (decl = syms; decl; decl = TREE_CHAIN (decl)) if (TREE_CODE (decl) == TYPE_DECL && (TREE_CODE (TREE_TYPE (decl)) == INTEGER_CST) && TYPE_PRECISION (TREE_TYPE (decl)) == 16 && TYPE_UNSIGNED (TREE_TYPE (decl))) /* @{i}This must be @{b}unsigned short@{ub}.@{ui} */ dbxout_symbol (decl); ... } @EndNode @Node "File Names and DBX" "gcc.guide/File Names and DBX" @Next "SDB and DWARF" @Prev "DBX Hooks" @Toc "Debugging Info" File Names in DBX Format ------------------------ This describes file names in DBX format. @{b}DBX_WORKING_DIRECTORY@{ub} Define this if DBX wants to have the current directory recorded in each object file. Note that the working directory is always recorded if GDB extensions are enabled. @{b}DBX_OUTPUT_MAIN_SOURCE_FILENAME (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} A C statement to output DBX debugging information to the stdio stream @{I}STREAM@{ui} which indicates that file @{I}NAME@{ui} is the main source file--the file specified as the input file for compilation. This macro is called only once, at the beginning of compilation. This macro need not be defined if the standard form of output for DBX debugging information is appropriate. @{b}DBX_OUTPUT_MAIN_SOURCE_DIRECTORY (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} A C statement to output DBX debugging information to the stdio stream @{I}STREAM@{ui} which indicates that the current directory during compilation is named @{I}NAME@{ui}. This macro need not be defined if the standard form of output for DBX debugging information is appropriate. @{b}DBX_OUTPUT_MAIN_SOURCE_FILE_END (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} A C statement to output DBX debugging information at the end of compilation of the main source file @{I}NAME@{ui}. If you don't define this macro, nothing special is output at the end of compilation, which is correct for most machines. @{b}DBX_OUTPUT_SOURCE_FILENAME (@{I}STREAM@{ui}, @{I}NAME@{ui})@{ub} A C statement to output DBX debugging information to the stdio stream @{I}STREAM@{ui} which indicates that file @{I}NAME@{ui} is the current source file. This output is generated each time input shifts to a different source file as a result of @{b}#include@{ub}, the end of an included file, or a @{b}#line@{ub} command. This macro need not be defined if the standard form of output for DBX debugging information is appropriate. @EndNode @Node "SDB and DWARF" "gcc.guide/SDB and DWARF" @Prev "File Names and DBX" @Toc "Debugging Info" Macros for SDB and DWARF Output ------------------------------- Here are macros for SDB and DWARF output. @{b}SDB_DEBUGGING_INFO@{ub} Define this macro if GNU CC should produce COFF-style debugging output for SDB in response to the @{b}-g@{ub} option. @{b}DWARF_DEBUGGING_INFO@{ub} Define this macro if GNU CC should produce dwarf format debugging output in response to the @{b}-g@{ub} option. @{b}PUT_SDB_...@{ub} Define these macros to override the assembler syntax for the special SDB assembler directives. See @{b}sdbout.c@{ub} for a list of these macros and their arguments. If the standard syntax is used, you need not define them yourself. @{b}SDB_DELIM@{ub} Some assemblers do not support a semicolon as a delimiter, even between SDB assembler directives. In that case, define this macro to be the delimiter to use (usually @{b}\\n@{ub}). It is not necessary to define a new set of @{b}PUT_SDB_@{I}OP@{ui}@{ub} macros if this is the only change required. @{b}SDB_GENERATE_FAKE@{ub} Define this macro to override the usual method of constructing a dummy name for anonymous structure and union types. See @{b}sdbout.c@{ub} for more information. @{b}SDB_ALLOW_UNKNOWN_REFERENCES@{ub} Define this macro to allow references to unknown structure, union, or enumeration tags to be emitted. Standard COFF does not allow handling of unknown references, MIPS ECOFF has support for it. @{b}SDB_ALLOW_FORWARD_REFERENCES@{ub} Define this macro to allow references to structure, union, or enumeration tags that have not yet been seen to be handled. Some assemblers choke if forward tags are used, while some require it. @EndNode @Node "Cross-compilation" "gcc.guide/Cross-compilation" @Next "Misc" @Prev "Debugging Info" @Toc "Target Macros" Cross Compilation and Floating Point ==================================== While all modern machines use 2's complement representation for integers, there are a variety of representations for floating point numbers. This means that in a cross-compiler the representation of floating point numbers in the compiled program may be different from that used in the machine doing the compilation. Because different representation systems may offer different amounts of range and precision, the cross compiler cannot safely use the host machine's floating point arithmetic. Therefore, floating point constants must be represented in the target machine's format. This means that the cross compiler cannot use @{b}atof@{ub} to parse a floating point constant; it must have its own special routine to use instead. Also, constant folding must emulate the target machine's arithmetic (or must not be done at all). The macros in the following table should be defined only if you are cross compiling between different floating point formats. Otherwise, don't define them. Then default definitions will be set up which use @{b}double@{ub} as the data type, @{b}==@{ub} to test for equality, etc. You don't need to worry about how many times you use an operand of any of these macros. The compiler never uses operands which have side effects. @{b}REAL_VALUE_TYPE@{ub} A macro for the C data type to be used to hold a floating point value in the target machine's format. Typically this would be a @{b}struct@{ub} containing an array of @{b}int@{ub}. @{b}REAL_VALUES_EQUAL (@{I}X@{ui}, @{I}Y@{ui})@{ub} A macro for a C expression which compares for equality the two values, @{I}X@{ui} and @{I}Y@{ui}, both of type @{b}REAL_VALUE_TYPE@{ub}. @{b}REAL_VALUES_LESS (@{I}X@{ui}, @{I}Y@{ui})@{ub} A macro for a C expression which tests whether @{I}X@{ui} is less than @{I}Y@{ui}, both values being of type @{b}REAL_VALUE_TYPE@{ub} and interpreted as floating point numbers in the target machine's representation. @{b}REAL_VALUE_LDEXP (@{I}X@{ui}, @{I}SCALE@{ui})@{ub} A macro for a C expression which performs the standard library function @{b}ldexp@{ub}, but using the target machine's floating point representation. Both @{I}X@{ui} and the value of the expression have type @{b}REAL_VALUE_TYPE@{ub}. The second argument, @{I}SCALE@{ui}, is an integer. @{b}REAL_VALUE_FIX (@{I}X@{ui})@{ub} A macro whose definition is a C expression to convert the target-machine floating point value @{I}X@{ui} to a signed integer. @{I}X@{ui} has type @{b}REAL_VALUE_TYPE@{ub}. @{b}REAL_VALUE_UNSIGNED_FIX (@{I}X@{ui})@{ub} A macro whose definition is a C expression to convert the target-machine floating point value @{I}X@{ui} to an unsigned integer. @{I}X@{ui} has type @{b}REAL_VALUE_TYPE@{ub}. @{b}REAL_VALUE_RNDZINT (@{I}X@{ui})@{ub} A macro whose definition is a C expression to round the target-machine floating point value @{I}X@{ui} towards zero to an integer value (but still as a floating point number). @{I}X@{ui} has type @{b}REAL_VALUE_TYPE@{ub}, and so does the value. @{b}REAL_VALUE_UNSIGNED_RNDZINT (@{I}X@{ui})@{ub} A macro whose definition is a C expression to round the target-machine floating point value @{I}X@{ui} towards zero to an unsigned integer value (but still represented as a floating point number). @{I}X@{ui} has type @{b}REAL_VALUE_TYPE@{ub}, and so does the value. @{b}REAL_VALUE_ATOF (@{I}STRING@{ui}, @{I}MODE@{ui})@{ub} A macro for a C expression which converts @{I}STRING@{ui}, an expression of type @{b}char *@{ub}, into a floating point number in the target machine's representation for mode @{I}MODE@{ui}. The value has type @{b}REAL_VALUE_TYPE@{ub}. @{b}REAL_INFINITY@{ub} Define this macro if infinity is a possible floating point value, and therefore division by 0 is legitimate. @{b}REAL_VALUE_ISINF (@{I}X@{ui})@{ub} A macro for a C expression which determines whether @{I}X@{ui}, a floating point value, is infinity. The value has type @{b}int@{ub}. By default, this is defined to call @{b}isinf@{ub}. @{b}REAL_VALUE_ISNAN (@{I}X@{ui})@{ub} A macro for a C expression which determines whether @{I}X@{ui}, a floating point value, is a "nan" (not-a-number). The value has type @{b}int@{ub}. By default, this is defined to call @{b}isnan@{ub}. Define the following additional macros if you want to make floating point constant folding work while cross compiling. If you don't define them, cross compilation is still possible, but constant folding will not happen for floating point values. @{b}REAL_ARITHMETIC (@{I}OUTPUT@{ui}, @{I}CODE@{ui}, @{I}X@{ui}, @{I}Y@{ui})@{ub} A macro for a C statement which calculates an arithmetic operation of the two floating point values @{I}X@{ui} and @{I}Y@{ui}, both of type @{b}REAL_VALUE_TYPE@{ub} in the target machine's representation, to produce a result of the same type and representation which is stored in @{I}OUTPUT@{ui} (which will be a variable). The operation to be performed is specified by @{I}CODE@{ui}, a tree code which will always be one of the following: @{b}PLUS_EXPR@{ub}, @{b}MINUS_EXPR@{ub}, @{b}MULT_EXPR@{ub}, @{b}RDIV_EXPR@{ub}, @{b}MAX_EXPR@{ub}, @{b}MIN_EXPR@{ub}. The expansion of this macro is responsible for checking for overflow. If overflow happens, the macro expansion should execute the statement @{b}return 0;@{ub}, which indicates the inability to perform the arithmetic operation requested. @{b}REAL_VALUE_NEGATE (@{I}X@{ui})@{ub} A macro for a C expression which returns the negative of the floating point value @{I}X@{ui}. Both @{I}X@{ui} and the value of the expression have type @{b}REAL_VALUE_TYPE@{ub} and are in the target machine's floating point representation. There is no way for this macro to report overflow, since overflow can't happen in the negation operation. @{b}REAL_VALUE_TRUNCATE (@{I}MODE@{ui}, @{I}X@{ui})@{ub} A macro for a C expression which converts the floating point value @{I}X@{ui} to mode @{I}MODE@{ui}. Both @{I}X@{ui} and the value of the expression are in the target machine's floating point representation and have type @{b}REAL_VALUE_TYPE@{ub}. However, the value should have an appropriate bit pattern to be output properly as a floating constant whose precision accords with mode @{I}MODE@{ui}. There is no way for this macro to report overflow. @{b}REAL_VALUE_TO_INT (@{I}LOW@{ui}, @{I}HIGH@{ui}, @{I}X@{ui})@{ub} A macro for a C expression which converts a floating point value @{I}X@{ui} into a double-precision integer which is then stored into @{I}LOW@{ui} and @{I}HIGH@{ui}, two variables of type @{I}INT@{ui}. @{b}REAL_VALUE_FROM_INT (@{I}X@{ui}, @{I}LOW@{ui}, @{I}HIGH@{ui})@{ub} A macro for a C expression which converts a double-precision integer found in @{I}LOW@{ui} and @{I}HIGH@{ui}, two variables of type @{I}INT@{ui}, into a floating point value which is then stored into @{I}X@{ui}. @EndNode @Node "Misc" "gcc.guide/Misc" @Prev "Cross-compilation" @Toc "Target Macros" Miscellaneous Parameters ======================== Here are several miscellaneous parameters. @{b}PREDICATE_CODES@{ub} Define this if you have defined special-purpose predicates in the file @{b}@{I}MACHINE@{ui}.c@{ub}. This macro is called within an initializer of an array of structures. The first field in the structure is the name of a predicate and the second field is an array of rtl codes. For each predicate, list all rtl codes that can be in expressions matched by the predicate. The list should have a trailing comma. Here is an example of two entries in the list for a typical RISC machine: #define PREDICATE_CODES \\ {"gen_reg_rtx_operand", {SUBREG, REG}}, \\ {"reg_or_short_cint_operand", {SUBREG, REG, CONST_INT}}, Defining this macro does not affect the generated code (however, incorrect definitions that omit an rtl code that may be matched by the predicate can cause the compiler to malfunction). Instead, it allows the table built by @{b}genrecog@{ub} to be more compact and efficient, thus speeding up the compiler. The most important predicates to include in the list specified by this macro are thoses used in the most insn patterns. @{b}CASE_VECTOR_MODE@{ub} An alias for a machine mode name. This is the machine mode that elements of a jump-table should have. @{b}CASE_VECTOR_PC_RELATIVE@{ub} Define this macro if jump-tables should contain relative addresses. @{b}CASE_DROPS_THROUGH@{ub} Define this if control falls through a @{b}case@{ub} insn when the index value is out of range. This means the specified default-label is actually ignored by the @{b}case@{ub} insn proper. @{b}CASE_VALUES_THRESHOLD@{ub} Define this to be the smallest number of different values for which it is best to use a jump-table instead of a tree of conditional branches. The default is four for machines with a @{b}casesi@{ub} instruction and five otherwise. This is best for most machines. @{b}WORD_REGISTER_OPERATIONS@{ub} Define this macro if operations between registers with integral mode smaller than a word are always performed on the entire register. Most RISC machines have this property and most CISC machines do not. @{b}LOAD_EXTEND_OP (@{I}MODE@{ui})@{ub} Define this macro to be a C expression indicating when insns that read memory in @{I}MODE@{ui}, an integral mode narrower than a word, set the bits outside of @{I}MODE@{ui} to be either the sign-extension or the zero-extension of the data read. Return @{b}SIGN_EXTEND@{ub} for values of @{I}MODE@{ui} for which the insn sign-extends, @{b}ZERO_EXTEND@{ub} for which it zero-extends, and @{b}NIL@{ub} for other modes. This macro is not called with @{I}MODE@{ui} non-integral or with a width greater than or equal to @{b}BITS_PER_WORD@{ub}, so you may return any value in this case. Do not define this macro if it would always return @{b}NIL@{ub}. On machines where this macro is defined, you will normally define it as the constant @{b}SIGN_EXTEND@{ub} or @{b}ZERO_EXTEND@{ub}. @{b}IMPLICIT_FIX_EXPR@{ub} An alias for a tree code that should be used by default for conversion of floating point values to fixed point. Normally, @{b}FIX_ROUND_EXPR@{ub} is used. @{b}FIXUNS_TRUNC_LIKE_FIX_TRUNC@{ub} Define this macro if the same instructions that convert a floating point number to a signed fixed point number also convert validly to an unsigned one. @{b}EASY_DIV_EXPR@{ub} An alias for a tree code that is the easiest kind of division to compile code for in the general case. It may be @{b}TRUNC_DIV_EXPR@{ub}, @{b}FLOOR_DIV_EXPR@{ub}, @{b}CEIL_DIV_EXPR@{ub} or @{b}ROUND_DIV_EXPR@{ub}. These four division operators differ in how they round the result to an integer. @{b}EASY_DIV_EXPR@{ub} is used when it is permissible to use any of those kinds of division and the choice should be made on the basis of efficiency. @{b}MOVE_MAX@{ub} The maximum number of bytes that a single instruction can move quickly from memory to memory. @{b}MAX_MOVE_MAX@{ub} The maximum number of bytes that a single instruction can move quickly from memory to memory. If this is undefined, the default is @{b}MOVE_MAX@{ub}. Otherwise, it is the constant value that is the largest value that @{b}MOVE_MAX@{ub} can have at run-time. @{b}SHIFT_COUNT_TRUNCATED@{ub} A C expression that is nonzero if on this machine the number of bits actually used for the count of a shift operation is equal to the number of bits needed to represent the size of the object being shifted. When this macro is non-zero, the compiler will assume that it is safe to omit a sign-extend, zero-extend, and certain bitwise `and' instructions that truncates the count of a shift operation. On machines that have instructions that act on bitfields at variable positions, which may include `bit test' instructions, a nonzero @{b}SHIFT_COUNT_TRUNCATED@{ub} also enables deletion of truncations of the values that serve as arguments to bitfield instructions. If both types of instructions truncate the count (for shifts) and position (for bitfield operations), or if no variable-position bitfield instructions exist, you should define this macro. However, on some machines, such as the 80386 and the 680x0, truncation only applies to shift operations and not the (real or pretended) bitfield operations. Define @{b}SHIFT_COUNT_TRUNCATED@{ub} to be zero on such machines. Instead, add patterns to the @{b}md@{ub} file that include the implied truncation of the shift instructions. You need not define this macro if it would always have the value of zero. @{b}TRULY_NOOP_TRUNCATION (@{I}OUTPREC@{ui}, @{I}INPREC@{ui})@{ub} A C expression which is nonzero if on this machine it is safe to "convert" an integer of @{I}INPREC@{ui} bits to one of @{I}OUTPREC@{ui} bits (where @{I}OUTPREC@{ui} is smaller than @{I}INPREC@{ui}) by merely operating on it as if it had only @{I}OUTPREC@{ui} bits. On many machines, this expression can be 1. When @{b}TRULY_NOOP_TRUNCATION@{ub} returns 1 for a pair of sizes for modes for which @{b}MODES_TIEABLE_P@{ub} is 0, suboptimal code can result. If this is the case, making @{b}TRULY_NOOP_TRUNCATION@{ub} return 0 in such cases may improve things. @{b}STORE_FLAG_VALUE@{ub} A C expression describing the value returned by a comparison operator with an integral mode and stored by a store-flag instruction (@{b} s@{I}COND@{ui}@{ub}) when the condition is true. This description must apply to @{i}all@{ui} the @{b} s@{I}COND@{ui}@{ub} patterns and all the comparison operators whose results have a @{b}MODE_INT@{ub} mode. A value of 1 or -1 means that the instruction implementing the comparison operator returns exactly 1 or -1 when the comparison is true and 0 when the comparison is false. Otherwise, the value indicates which bits of the result are guaranteed to be 1 when the comparison is true. This value is interpreted in the mode of the comparison operation, which is given by the mode of the first operand in the @{b} s@{I}COND@{ui}@{ub} pattern. Either the low bit or the sign bit of @{b}STORE_FLAG_VALUE@{ub} be on. Presently, only those bits are used by the compiler. If @{b}STORE_FLAG_VALUE@{ub} is neither 1 or -1, the compiler will generate code that depends only on the specified bits. It can also replace comparison operators with equivalent operations if they cause the required bits to be set, even if the remaining bits are undefined. For example, on a machine whose comparison operators return an @{b}SImode@{ub} value and where @{b}STORE_FLAG_VALUE@{ub} is defined as @{b}0x80000000@{ub}, saying that just the sign bit is relevant, the expression (ne:SI (and:SI @{I}X@{ui} (const_int @{I}POWER-OF-2@{ui})) (const_int 0)) can be converted to (ashift:SI @{I}X@{ui} (const_int @{I}N@{ui})) where @{I}N@{ui} is the appropriate shift count to move the bit being tested into the sign bit. There is no way to describe a machine that always sets the low-order bit for a true value, but does not guarantee the value of any other bits, but we do not know of any machine that has such an instruction. If you are trying to port GNU CC to such a machine, include an instruction to perform a logical-and of the result with 1 in the pattern for the comparison operators and let us know (see @{"How to Report Bugs" Link "Bug Reporting"}). Often, a machine will have multiple instructions that obtain a value from a comparison (or the condition codes). Here are rules to guide the choice of value for @{b}STORE_FLAG_VALUE@{ub}, and hence the instructions to be used: @{b}*@{ub} Use the shortest sequence that yields a valid definition for @{b}STORE_FLAG_VALUE@{ub}. It is more efficient for the compiler to "normalize" the value (convert it to, e.g., 1 or 0) than for the comparison operators to do so because there may be opportunities to combine the normalization with other operations. @{b}*@{ub} For equal-length sequences, use a value of 1 or -1, with -1 being slightly preferred on machines with expensive jumps and 1 preferred on other machines. @{b}*@{ub} As a second choice, choose a value of @{b}0x80000001@{ub} if instructions exist that set both the sign and low-order bits but do not define the others. @{b}*@{ub} Otherwise, use a value of @{b}0x80000000@{ub}. Many machines can produce both the value chosen for @{b}STORE_FLAG_VALUE@{ub} and its negation in the same number of instructions. On those machines, you should also define a pattern for those cases, e.g., one matching (set @{I}A@{ui} (neg:@{I}M@{ui} (ne:@{I}M@{ui} @{I}B@{ui} @{I}C@{ui}))) Some machines can also perform @{b}and@{ub} or @{b}plus@{ub} operations on condition code values with less instructions than the corresponding @{b} s@{I}COND@{ui}@{ub} insn followed by @{b}and@{ub} or @{b}plus@{ub}. On those machines, define the appropriate patterns. Use the names @{b}incscc@{ub} and @{b}decscc@{ub}, respectively, for the the patterns which perform @{b}plus@{ub} or @{b}minus@{ub} operations on condition code values. See @{b}rs6000.md@{ub} for some examples. The GNU Superoptizer can be used to find such instruction sequences on other machines. You need not define @{b}STORE_FLAG_VALUE@{ub} if the machine has no store-flag instructions. @{b}FLOAT_STORE_FLAG_VALUE@{ub} A C expression that gives a non-zero floating point value that is returned when comparison operators with floating-point results are true. Define this macro on machine that have comparison operations that return floating-point values. If there are no such operations, do not define this macro. @{b}Pmode@{ub} An alias for the machine mode for pointers. On most machines, define this to be the integer mode corresponding to the width of a hardware pointer; @{b}SImode@{ub} on 32-bit machine or @{b}DImode@{ub} on 64-bit machines. On some machines you must define this to be one of the partial integer modes, such as @{b}PSImode@{ub}. The width of @{b}Pmode@{ub} must be at least as large as the value of @{b}POINTER_SIZE@{ub}. If it is not equal, you must define the macro @{b}POINTERS_EXTEND_UNSIGNED@{ub} to specify how pointers are extended to @{b}Pmode@{ub}. @{b}FUNCTION_MODE@{ub} An alias for the machine mode used for memory references to functions being called, in @{b}call@{ub} RTL expressions. On most machines this should be @{b}QImode@{ub}. @{b}INTEGRATE_THRESHOLD (@{I}DECL@{ui})@{ub} A C expression for the maximum number of instructions above which the function @{I}DECL@{ui} should not be inlined. @{I}DECL@{ui} is a @{b}FUNCTION_DECL@{ub} node. The default definition of this macro is 64 plus 8 times the number of arguments that the function accepts. Some people think a larger threshold should be used on RISC machines. @{b}SCCS_DIRECTIVE@{ub} Define this if the preprocessor should ignore @{b}#sccs@{ub} directives and print no error message. @{b}NO_IMPLICIT_EXTERN_C@{ub} Define this macro if the system header files support C++ as well as C. This macro inhibits the usual method of using system header files in C++, which is to pretend that the file's contents are enclosed in @{b}extern "C" {...}@{ub}. @{b}HANDLE_PRAGMA (@{I}STREAM@{ui})@{ub} Define this macro if you want to implement any pragmas. If defined, it should be a C statement to be executed when @{b}#pragma@{ub} is seen. The argument @{I}STREAM@{ui} is the stdio input stream from which the source text can be read. It is generally a bad idea to implement new uses of @{b}#pragma@{ub}. The only reason to define this macro is for compatibility with other compilers that do support @{b}#pragma@{ub} for the sake of any user programs which already use it. @{b}VALID_MACHINE_DECL_ATTRIBUTE (@{I}DECL@{ui}, @{I}ATTRIBUTES@{ui}, @{I}IDENTIFIER@{ui}, @{I}ARGS@{ui})@{ub} If defined, a C expression whose value is nonzero if @{I}IDENTIFIER@{ui} with arguments @{I}ARGS@{ui} is a valid machine specific attribute for @{I}DECL@{ui}. The attributes in @{I}ATTRIBUTES@{ui} have previously been assigned to @{I}DECL@{ui}. @{b}VALID_MACHINE_TYPE_ATTRIBUTE (@{I}TYPE@{ui}, @{I}ATTRIBUTES@{ui}, @{I}IDENTIFIER@{ui}, @{I}ARGS@{ui})@{ub} If defined, a C expression whose value is nonzero if @{I}IDENTIFIER@{ui} with arguments @{I}ARGS@{ui} is a valid machine specific attribute for @{I}TYPE@{ui}. The attributes in @{I}ATTRIBUTES@{ui} have previously been assigned to @{I}TYPE@{ui}. @{b}COMP_TYPE_ATTRIBUTES (@{I}TYPE1@{ui}, @{I}TYPE2@{ui})@{ub} If defined, a C expression whose value is zero if the attributes on @{I}TYPE1@{ui} and @{I}TYPE2@{ui} are incompatible, one if they are compatible, and two if they are nearly compatible (which causes a warning to be generated). @{b}SET_DEFAULT_TYPE_ATTRIBUTES (@{I}TYPE@{ui})@{ub} If defined, a C statement that assigns default attributes to newly defined @{I}TYPE@{ui}. @{b}DOLLARS_IN_IDENTIFIERS@{ub} Define this macro to control use of the character @{b}$@{ub} in identifier names. The value should be 0, 1, or 2. 0 means @{b}$@{ub} is not allowed by default; 1 means it is allowed by default if @{b}-traditional@{ub} is used; 2 means it is allowed by default provided @{b}-ansi@{ub} is not used. 1 is the default; there is no need to define this macro in that case. @{b}NO_DOLLAR_IN_LABEL@{ub} Define this macro if the assembler does not accept the character @{b}$@{ub} in label names. By default constructors and destructors in G++ have @{b}$@{ub} in the identifiers. If this macro is defined, @{b}.@{ub} is used instead. @{b}NO_DOT_IN_LABEL@{ub} Define this macro if the assembler does not accept the character @{b}.@{ub} in label names. By default constructors and destructors in G++ have names that use @{b}.@{ub}. If this macro is defined, these names are rewritten to avoid @{b}.@{ub}. @{b}DEFAULT_MAIN_RETURN@{ub} Define this macro if the target system expects every program's @{b}main@{ub} function to return a standard "success" value by default (if no other value is explicitly returned). The definition should be a C statement (sans semicolon) to generate the appropriate rtl instructions. It is used only when compiling the end of @{b}main@{ub}. @{b}HAVE_ATEXIT@{ub} Define this if the target system supports the function @{b}atexit@{ub} from the ANSI C standard. If this is not defined, and @{b}INIT_SECTION_ASM_OP@{ub} is not defined, a default @{b}exit@{ub} function will be provided to support C++. @{b}EXIT_BODY@{ub} Define this if your @{b}exit@{ub} function needs to do something besides calling an external function @{b}_cleanup@{ub} before terminating with @{b}_exit@{ub}. The @{b}EXIT_BODY@{ub} macro is only needed if netiher @{b}HAVE_ATEXIT@{ub} nor @{b}INIT_SECTION_ASM_OP@{ub} are defined. @{b}INSN_SETS_ARE_DELAYED (@{I}INSN@{ui})@{ub} Define this macro as a C expression that is nonzero if it is safe for the delay slot scheduler to place instructions in the delay slot of @{I}INSN@{ui}, even if they appear to use a resource set or clobbered in @{I}INSN@{ui}. @{I}INSN@{ui} is always a @{b}jump_insn@{ub} or an @{b}insn@{ub}; GNU CC knows that every @{b}call_insn@{ub} has this behavior. On machines where some @{b}insn@{ub} or @{b}jump_insn@{ub} is really a function call and hence has this behavior, you should define this macro. You need not define this macro if it would always return zero. @{b}INSN_REFERENCES_ARE_DELAYED (@{I}INSN@{ui})@{ub} Define this macro as a C expression that is nonzero if it is safe for the delay slot scheduler to place instructions in the delay slot of @{I}INSN@{ui}, even if they appear to set or clobber a resource referenced in @{I}INSN@{ui}. @{I}INSN@{ui} is always a @{b}jump_insn@{ub} or an @{b}insn@{ub}. On machines where some @{b}insn@{ub} or @{b}jump_insn@{ub} is really a function call and its operands are registers whose use is actually in the subroutine it calls, you should define this macro. Doing so allows the delay slot scheduler to move instructions which copy arguments into the argument registers into the delay slot of @{I}INSN@{ui}. You need not define this macro if it would always return zero. @{b}MACHINE_DEPENDENT_REORG (@{I}INSN@{ui})@{ub} In rare cases, correct code generation requires extra machine dependent processing between the second jump optimization pass and delayed branch scheduling. On those machines, define this macro as a C statement to act on the code starting at @{I}INSN@{ui}. @EndNode @Node "Config" "gcc.guide/Config" @Next "Fragments" @Prev "Target Macros" @Toc "Main" The Configuration File ********************** The configuration file @{b}xm-@{I}MACHINE@{ui}.h@{ub} contains macro definitions that describe the machine and system on which the compiler is running, unlike the definitions in @{b}@{I}MACHINE@{ui}.h@{ub}, which describe the machine for which the compiler is producing output. Most of the values in @{b}xm-@{I}MACHINE@{ui}.h@{ub} are actually the same on all machines that GNU CC runs on, so large parts of all configuration files are identical. But there are some macros that vary: @{b}USG@{ub} Define this macro if the host system is System V. @{b}VMS@{ub} Define this macro if the host system is VMS. @{b}FATAL_EXIT_CODE@{ub} A C expression for the status code to be returned when the compiler exits after serious errors. @{b}SUCCESS_EXIT_CODE@{ub} A C expression for the status code to be returned when the compiler exits without serious errors. @{b}HOST_WORDS_BIG_ENDIAN@{ub} Defined if the host machine stores words of multi-word values in big-endian order. (GNU CC does not depend on the host byte ordering within a word.) @{b}HOST_FLOAT_WORDS_BIG_ENDIAN@{ub} Define this macro to be 1 if the host machine stores @{b}DFmode@{ub}, @{b}XFmode@{ub} or @{b}TFmode@{ub} floating point numbers in memory with the word containing the sign bit at the lowest address; otherwise, define it to be zero. This macro need not be defined if the ordering is the same as for multi-word integers. @{b}HOST_FLOAT_FORMAT@{ub} A numeric code distinguishing the floating point format for the host machine. See @{b}TARGET_FLOAT_FORMAT@{ub} in @{"Storage Layout" Link "Storage Layout"} for the alternatives and default. @{b}HOST_BITS_PER_CHAR@{ub} A C expression for the number of bits in @{b}char@{ub} on the host machine. @{b}HOST_BITS_PER_SHORT@{ub} A C expression for the number of bits in @{b}short@{ub} on the host machine. @{b}HOST_BITS_PER_INT@{ub} A C expression for the number of bits in @{b}int@{ub} on the host machine. @{b}HOST_BITS_PER_LONG@{ub} A C expression for the number of bits in @{b}long@{ub} on the host machine. @{b}ONLY_INT_FIELDS@{ub} Define this macro to indicate that the host compiler only supports @{b}int@{ub} bit fields, rather than other integral types, including @{b}enum@{ub}, as do most C compilers. @{b}OBSTACK_CHUNK_SIZE@{ub} A C expression for the size of ordinary obstack chunks. If you don't define this, a usually-reasonable default is used. @{b}OBSTACK_CHUNK_ALLOC@{ub} The function used to allocate obstack chunks. If you don't define this, @{b}xmalloc@{ub} is used. @{b}OBSTACK_CHUNK_FREE@{ub} The function used to free obstack chunks. If you don't define this, @{b}free@{ub} is used. @{b}USE_C_ALLOCA@{ub} Define this macro to indicate that the compiler is running with the @{b}alloca@{ub} implemented in C. This version of @{b}alloca@{ub} can be found in the file @{b}alloca.c@{ub}; to use it, you must also alter the @{b}Makefile@{ub} variable @{b}ALLOCA@{ub}. (This is done automatically for the systems on which we know it is needed.) If you do define this macro, you should probably do it as follows: #ifndef __GNUC__ #define USE_C_ALLOCA #else #define alloca __builtin_alloca #endif so that when the compiler is compiled with GNU CC it uses the more efficient built-in @{b}alloca@{ub} function. @{b}FUNCTION_CONVERSION_BUG@{ub} Define this macro to indicate that the host compiler does not properly handle converting a function value to a pointer-to-function when it is used in an expression. @{b}HAVE_VPRINTF@{ub} Define this if the library function @{b}vprintf@{ub} is available on your system. @{b}MULTIBYTE_CHARS@{ub} Define this macro to enable support for multibyte characters in the input to GNU CC. This requires that the host system support the ANSI C library functions for converting multibyte characters to wide characters. @{b}HAVE_PUTENV@{ub} Define this if the library function @{b}putenv@{ub} is available on your system. @{b}POSIX@{ub} Define this if your system is POSIX.1 compliant. @{b}NO_SYS_SIGLIST@{ub} Define this if your system @{i}does not@{ui} provide the variable @{b}sys_siglist@{ub}. @{b}DONT_DECLARE_SYS_SIGLIST@{ub} Define this if your system has the variable @{b}sys_siglist@{ub}, and there is already a declaration of it in the system header files. @{b}USE_PROTOTYPES@{ub} Define this to be 1 if you know that the host compiler supports prototypes, even if it doesn't define __STDC__, or define it to be 0 if you do not want any prototypes used in compiling GNU CC. If @{b}USE_PROTOTYPES@{ub} is not defined, it will be determined automatically whether your compiler supports prototypes by checking if @{b}__STDC__@{ub} is defined. @{b}NO_MD_PROTOTYPES@{ub} Define this if you wish suppression of prototypes generated from the machine description file, but to use other prototypes within GNU CC. If @{b}USE_PROTOTYPES@{ub} is defined to be 0, or the host compiler does not support prototypes, this macro has no effect. @{b}MD_CALL_PROTOTYPES@{ub} Define this if you wish to generate prototypes for the @{b}gen_call@{ub} or @{b}gen_call_value@{ub} functions generated from the machine description file. If @{b}USE_PROTOTYPES@{ub} is defined to be 0, or the host compiler does not support prototypes, or @{b}NO_MD_PROTOTYPES@{ub} is defined, this macro has no effect. As soon as all of the machine descriptions are modified to have the appropriate number of arguments, this macro will be removed. Some systems do provide this variable, but with a different name such as @{b}_sys_siglist@{ub}. On these systems, you can define @{b}sys_siglist@{ub} as a macro which expands into the name actually provided. @{b}NO_STAB_H@{ub} Define this if your system does not have the include file @{b}stab.h@{ub}. If @{b}USG@{ub} is defined, @{b}NO_STAB_H@{ub} is assumed. @{b}PATH_SEPARATOR@{ub} Define this macro to be a C character constant representing the character used to separate components in paths. The default value is. the colon character @{b}DIR_SEPARATOR@{ub} If your system uses some character other than slash to separate directory names within a file specification, define this macro to be a C character constant specifying that character. When GNU CC displays file names, the character you specify will be used. GNU CC will test for both slash and the character you specify when parsing filenames. @{b}OBJECT_SUFFIX@{ub} Define this macro to be a C string representing the suffix for object files on your machine. If you do not define this macro, GNU CC will use @{b}.o@{ub} as the suffix for object files. @{b}EXECUTABLE_SUFFIX@{ub} Define this macro to be a C string representing the suffix for executable files on your machine. If you do not define this macro, GNU CC will use the null string as the suffix for object files. @{b}COLLECT_EXPORT_LIST@{ub} If defined, @{b}collect2@{ub} will scan the individual object files specified on its command line and create an export list for the linker. Define this macro for systems like AIX, where the linker discards object files that are not referenced from @{b}main@{ub} and uses export lists. In addition, configuration files for system V define @{b}bcopy@{ub}, @{b}bzero@{ub} and @{b}bcmp@{ub} as aliases. Some files define @{b}alloca@{ub} as a macro when compiled with GNU CC, in order to take advantage of the benefit of GNU CC's built-in @{b}alloca@{ub}. @EndNode @Node "Fragments" "gcc.guide/Fragments" @Next "Index" @Prev "Config" @Toc "Main" Makefile Fragments ****************** When you configure GNU CC using the @{b}configure@{ub} script (see @{"Installation" Link "Installation"}), it will construct the file @{b}Makefile@{ub} from the template file @{b}Makefile.in@{ub}. When it does this, it will incorporate makefile fragment files from the @{b}config@{ub} directory, named @{b}t-@{I}TARGET@{ui}@{ub} and @{b}x-@{I}HOST@{ui}@{ub}. If these files do not exist, it means nothing needs to be added for a given target or host. @{" Target Fragment " Link "Target Fragment"} Writing the @{b}t-@{I}TARGET@{ui}@{ub} file. @{" Host Fragment " Link "Host Fragment"} Writing the @{b}x-@{I}HOST@{ui}@{ub} file. @EndNode @Node "Target Fragment" "gcc.guide/Target Fragment" @Next "Host Fragment" @Toc "Fragments" The Target Makefile Fragment ============================ The target makefile fragment, @{b}t-@{I}TARGET@{ui}@{ub}, defines special target dependent variables and targets used in the @{b}Makefile@{ub}: @{b}LIBGCC1@{ub} The rule to use to build @{b}libgcc1.a@{ub}. If your target does not need to use the functions in @{b}libgcc1.a@{ub}, set this to empty. See @{"Interface" Link "Interface"}. @{b}CROSS_LIBGCC1@{ub} The rule to use to build @{b}libgcc1.a@{ub} when building a cross compiler. If your target does not need to use the functions in @{b}libgcc1.a@{ub}, set this to empty. See @{"Cross Runtime" Link "Cross Runtime"}. @{b}LIBGCC2_CFLAGS@{ub} Compiler flags to use when compiling @{b}libgcc2.c@{ub}. @{b}LIB2FUNCS_EXTRA@{ub} A list of source file names to be compiled or assembled and inserted into @{b}libgcc.a@{ub}. @{b}CRTSTUFF_T_CFLAGS@{ub} Special flags used when compiling @{b}crtstuff.c@{ub}. See @{"Initialization" Link "Initialization"}. @{b}MULTILIB_OPTIONS@{ub} For some targets, invoking GNU CC in different ways produces objects that can not be linked together. For example, for some targets GNU CC produces both big and little endian code. For these targets, you must arrange for multiple versions of @{b}libgcc.a@{ub} to be compiled, one for each set of incompatible options. When GNU CC invokes the linker, it arranges to link in the right version of @{b}libgcc.a@{ub}, based on the command line options used. The @{b}MULTILIB_OPTIONS@{ub} macro lists the set of options for which special versions of @{b}libgcc.a@{ub} must be built. Write options that are mutually incompatible side by side, separated by a slash. Write options that may be used together separated by a space. The build procedure will build all combinations of compatible options. For example, if you set @{b}MULTILIB_OPTIONS@{ub} to @{b}m68000/m68020 msoft-float@{ub}, @{b}Makefile@{ub} will build special versions of @{b}libgcc.a@{ub} using the options @{b}-m68000@{ub}, @{b}-m68020@{ub}, @{b}-msoft-float@{ub}, @{b}-m68000 -msoft-float@{ub}, and @{b}-m68020 -msoft-float@{ub}. @{b}MULTILIB_DIRNAMES@{ub} If @{b}MULTILIB_OPTIONS@{ub} is used, this variable specifies the directory names that should be used to hold the various libraries. Write one element in @{b}MULTILIB_DIRNAMES@{ub} for each element in @{b}MULTILIB_OPTIONS@{ub}. If @{b}MULTILIB_DIRNAMES@{ub} is not used, the default value will be @{b}MULTILIB_OPTIONS@{ub}, with all slashes treated as spaces. For example, if @{b}MULTILIB_OPTIONS@{ub} is @{b}m68000/m68020 msoft-float@{ub}, then the default value of @{b}MULTILIB_DIRNAMES@{ub} is @{b}m68000 m68020 msoft-float@{ub}. You may specify a different value if you desire a different set of directory names. @{b}MULTILIB_MATCHES@{ub} Sometimes the same option may be written in two different ways. If an option is listed in @{b}MULTILIB_OPTIONS@{ub}, GNU CC needs to know about any synonyms. In that case, set @{b}MULTILIB_MATCHES@{ub} to a list of items of the form @{b}option=option@{ub} to describe all relevant synonyms. For example, @{b}m68000=mc68000 m68020=mc68020@{ub}. @EndNode @Node "Host Fragment" "gcc.guide/Host Fragment" @Prev "Target Fragment" @Toc "Fragments" The Host Makefile Fragment ========================== The host makefile fragment, @{b}x-@{I}HOST@{ui}@{ub}, defines special host dependent variables and targets used in the @{b}Makefile@{ub}: @{b}CC@{ub} The compiler to use when building the first stage. @{b}CLIB@{ub} Additional host libraries to link with. @{b}OLDCC@{ub} The compiler to use when building @{b}libgcc1.a@{ub} for a native compilation. @{b}OLDAR@{ub} The version of @{b}ar@{ub} to use when building @{b}libgcc1.a@{ub} for a native compilation. @{b}INSTALL@{ub} The install program to use. @EndNode @Node "Index" "gcc.guide/Index" @Prev "Fragments" @Toc "Main" Index ***** @Index "Index" @{" ! in constraint " Link "Multi-Alternative"} Multi-Alternative @{" # in constraint " Link "Modifiers"} Modifiers @{" # in template " Link "Output Template"} Output Template @{" #pragma " Link "Misc"} Misc @{" #pragma implementation, implied " Link "C++ Interface"} C++ Interface @{" #pragma, reason for not using " Link "Function Attributes"} Function Attributes @{" $ " Link "Dollar Signs"} Dollar Signs @{" % in constraint " Link "Modifiers"} Modifiers @{" % in template " Link "Output Template"} Output Template @{" & in constraint " Link "Modifiers"} Modifiers @{" ' " Link "Incompatibilities"} Incompatibilities @{" (nil) " Link "RTL Objects"} RTL Objects @{" * in constraint " Link "Modifiers"} Modifiers @{" * in template " Link "Output Statement"} Output Statement @{" + in constraint " Link "Modifiers"} Modifiers @{" -lgcc, use with -nodefaultlibs " Link "Link Options"} Link Options @{" -lgcc, use with -nostdlib " Link "Link Options"} Link Options @{" -nodefaultlibs and unresolved references " Link "Link Options"} Link Options @{" -nostdlib and unresolved references " Link "Link Options"} Link Options @{" // " Link "C++ Comments"} C++ Comments @{" /i in RTL dump " Link "Flags"} Flags @{" /s in RTL dump " Link "Flags"} Flags @{" /u in RTL dump " Link "Flags"} Flags @{" /v in RTL dump " Link "Flags"} Flags @{" 0 in constraint " Link "Simple Constraints"} Simple Constraints @{" < in constraint " Link "Simple Constraints"} Simple Constraints @{" in constraint " Link "Simple Constraints"} Simple Constraints @{" >? " Link "Min and Max"} Min and Max @{" ? in constraint " Link "Multi-Alternative"} Multi-Alternative @{" ?: extensions <1> " Link "Lvalues"} Lvalues @{" ?: extensions " Link "Conditionals"} Conditionals @{" ?: side effect " Link "Conditionals"} Conditionals @{" \ " Link "Output Template"} Output Template @{" _ in variables in macros " Link "Naming Types"} Naming Types @{" __bb_init_func " Link "Profiling"} Profiling @{" __builtin_apply " Link "Constructing Calls"} Constructing Calls @{" __builtin_apply_args " Link "Constructing Calls"} Constructing Calls @{" __builtin_args_info " Link "Varargs"} Varargs @{" __builtin_classify_type " Link "Varargs"} Varargs @{" __builtin_next_arg " Link "Varargs"} Varargs @{" __builtin_return " Link "Constructing Calls"} Constructing Calls @{" __builtin_saveregs " Link "Varargs"} Varargs @{" __CTOR_LIST__ " Link "Initialization"} Initialization @{" __DTOR_LIST__ " Link "Initialization"} Initialization @{" __main " Link "Collect2"} Collect2 @{" abort <1> " Link "Portability"} Portability @{" abort " Link "C Dialect Options"} C Dialect Options @{" abs <1> " Link "C Dialect Options"} C Dialect Options @{" abs " Link "Arithmetic"} Arithmetic @{" abs and attributes " Link "Expressions"} Expressions @{" absM2 instruction pattern " Link "Standard Names"} Standard Names @{" absolute value " Link "Arithmetic"} Arithmetic @{" access to operands " Link "Accessors"} Accessors @{" accessors " Link "Accessors"} Accessors @{" ACCUMULATE_OUTGOING_ARGS " Link "Stack Arguments"} Stack Arguments @{" ACCUMULATE_OUTGOING_ARGS and stack frames " Link "Function Entry"} Function Entry @{" ADDITIONAL_REGISTER_NAMES " Link "Instruction Output"} Instruction Output @{" addM3 instruction pattern " Link "Standard Names"} Standard Names @{" addr_diff_vec " Link "Side Effects"} Side Effects @{" addr_diff_vec, length of " Link "Insn Lengths"} Insn Lengths @{" addr_vec " Link "Side Effects"} Side Effects @{" addr_vec, length of " Link "Insn Lengths"} Insn Lengths @{" address " Link "RTL Template"} RTL Template @{" address constraints " Link "Simple Constraints"} Simple Constraints @{" address of a label " Link "Labels as Values"} Labels as Values @{" ADDRESS_COST " Link "Costs"} Costs @{" address_operand " Link "Simple Constraints"} Simple Constraints @{" addressing modes " Link "Addressing Modes"} Addressing Modes @{" ADJUST_COST " Link "Costs"} Costs @{" ADJUST_INSN_LENGTH " Link "Insn Lengths"} Insn Lengths @{" aggregates as return values " Link "Aggregate Return"} Aggregate Return @{" alias attribute " Link "Function Attributes"} Function Attributes @{" aligned attribute <1> " Link "Type Attributes"} Type Attributes @{" aligned attribute " Link "Variable Attributes"} Variable Attributes @{" alignment " Link "Alignment"} Alignment @{" ALL_REGS " Link "Register Classes"} Register Classes @{" Alliant " Link "Interoperation"} Interoperation @{" alloca " Link "C Dialect Options"} C Dialect Options @{" alloca and SunOs " Link "Installation"} Installation @{" alloca vs variable-length arrays " Link "Variable Length"} Variable Length @{" alloca, for SunOs " Link "Sun Install"} Sun Install @{" alloca, for Unos " Link "Configurations"} Configurations @{" allocate_stack instruction pattern " Link "Standard Names"} Standard Names @{" ALLOCATE_TRAMPOLINE " Link "Trampolines"} Trampolines @{" alternate keywords " Link "Alternate Keywords"} Alternate Keywords @{" AMD29K options " Link "AMD29K Options"} AMD29K Options @{" analysis, data flow " Link "Passes"} Passes @{" and " Link "Arithmetic"} Arithmetic @{" and and attributes " Link "Expressions"} Expressions @{" and, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @{" andM3 instruction pattern " Link "Standard Names"} Standard Names @{" ANSI support " Link "C Dialect Options"} C Dialect Options @{" apostrophes " Link "Incompatibilities"} Incompatibilities @{" APPLY_RESULT_SIZE " Link "Scalar Return"} Scalar Return @{" ARG_POINTER_REGNUM " Link "Frame Registers"} Frame Registers @{" ARG_POINTER_REGNUM and virtual registers " Link "Regs and Memory"} Regs and Memory @{" arg_pointer_rtx " Link "Frame Registers"} Frame Registers @{" ARGS_GROW_DOWNWARD " Link "Frame Layout"} Frame Layout @{" argument passing " Link "Interface"} Interface @{" arguments in frame (88k) " Link "M88K Options"} M88K Options @{" arguments in registers " Link "Register Arguments"} Register Arguments @{" arguments on stack " Link "Stack Arguments"} Stack Arguments @{" arithmetic libraries " Link "Interface"} Interface @{" arithmetic shift " Link "Arithmetic"} Arithmetic @{" arithmetic simplifications " Link "Passes"} Passes @{" arithmetic, in RTL " Link "Arithmetic"} Arithmetic @{" ARM options " Link "ARM Options"} ARM Options @{" arrays of length zero " Link "Zero Length"} Zero Length @{" arrays of variable length " Link "Variable Length"} Variable Length @{" arrays, non-lvalue " Link "Subscripting"} Subscripting @{" ashift " Link "Arithmetic"} Arithmetic @{" ashift and attributes " Link "Expressions"} Expressions @{" ashiftrt " Link "Arithmetic"} Arithmetic @{" ashiftrt and attributes " Link "Expressions"} Expressions @{" ashlM3 instruction pattern " Link "Standard Names"} Standard Names @{" ashrM3 instruction pattern " Link "Standard Names"} Standard Names @{" asm expressions " Link "Extended Asm"} Extended Asm @{" ASM_APP_OFF " Link "File Framework"} File Framework @{" ASM_APP_ON " Link "File Framework"} File Framework @{" ASM_BYTE_OP " Link "Data Output"} Data Output @{" ASM_CLOSE_PAREN " Link "Data Output"} Data Output @{" ASM_COMMENT_START " Link "File Framework"} File Framework @{" ASM_DECLARE_FUNCTION_NAME " Link "Label Output"} Label Output @{" ASM_DECLARE_FUNCTION_SIZE " Link "Label Output"} Label Output @{" ASM_DECLARE_OBJECT_NAME " Link "Label Output"} Label Output @{" ASM_FILE_END " Link "File Framework"} File Framework @{" ASM_FILE_START " Link "File Framework"} File Framework @{" ASM_FINAL_SPEC " Link "Driver"} Driver @{" ASM_FINISH_DECLARE_OBJECT " Link "Label Output"} Label Output @{" ASM_FORMAT_PRIVATE_NAME " Link "Label Output"} Label Output @{" asm_fprintf " Link "Instruction Output"} Instruction Output @{" ASM_GENERATE_INTERNAL_LABEL " Link "Label Output"} Label Output @{" ASM_GLOBALIZE_LABEL " Link "Label Output"} Label Output @{" ASM_IDENTIFY_GCC " Link "File Framework"} File Framework @{" asm_input " Link "Side Effects"} Side Effects @{" ASM_NO_SKIP_IN_TEXT " Link "Alignment Output"} Alignment Output @{" asm_noperands " Link "Insns"} Insns @{" ASM_OPEN_PAREN " Link "Data Output"} Data Output @{" asm_operands, RTL sharing " Link "Sharing"} Sharing @{" asm_operands, usage " Link "Assembler"} Assembler @{" ASM_OUTPUT_ADDR_DIFF_ELT " Link "Dispatch Tables"} Dispatch Tables @{" ASM_OUTPUT_ADDR_VEC_ELT " Link "Dispatch Tables"} Dispatch Tables @{" ASM_OUTPUT_ALIGN " Link "Alignment Output"} Alignment Output @{" ASM_OUTPUT_ALIGN_CODE " Link "Alignment Output"} Alignment Output @{" ASM_OUTPUT_ALIGNED_COMMON " Link "Uninitialized Data"} Uninitialized Data @{" ASM_OUTPUT_ALIGNED_LOCAL " Link "Uninitialized Data"} Uninitialized Data @{" ASM_OUTPUT_ASCII " Link "Data Output"} Data Output @{" ASM_OUTPUT_BYTE " Link "Data Output"} Data Output @{" ASM_OUTPUT_CASE_END " Link "Dispatch Tables"} Dispatch Tables @{" ASM_OUTPUT_CASE_LABEL " Link "Dispatch Tables"} Dispatch Tables @{" ASM_OUTPUT_CHAR " Link "Data Output"} Data Output @{" ASM_OUTPUT_COMMON " Link "Uninitialized Data"} Uninitialized Data @{" ASM_OUTPUT_CONSTRUCTOR " Link "Macros for Initialization"} Macros for Initialization @{" ASM_OUTPUT_DEF " Link "Label Output"} Label Output @{" ASM_OUTPUT_DESTRUCTOR " Link "Macros for Initialization"} Macros for Initialization @{" ASM_OUTPUT_DOUBLE " Link "Data Output"} Data Output @{" ASM_OUTPUT_DOUBLE_INT " Link "Data Output"} Data Output @{" ASM_OUTPUT_EXTERNAL " Link "Label Output"} Label Output @{" ASM_OUTPUT_EXTERNAL_LIBCALL " Link "Label Output"} Label Output @{" ASM_OUTPUT_FLOAT " Link "Data Output"} Data Output @{" ASM_OUTPUT_IDENT " Link "File Framework"} File Framework @{" ASM_OUTPUT_INT " Link "Data Output"} Data Output @{" ASM_OUTPUT_INTERNAL_LABEL " Link "Label Output"} Label Output @{" ASM_OUTPUT_LABEL " Link "Label Output"} Label Output @{" ASM_OUTPUT_LABELREF " Link "Label Output"} Label Output @{" ASM_OUTPUT_LOCAL " Link "Uninitialized Data"} Uninitialized Data @{" ASM_OUTPUT_LONG_DOUBLE " Link "Data Output"} Data Output @{" ASM_OUTPUT_LOOP_ALIGN " Link "Alignment Output"} Alignment Output @{" ASM_OUTPUT_OPCODE " Link "Instruction Output"} Instruction Output @{" ASM_OUTPUT_POOL_PROLOGUE " Link "Data Output"} Data Output @{" ASM_OUTPUT_QUADRUPLE_INT " Link "Data Output"} Data Output @{" ASM_OUTPUT_REG_POP " Link "Instruction Output"} Instruction Output @{" ASM_OUTPUT_REG_PUSH " Link "Instruction Output"} Instruction Output @{" ASM_OUTPUT_SECTION_NAME " Link "File Framework"} File Framework @{" ASM_OUTPUT_SHARED_COMMON " Link "Uninitialized Data"} Uninitialized Data @{" ASM_OUTPUT_SHARED_LOCAL " Link "Uninitialized Data"} Uninitialized Data @{" ASM_OUTPUT_SHORT " Link "Data Output"} Data Output @{" ASM_OUTPUT_SKIP " Link "Alignment Output"} Alignment Output @{" ASM_OUTPUT_SOURCE_FILENAME " Link "File Framework"} File Framework @{" ASM_OUTPUT_SOURCE_LINE " Link "File Framework"} File Framework @{" ASM_OUTPUT_SPECIAL_POOL_ENTRY " Link "Data Output"} Data Output @{" ASM_SPEC " Link "Driver"} Driver @{" ASM_STABD_OP " Link "DBX Options"} DBX Options @{" ASM_STABN_OP " Link "DBX Options"} DBX Options @{" ASM_STABS_OP " Link "DBX Options"} DBX Options @{" ASM_WEAKEN_LABEL " Link "Label Output"} Label Output @{" assemble_name " Link "Label Output"} Label Output @{" assembler format " Link "File Framework"} File Framework @{" assembler instructions " Link "Extended Asm"} Extended Asm @{" assembler instructions in RTL " Link "Assembler"} Assembler @{" assembler names for identifiers " Link "Asm Labels"} Asm Labels @{" assembler syntax, 88k " Link "M88K Options"} M88K Options @{" ASSEMBLER_DIALECT " Link "Instruction Output"} Instruction Output @{" assembly code, invalid " Link "Bug Criteria"} Bug Criteria @{" assigning attribute values to insns " Link "Tagging Insns"} Tagging Insns @{" asterisk in template " Link "Output Statement"} Output Statement @{" atof " Link "Cross-compilation"} Cross-compilation @{" attr " Link "Tagging Insns"} Tagging Insns @{" attr_flag " Link "Expressions"} Expressions @{" attribute expressions " Link "Expressions"} Expressions @{" attribute of types " Link "Type Attributes"} Type Attributes @{" attribute of variables " Link "Variable Attributes"} Variable Attributes @{" attribute specifications " Link "Attr Example"} Attr Example @{" attribute specifications example " Link "Attr Example"} Attr Example @{" attributes, defining " Link "Defining Attributes"} Defining Attributes @{" autoincrement addressing, availability " Link "Portability"} Portability @{" autoincrement/decrement addressing " Link "Simple Constraints"} Simple Constraints @{" autoincrement/decrement analysis " Link "Passes"} Passes @{" automatic inline for C++ member fns " Link "Inline"} Inline @{" backslash " Link "Output Template"} Output Template @{" backtrace for bug reports " Link "Bug Reporting"} Bug Reporting @{" barrier " Link "Insns"} Insns @{" BASE_REG_CLASS " Link "Register Classes"} Register Classes @{" basic blocks " Link "Passes"} Passes @{" bcmp " Link "Config"} Config @{" bCOND instruction pattern " Link "Standard Names"} Standard Names @{" bcopy, implicit usage " Link "Library Calls"} Library Calls @{" BIGGEST_ALIGNMENT " Link "Storage Layout"} Storage Layout @{" BIGGEST_FIELD_ALIGNMENT " Link "Storage Layout"} Storage Layout @{" Bison parser generator " Link "Installation"} Installation @{" bit fields " Link "Bit Fields"} Bit Fields @{" bit shift overflow (88k) " Link "M88K Options"} M88K Options @{" BITFIELD_NBYTES_LIMITED " Link "Storage Layout"} Storage Layout @{" BITS_BIG_ENDIAN " Link "Storage Layout"} Storage Layout @{" BITS_BIG_ENDIAN, effect on sign_extract " Link "Bit Fields"} Bit Fields @{" BITS_PER_UNIT " Link "Storage Layout"} Storage Layout @{" BITS_PER_WORD " Link "Storage Layout"} Storage Layout @{" bitwise complement " Link "Arithmetic"} Arithmetic @{" bitwise exclusive-or " Link "Arithmetic"} Arithmetic @{" bitwise inclusive-or " Link "Arithmetic"} Arithmetic @{" bitwise logical-and " Link "Arithmetic"} Arithmetic @{" BLKmode " Link "Machine Modes"} Machine Modes @{" BLKmode, and function return values " Link "Calls"} Calls @{" BLOCK_PROFILER " Link "Profiling"} Profiling @{" BLOCK_PROFILER_CODE " Link "Profiling"} Profiling @{" BRANCH_COST " Link "Costs"} Costs @{" break_out_memory_refs " Link "Addressing Modes"} Addressing Modes @{" bug criteria " Link "Bug Criteria"} Bug Criteria @{" bug report mailing lists " Link "Bug Lists"} Bug Lists @{" bugs " Link "Bugs"} Bugs @{" bugs, known " Link "Trouble"} Trouble @{" builtin functions " Link "C Dialect Options"} C Dialect Options @{" byte writes (29k) " Link "AMD29K Options"} AMD29K Options @{" byte_mode " Link "Machine Modes"} Machine Modes @{" BYTES_BIG_ENDIAN " Link "Storage Layout"} Storage Layout @{" bzero " Link "Config"} Config @{" bzero, implicit usage " Link "Library Calls"} Library Calls @{" C compilation options " Link "Invoking GCC"} Invoking GCC @{" C intermediate output, nonexistent " Link "G++ and GCC"} G++ and GCC @{" C language extensions " Link "C Extensions"} C Extensions @{" C language, traditional " Link "C Dialect Options"} C Dialect Options @{" C statements for assembler output " Link "Output Statement"} Output Statement @{" c++ " Link "Invoking G++"} Invoking G++ @{" C++ " Link "G++ and GCC"} G++ and GCC @{" C++ comments " Link "C++ Comments"} C++ Comments @{" C++ compilation options " Link "Invoking GCC"} Invoking GCC @{" C++ interface and implementation headers " Link "C++ Interface"} C++ Interface @{" C++ language extensions " Link "C++ Extensions"} C++ Extensions @{" C++ member fns, automatically inline " Link "Inline"} Inline @{" C++ misunderstandings " Link "C++ Misunderstandings"} C++ Misunderstandings @{" C++ named return value " Link "Naming Results"} Naming Results @{" C++ options, command line " Link "C++ Dialect Options"} C++ Dialect Options @{" C++ pragmas, effect on inlining " Link "C++ Interface"} C++ Interface @{" C++ signatures " Link "C++ Signatures"} C++ Signatures @{" C++ source file suffixes " Link "Invoking G++"} Invoking G++ @{" C++ static data, declaring and defining " Link "Static Definitions"} Static Definitions @{" C++ subtype polymorphism " Link "C++ Signatures"} C++ Signatures @{" C++ type abstraction " Link "C++ Signatures"} C++ Signatures @{" C_INCLUDE_PATH " Link "Environment Variables"} Environment Variables @{" call " Link "Side Effects"} Side Effects @{" call instruction pattern " Link "Standard Names"} Standard Names @{" call usage " Link "Calls"} Calls @{" call-clobbered register " Link "Register Basics"} Register Basics @{" call-saved register " Link "Register Basics"} Register Basics @{" call-used register " Link "Register Basics"} Register Basics @{" call_insn " Link "Insns"} Insns @{" call_insn and /u " Link "Flags"} Flags @{" CALL_INSN_FUNCTION_USAGE " Link "Insns"} Insns @{" call_pop instruction pattern " Link "Standard Names"} Standard Names @{" CALL_USED_REGISTERS " Link "Register Basics"} Register Basics @{" call_used_regs " Link "Register Basics"} Register Basics @{" call_value instruction pattern " Link "Standard Names"} Standard Names @{" call_value_pop instruction pattern " Link "Standard Names"} Standard Names @{" CALLER_SAVE_PROFITABLE " Link "Caller Saves"} Caller Saves @{" calling conventions " Link "Stack and Calling"} Stack and Calling @{" calling functions in RTL " Link "Calls"} Calls @{" CAN_DEBUG_WITHOUT_FP " Link "Run-time Target"} Run-time Target @{" CAN_ELIMINATE " Link "Elimination"} Elimination @{" canonicalization of instructions " Link "Insn Canonicalizations"} Insn Canonicalizations @{" CANONICALIZE_COMPARISON " Link "Condition Code"} Condition Code @{" case labels in initializers " Link "Labeled Elements"} Labeled Elements @{" case ranges " Link "Case Ranges"} Case Ranges @{" case sensitivity and VMS " Link "VMS Misc"} VMS Misc @{" CASE_DROPS_THROUGH " Link "Misc"} Misc @{" CASE_VALUES_THRESHOLD " Link "Misc"} Misc @{" CASE_VECTOR_MODE " Link "Misc"} Misc @{" CASE_VECTOR_PC_RELATIVE " Link "Misc"} Misc @{" casesi instruction pattern " Link "Standard Names"} Standard Names @{" cast to a union " Link "Cast to Union"} Cast to Union @{" casts as lvalues " Link "Lvalues"} Lvalues @{" CC " Link "Host Fragment"} Host Fragment @{" cc0 " Link "Regs and Memory"} Regs and Memory @{" cc0, RTL sharing " Link "Sharing"} Sharing @{" cc0_rtx " Link "Regs and Memory"} Regs and Memory @{" CC1_SPEC " Link "Driver"} Driver @{" CC1PLUS_SPEC " Link "Driver"} Driver @{" cc_status " Link "Condition Code"} Condition Code @{" CC_STATUS_MDEP " Link "Condition Code"} Condition Code @{" CC_STATUS_MDEP_INIT " Link "Condition Code"} Condition Code @{" CCmode " Link "Machine Modes"} Machine Modes @{" CDImode " Link "Machine Modes"} Machine Modes @{" change_address " Link "Standard Names"} Standard Names @{" CHAR_TYPE_SIZE " Link "Type Layout"} Type Layout @{" CHECK_FLOAT_VALUE " Link "Storage Layout"} Storage Layout @{" CHImode " Link "Machine Modes"} Machine Modes @{" class definitions, register " Link "Register Classes"} Register Classes @{" class preference constraints " Link "Class Preferences"} Class Preferences @{" CLASS_LIKELY_SPILLED_P " Link "Register Classes"} Register Classes @{" CLASS_MAX_NREGS " Link "Register Classes"} Register Classes @{" classes of RTX codes " Link "Accessors"} Accessors @{" CLEAR_INSN_CACHE " Link "Trampolines"} Trampolines @{" CLIB " Link "Host Fragment"} Host Fragment @{" clobber " Link "Side Effects"} Side Effects @{" cmpM instruction pattern " Link "Standard Names"} Standard Names @{" cmpstrM instruction pattern " Link "Standard Names"} Standard Names @{" code generation conventions " Link "Code Gen Options"} Code Gen Options @{" code generation RTL sequences " Link "Expander Definitions"} Expander Definitions @{" code motion " Link "Passes"} Passes @{" code_label " Link "Insns"} Insns @{" code_label and /i " Link "Flags"} Flags @{" CODE_LABEL_NUMBER " Link "Insns"} Insns @{" codes, RTL expression " Link "RTL Objects"} RTL Objects @{" COImode " Link "Machine Modes"} Machine Modes @{" COLLECT_EXPORT_LIST " Link "Config"} Config @{" combiner pass " Link "Regs and Memory"} Regs and Memory @{" command options " Link "Invoking GCC"} Invoking GCC @{" comments, C++ style " Link "C++ Comments"} C++ Comments @{" common subexpression elimination " Link "Passes"} Passes @{" COMP_TYPE_ATTRIBUTES " Link "Misc"} Misc @{" compare " Link "Arithmetic"} Arithmetic @{" compare, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @{" compilation in a separate directory " Link "Other Dir"} Other Dir @{" compiler bugs, reporting " Link "Bug Reporting"} Bug Reporting @{" compiler compared to C++ preprocessor " Link "G++ and GCC"} G++ and GCC @{" compiler options, C++ " Link "C++ Dialect Options"} C++ Dialect Options @{" compiler passes and files " Link "Passes"} Passes @{" compiler version, specifying " Link "Target Options"} Target Options @{" COMPILER_PATH " Link "Environment Variables"} Environment Variables @{" complement, bitwise " Link "Arithmetic"} Arithmetic @{" complex numbers " Link "Complex"} Complex @{" compound expressions as lvalues " Link "Lvalues"} Lvalues @{" computed gotos " Link "Labels as Values"} Labels as Values @{" computing the length of an insn " Link "Insn Lengths"} Insn Lengths @{" cond " Link "Comparisons"} Comparisons @{" cond and attributes " Link "Expressions"} Expressions @{" condition code register " Link "Regs and Memory"} Regs and Memory @{" condition code status " Link "Condition Code"} Condition Code @{" condition codes " Link "Comparisons"} Comparisons @{" conditional expressions as lvalues " Link "Lvalues"} Lvalues @{" conditional expressions, extensions " Link "Conditionals"} Conditionals @{" CONDITIONAL_REGISTER_USAGE " Link "Register Basics"} Register Basics @{" conditions, in patterns " Link "Patterns"} Patterns @{" configuration file " Link "Config"} Config @{" configurations supported by GNU CC " Link "Configurations"} Configurations @{" conflicting types " Link "Disappointments"} Disappointments @{" const applied to function " Link "Function Attributes"} Function Attributes @{" const function attribute " Link "Function Attributes"} Function Attributes @{" const0_rtx " Link "Constants"} Constants @{" CONST0_RTX " Link "Constants"} Constants @{" const1_rtx " Link "Constants"} Constants @{" CONST1_RTX " Link "Constants"} Constants @{" const2_rtx " Link "Constants"} Constants @{" CONST2_RTX " Link "Constants"} Constants @{" CONST_CALL_P " Link "Flags"} Flags @{" CONST_COSTS " Link "Costs"} Costs @{" const_double " Link "Constants"} Constants @{" const_double, RTL sharing " Link "Sharing"} Sharing @{" CONST_DOUBLE_CHAIN " Link "Constants"} Constants @{" CONST_DOUBLE_LOW " Link "Constants"} Constants @{" CONST_DOUBLE_MEM " Link "Constants"} Constants @{" CONST_DOUBLE_OK_FOR_LETTER_P " Link "Register Classes"} Register Classes @{" const_int " Link "Constants"} Constants @{" const_int and attribute tests " Link "Expressions"} Expressions @{" const_int and attributes " Link "Expressions"} Expressions @{" const_int, RTL sharing " Link "Sharing"} Sharing @{" CONST_OK_FOR_LETTER_P " Link "Register Classes"} Register Classes @{" const_string " Link "Constants"} Constants @{" const_string and attributes " Link "Expressions"} Expressions @{" const_true_rtx " Link "Constants"} Constants @{" constant attributes " Link "Constant Attributes"} Constant Attributes @{" constant folding " Link "Passes"} Passes @{" constant folding and floating point " Link "Cross-compilation"} Cross-compilation @{" constant propagation " Link "Passes"} Passes @{" CONSTANT_ADDRESS_P " Link "Addressing Modes"} Addressing Modes @{" CONSTANT_ALIGNMENT " Link "Storage Layout"} Storage Layout @{" CONSTANT_P " Link "Addressing Modes"} Addressing Modes @{" CONSTANT_POOL_ADDRESS_P " Link "Flags"} Flags @{" constants in constraints " Link "Simple Constraints"} Simple Constraints @{" constm1_rtx " Link "Constants"} Constants @{" constraint modifier characters " Link "Modifiers"} Modifiers @{" constraint, matching " Link "Simple Constraints"} Simple Constraints @{" constraints " Link "Constraints"} Constraints @{" constraints, machine specific " Link "Machine Constraints"} Machine Constraints @{" constructing calls " Link "Constructing Calls"} Constructing Calls @{" constructor expressions " Link "Constructors"} Constructors @{" constructor function attribute " Link "Function Attributes"} Function Attributes @{" constructors vs goto " Link "Destructors and Goto"} Destructors and Goto @{" constructors, automatic calls " Link "Collect2"} Collect2 @{" constructors, output of " Link "Initialization"} Initialization @{" contributors " Link "Contributors"} Contributors @{" controlling register usage " Link "Register Basics"} Register Basics @{" controlling the compilation driver " Link "Driver"} Driver @{" conventions, run-time " Link "Interface"} Interface @{" conversions " Link "Conversions"} Conversions @{" Convex options " Link "Convex Options"} Convex Options @{" copy_rtx_if_shared " Link "Sharing"} Sharing @{" core dump " Link "Bug Criteria"} Bug Criteria @{" cos " Link "C Dialect Options"} C Dialect Options @{" costs of instructions " Link "Costs"} Costs @{" COSTS_N_INSNS " Link "Costs"} Costs @{" CPLUS_INCLUDE_PATH " Link "Environment Variables"} Environment Variables @{" CPP_PREDEFINES " Link "Run-time Target"} Run-time Target @{" CPP_SPEC " Link "Driver"} Driver @{" CQImode " Link "Machine Modes"} Machine Modes @{" cross compilation and floating point " Link "Cross-compilation"} Cross-compilation @{" cross compiling " Link "Target Options"} Target Options @{" cross-compiler, installation " Link "Cross-Compiler"} Cross-Compiler @{" cross-jumping " Link "Passes"} Passes @{" CROSS_LIBGCC1 " Link "Target Fragment"} Target Fragment @{" CRTSTUFF_T_CFLAGS " Link "Target Fragment"} Target Fragment @{" CSImode " Link "Machine Modes"} Machine Modes @{" CTImode " Link "Machine Modes"} Machine Modes @{" CUMULATIVE_ARGS " Link "Register Arguments"} Register Arguments @{" current_function_epilogue_delay_list " Link "Function Entry"} Function Entry @{" current_function_outgoing_args_size " Link "Stack Arguments"} Stack Arguments @{" current_function_pops_args " Link "Function Entry"} Function Entry @{" current_function_pretend_args_size " Link "Function Entry"} Function Entry @{" d in constraint " Link "Simple Constraints"} Simple Constraints @{" data flow analysis " Link "Passes"} Passes @{" DATA_ALIGNMENT " Link "Storage Layout"} Storage Layout @{" data_section " Link "Sections"} Sections @{" DATA_SECTION_ASM_OP " Link "Sections"} Sections @{" DBR_OUTPUT_SEQEND " Link "Instruction Output"} Instruction Output @{" dbr_sequence_length " Link "Instruction Output"} Instruction Output @{" DBX " Link "Interoperation"} Interoperation @{" DBX_BLOCKS_FUNCTION_RELATIVE " Link "DBX Options"} DBX Options @{" DBX_CONTIN_CHAR " Link "DBX Options"} DBX Options @{" DBX_CONTIN_LENGTH " Link "DBX Options"} DBX Options @{" DBX_DEBUGGING_INFO " Link "DBX Options"} DBX Options @{" DBX_FUNCTION_FIRST " Link "DBX Options"} DBX Options @{" DBX_LBRAC_FIRST " Link "DBX Options"} DBX Options @{" DBX_MEMPARM_STABS_LETTER " Link "DBX Options"} DBX Options @{" DBX_NO_XREFS " Link "DBX Options"} DBX Options @{" DBX_OUTPUT_ENUM " Link "DBX Hooks"} DBX Hooks @{" DBX_OUTPUT_FUNCTION_END " Link "DBX Hooks"} DBX Hooks @{" DBX_OUTPUT_LBRAC " Link "DBX Hooks"} DBX Hooks @{" DBX_OUTPUT_MAIN_SOURCE_DIRECTORY " Link "File Names and DBX"} File Names and DBX @{" DBX_OUTPUT_MAIN_SOURCE_FILE_END " Link "File Names and DBX"} File Names and DBX @{" DBX_OUTPUT_MAIN_SOURCE_FILENAME " Link "File Names and DBX"} File Names and DBX @{" DBX_OUTPUT_RBRAC " Link "DBX Hooks"} DBX Hooks @{" DBX_OUTPUT_SOURCE_FILENAME " Link "File Names and DBX"} File Names and DBX @{" DBX_OUTPUT_STANDARD_TYPES " Link "DBX Hooks"} DBX Hooks @{" DBX_REGISTER_NUMBER " Link "All Debuggers"} All Debuggers @{" DBX_REGPARM_STABS_CODE " Link "DBX Options"} DBX Options @{" DBX_REGPARM_STABS_LETTER " Link "DBX Options"} DBX Options @{" DBX_STATIC_CONST_VAR_CODE " Link "DBX Options"} DBX Options @{" DBX_STATIC_STAB_DATA_SECTION " Link "DBX Options"} DBX Options @{" DBX_TYPE_DECL_STABS_CODE " Link "DBX Options"} DBX Options @{" DBX_WORKING_DIRECTORY " Link "File Names and DBX"} File Names and DBX @{" DCmode " Link "Machine Modes"} Machine Modes @{" De Morgan's law " Link "Insn Canonicalizations"} Insn Canonicalizations @{" dead code " Link "Passes"} Passes @{" dead_or_set_p " Link "Peephole Definitions"} Peephole Definitions @{" deallocating variable length arrays " Link "Variable Length"} Variable Length @{" death notes " Link "Obsolete Register Macros"} Obsolete Register Macros @{" debug_rtx " Link "Bug Reporting"} Bug Reporting @{" DEBUG_SYMS_TEXT " Link "DBX Options"} DBX Options @{" DEBUGGER_ARG_OFFSET " Link "All Debuggers"} All Debuggers @{" DEBUGGER_AUTO_OFFSET " Link "All Debuggers"} All Debuggers @{" debugging information generation " Link "Passes"} Passes @{" debugging information options " Link "Debugging Options"} Debugging Options @{" debugging, 88k OCS " Link "M88K Options"} M88K Options @{" declaration scope " Link "Incompatibilities"} Incompatibilities @{" declarations inside expressions " Link "Statement Exprs"} Statement Exprs @{" declarations, RTL " Link "RTL Declarations"} RTL Declarations @{" declaring attributes of functions " Link "Function Attributes"} Function Attributes @{" declaring static data in C++ " Link "Static Definitions"} Static Definitions @{" default implementation, signature member function " Link "C++ Signatures"} C++ Signatures @{" DEFAULT_CALLER_SAVES " Link "Caller Saves"} Caller Saves @{" DEFAULT_GDB_EXTENSIONS " Link "DBX Options"} DBX Options @{" DEFAULT_MAIN_RETURN " Link "Misc"} Misc @{" DEFAULT_PCC_STRUCT_RETURN " Link "Aggregate Return"} Aggregate Return @{" DEFAULT_SHORT_ENUMS " Link "Type Layout"} Type Layout @{" DEFAULT_SIGNED_CHAR " Link "Type Layout"} Type Layout @{" define_asm_attributes " Link "Tagging Insns"} Tagging Insns @{" define_attr " Link "Defining Attributes"} Defining Attributes @{" define_delay " Link "Delay Slots"} Delay Slots @{" define_expand " Link "Expander Definitions"} Expander Definitions @{" define_function_unit " Link "Function Units"} Function Units @{" define_insn " Link "Patterns"} Patterns @{" define_insn example " Link "Example"} Example @{" define_peephole " Link "Expander Definitions"} Expander Definitions @{" define_split " Link "Insn Splitting"} Insn Splitting @{" defining attributes and their values " Link "Defining Attributes"} Defining Attributes @{" defining jump instruction patterns " Link "Jump Patterns"} Jump Patterns @{" defining peephole optimizers " Link "Peephole Definitions"} Peephole Definitions @{" defining RTL sequences for code generation " Link "Expander Definitions"} Expander Definitions @{" defining static data in C++ " Link "Static Definitions"} Static Definitions @{" delay slots, defining " Link "Delay Slots"} Delay Slots @{" DELAY_SLOTS_FOR_EPILOGUE " Link "Function Entry"} Function Entry @{" delayed branch scheduling " Link "Passes"} Passes @{" dependencies for make as output " Link "Environment Variables"} Environment Variables @{" dependencies, make " Link "Preprocessor Options"} Preprocessor Options @{" DEPENDENCIES_OUTPUT " Link "Environment Variables"} Environment Variables @{" Dependent Patterns " Link "Dependent Patterns"} Dependent Patterns @{" destructor function attribute " Link "Function Attributes"} Function Attributes @{" destructors vs goto " Link "Destructors and Goto"} Destructors and Goto @{" destructors, output of " Link "Initialization"} Initialization @{" detecting -traditional " Link "C Dialect Options"} C Dialect Options @{" DFmode " Link "Machine Modes"} Machine Modes @{" dialect options " Link "C Dialect Options"} C Dialect Options @{" digits in constraint " Link "Simple Constraints"} Simple Constraints @{" DImode " Link "Machine Modes"} Machine Modes @{" DIR_SEPARATOR " Link "Config"} Config @{" directory options " Link "Directory Options"} Directory Options @{" disabling certain registers " Link "Register Basics"} Register Basics @{" dispatch table " Link "Dispatch Tables"} Dispatch Tables @{" div " Link "Arithmetic"} Arithmetic @{" div and attributes " Link "Expressions"} Expressions @{" DIVDI3_LIBCALL " Link "Library Calls"} Library Calls @{" divide instruction, 88k " Link "M88K Options"} M88K Options @{" division " Link "Arithmetic"} Arithmetic @{" divM3 instruction pattern " Link "Standard Names"} Standard Names @{" divmodM4 instruction pattern " Link "Standard Names"} Standard Names @{" DIVSI3_LIBCALL " Link "Library Calls"} Library Calls @{" dollar signs in identifier names " Link "Dollar Signs"} Dollar Signs @{" DOLLARS_IN_IDENTIFIERS " Link "Misc"} Misc @{" DONE " Link "Expander Definitions"} Expander Definitions @{" DONT_DECLARE_SYS_SIGLIST " Link "Config"} Config @{" DONT_REDUCE_ADDR " Link "Costs"} Costs @{" double-word arithmetic " Link "Long Long"} Long Long @{" DOUBLE_TYPE_SIZE " Link "Type Layout"} Type Layout @{" downward funargs " Link "Nested Functions"} Nested Functions @{" driver " Link "Driver"} Driver @{" DW bit (29k) " Link "AMD29K Options"} AMD29K Options @{" DWARF_DEBUGGING_INFO " Link "SDB and DWARF"} SDB and DWARF @{" DYNAMIC_CHAIN_ADDRESS " Link "Frame Layout"} Frame Layout @{" E in constraint " Link "Simple Constraints"} Simple Constraints @{" EASY_DIV_EXPR " Link "Misc"} Misc @{" EDOM, implicit usage " Link "Library Calls"} Library Calls @{" ELIGIBLE_FOR_EPILOGUE_DELAY " Link "Function Entry"} Function Entry @{" ELIMINABLE_REGS " Link "Elimination"} Elimination @{" empty constraints " Link "No Constraints"} No Constraints @{" EMPTY_FIELD_BOUNDARY " Link "Storage Layout"} Storage Layout @{" ENCODE_SECTION_INFO " Link "Sections"} Sections @{" ENCODE_SECTION_INFO and address validation " Link "Addressing Modes"} Addressing Modes @{" ENCODE_SECTION_INFO usage " Link "Instruction Output"} Instruction Output @{" ENDFILE_SPEC " Link "Driver"} Driver @{" endianness " Link "Portability"} Portability @{" enum machine_mode " Link "Machine Modes"} Machine Modes @{" enum reg_class " Link "Register Classes"} Register Classes @{" enumeration clash warnings " Link "Warning Options"} Warning Options @{" environment variables " Link "Environment Variables"} Environment Variables @{" epilogue " Link "Function Entry"} Function Entry @{" eq " Link "Comparisons"} Comparisons @{" eq and attributes " Link "Expressions"} Expressions @{" eq_attr " Link "Expressions"} Expressions @{" equal " Link "Comparisons"} Comparisons @{" errno, implicit usage " Link "Library Calls"} Library Calls @{" error messages " Link "Warnings and Errors"} Warnings and Errors @{" escape sequences, traditional " Link "C Dialect Options"} C Dialect Options @{" exclamation point " Link "Multi-Alternative"} Multi-Alternative @{" exclusive-or, bitwise " Link "Arithmetic"} Arithmetic @{" EXECUTABLE_SUFFIX " Link "Config"} Config @{" exit " Link "C Dialect Options"} C Dialect Options @{" exit status and VMS " Link "VMS Misc"} VMS Misc @{" EXIT_BODY " Link "Misc"} Misc @{" EXIT_IGNORE_STACK " Link "Function Entry"} Function Entry @{" EXPAND_BUILTIN_SAVEREGS " Link "Varargs"} Varargs @{" expander definitions " Link "Expander Definitions"} Expander Definitions @{" explicit register variables " Link "Explicit Reg Vars"} Explicit Reg Vars @{" expr_list " Link "Insns"} Insns @{" expression codes " Link "RTL Objects"} RTL Objects @{" expressions containing statements " Link "Statement Exprs"} Statement Exprs @{" expressions, compound, as lvalues " Link "Lvalues"} Lvalues @{" expressions, conditional, as lvalues " Link "Lvalues"} Lvalues @{" expressions, constructor " Link "Constructors"} Constructors @{" extended asm " Link "Extended Asm"} Extended Asm @{" extendMN instruction pattern " Link "Standard Names"} Standard Names @{" extensible constraints " Link "Simple Constraints"} Simple Constraints @{" extensions, ?: <1> " Link "Lvalues"} Lvalues @{" extensions, ?: " Link "Conditionals"} Conditionals @{" extensions, C language " Link "C Extensions"} C Extensions @{" extensions, C++ language " Link "C++ Extensions"} C++ Extensions @{" extern int target_flags " Link "Run-time Target"} Run-time Target @{" external declaration scope " Link "Incompatibilities"} Incompatibilities @{" EXTRA_CC_MODES " Link "Condition Code"} Condition Code @{" EXTRA_CC_NAMES " Link "Condition Code"} Condition Code @{" EXTRA_CONSTRAINT " Link "Register Classes"} Register Classes @{" EXTRA_SECTION_FUNCTIONS " Link "Sections"} Sections @{" EXTRA_SECTIONS " Link "Sections"} Sections @{" extv instruction pattern " Link "Standard Names"} Standard Names @{" extzv instruction pattern " Link "Standard Names"} Standard Names @{" F in constraint " Link "Simple Constraints"} Simple Constraints @{" fabs " Link "C Dialect Options"} C Dialect Options @{" FAIL " Link "Expander Definitions"} Expander Definitions @{" fatal signal " Link "Bug Criteria"} Bug Criteria @{" FATAL_EXIT_CODE " Link "Config"} Config @{" features, optional, in system conventions " Link "Run-time Target"} Run-time Target @{" ffs <1> " Link "Arithmetic"} Arithmetic @{" ffs " Link "C Dialect Options"} C Dialect Options @{" ffsM2 instruction pattern " Link "Standard Names"} Standard Names @{" file name suffix " Link "Overall Options"} Overall Options @{" file names " Link "Link Options"} Link Options @{" files and passes of the compiler " Link "Passes"} Passes @{" final pass " Link "Passes"} Passes @{" FINAL_PRESCAN_INSN " Link "Instruction Output"} Instruction Output @{" FINAL_REG_PARM_STACK_SPACE " Link "Stack Arguments"} Stack Arguments @{" final_scan_insn " Link "Function Entry"} Function Entry @{" final_sequence " Link "Instruction Output"} Instruction Output @{" FINALIZE_PIC " Link "PIC"} PIC @{" FIRST_INSN_ADDRESS " Link "Insn Lengths"} Insn Lengths @{" FIRST_PARM_OFFSET " Link "Frame Layout"} Frame Layout @{" FIRST_PARM_OFFSET and virtual registers " Link "Regs and Memory"} Regs and Memory @{" FIRST_PSEUDO_REGISTER " Link "Register Basics"} Register Basics @{" FIRST_STACK_REG " Link "Stack Registers"} Stack Registers @{" FIRST_VIRTUAL_REGISTER " Link "Regs and Memory"} Regs and Memory @{" fix " Link "Conversions"} Conversions @{" fix_truncMN2 instruction pattern " Link "Standard Names"} Standard Names @{" fixed register " Link "Register Basics"} Register Basics @{" FIXED_REGISTERS " Link "Register Basics"} Register Basics @{" fixed_regs " Link "Register Basics"} Register Basics @{" fixMN2 instruction pattern " Link "Standard Names"} Standard Names @{" FIXUNS_TRUNC_LIKE_FIX_TRUNC " Link "Misc"} Misc @{" fixuns_truncMN2 instruction pattern " Link "Standard Names"} Standard Names @{" fixunsMN2 instruction pattern " Link "Standard Names"} Standard Names @{" flags in RTL expression " Link "Flags"} Flags @{" float " Link "Conversions"} Conversions @{" float as function value type " Link "Incompatibilities"} Incompatibilities @{" FLOAT_ARG_TYPE " Link "Library Calls"} Library Calls @{" float_extend " Link "Conversions"} Conversions @{" FLOAT_STORE_FLAG_VALUE " Link "Misc"} Misc @{" float_truncate " Link "Conversions"} Conversions @{" FLOAT_TYPE_SIZE " Link "Type Layout"} Type Layout @{" FLOAT_VALUE_TYPE " Link "Library Calls"} Library Calls @{" FLOAT_WORDS_BIG_ENDIAN " Link "Storage Layout"} Storage Layout @{" FLOATIFY " Link "Library Calls"} Library Calls @{" floating point and cross compilation " Link "Cross-compilation"} Cross-compilation @{" floatMN2 instruction pattern " Link "Standard Names"} Standard Names @{" floatunsMN2 instruction pattern " Link "Standard Names"} Standard Names @{" force_reg " Link "Standard Names"} Standard Names @{" format function attribute " Link "Function Attributes"} Function Attributes @{" forwarding calls " Link "Constructing Calls"} Constructing Calls @{" frame layout " Link "Frame Layout"} Frame Layout @{" FRAME_GROWS_DOWNWARD " Link "Frame Layout"} Frame Layout @{" FRAME_GROWS_DOWNWARD and virtual registers " Link "Regs and Memory"} Regs and Memory @{" frame_pointer_needed " Link "Function Entry"} Function Entry @{" FRAME_POINTER_REGNUM " Link "Frame Registers"} Frame Registers @{" FRAME_POINTER_REGNUM and virtual registers " Link "Regs and Memory"} Regs and Memory @{" FRAME_POINTER_REQUIRED " Link "Elimination"} Elimination @{" frame_pointer_rtx " Link "Frame Registers"} Frame Registers @{" fscanf, and constant strings " Link "Incompatibilities"} Incompatibilities @{" ftruncM2 instruction pattern " Link "Standard Names"} Standard Names @{" function attributes " Link "Function Attributes"} Function Attributes @{" function call conventions " Link "Interface"} Interface @{" function entry and exit " Link "Function Entry"} Function Entry @{" function pointers, arithmetic " Link "Pointer Arith"} Pointer Arith @{" function prototype declarations " Link "Function Prototypes"} Function Prototypes @{" function units, for scheduling " Link "Function Units"} Function Units @{" function, size of pointer to " Link "Pointer Arith"} Pointer Arith @{" function-call insns " Link "Calls"} Calls @{" FUNCTION_ARG " Link "Register Arguments"} Register Arguments @{" FUNCTION_ARG_ADVANCE " Link "Register Arguments"} Register Arguments @{" FUNCTION_ARG_BOUNDARY " Link "Register Arguments"} Register Arguments @{" FUNCTION_ARG_CALLEE_COPIES " Link "Register Arguments"} Register Arguments @{" FUNCTION_ARG_PADDING " Link "Register Arguments"} Register Arguments @{" FUNCTION_ARG_PARTIAL_NREGS " Link "Register Arguments"} Register Arguments @{" FUNCTION_ARG_PASS_BY_REFERENCE " Link "Register Arguments"} Register Arguments @{" FUNCTION_ARG_REGNO_P " Link "Register Arguments"} Register Arguments @{" FUNCTION_BLOCK_PROFILER " Link "Profiling"} Profiling @{" FUNCTION_BOUNDARY " Link "Storage Layout"} Storage Layout @{" FUNCTION_CONVERSION_BUG " Link "Config"} Config @{" FUNCTION_EPILOGUE " Link "Function Entry"} Function Entry @{" FUNCTION_EPILOGUE and trampolines " Link "Trampolines"} Trampolines @{" FUNCTION_INCOMING_ARG " Link "Register Arguments"} Register Arguments @{" FUNCTION_MODE " Link "Misc"} Misc @{" FUNCTION_OUTGOING_VALUE " Link "Scalar Return"} Scalar Return @{" FUNCTION_PROFILER " Link "Profiling"} Profiling @{" FUNCTION_PROLOGUE " Link "Function Entry"} Function Entry @{" FUNCTION_PROLOGUE and trampolines " Link "Trampolines"} Trampolines @{" FUNCTION_VALUE " Link "Scalar Return"} Scalar Return @{" FUNCTION_VALUE_REGNO_P " Link "Scalar Return"} Scalar Return @{" functions in arbitrary sections " Link "Function Attributes"} Function Attributes @{" functions that are passed arguments in registers on the 386 " Link "Function Attributes"} Function Attributes @{" functions that do not pop the argument stack on the 386 " Link "Function Attributes"} Function Attributes @{" functions that do pop the argument stack on the 386 " Link "Function Attributes"} Function Attributes @{" functions that have no side effects " Link "Function Attributes"} Function Attributes @{" functions that never return " Link "Function Attributes"} Function Attributes @{" functions that pop the argument stack on the 386 " Link "Function Attributes"} Function Attributes @{" functions with printf or scanf style arguments " Link "Function Attributes"} Function Attributes @{" functions, leaf " Link "Leaf Functions"} Leaf Functions @{" G in constraint " Link "Simple Constraints"} Simple Constraints @{" g in constraint " Link "Simple Constraints"} Simple Constraints @{" G++ " Link "G++ and GCC"} G++ and GCC @{" g++ " Link "Invoking G++"} Invoking G++ @{" g++ 1.XX " Link "Invoking G++"} Invoking G++ @{" g++ older version " Link "Invoking G++"} Invoking G++ @{" g++, separate compiler " Link "Invoking G++"} Invoking G++ @{" GCC " Link "G++ and GCC"} G++ and GCC @{" GCC_EXEC_PREFIX " Link "Environment Variables"} Environment Variables @{" ge " Link "Comparisons"} Comparisons @{" ge and attributes " Link "Expressions"} Expressions @{" GEN_ERRNO_RTX " Link "Library Calls"} Library Calls @{" gencodes " Link "Passes"} Passes @{" genconfig " Link "Passes"} Passes @{" general_operand " Link "RTL Template"} RTL Template @{" GENERAL_REGS " Link "Register Classes"} Register Classes @{" generalized lvalues " Link "Lvalues"} Lvalues @{" generating assembler output " Link "Output Statement"} Output Statement @{" generating insns " Link "RTL Template"} RTL Template @{" genflags " Link "Passes"} Passes @{" genflags, crash on Sun 4 " Link "Installation Problems"} Installation Problems @{" get_attr " Link "Expressions"} Expressions @{" get_attr_length " Link "Insn Lengths"} Insn Lengths @{" GET_CLASS_NARROWEST_MODE " Link "Machine Modes"} Machine Modes @{" GET_CODE " Link "RTL Objects"} RTL Objects @{" get_frame_size " Link "Elimination"} Elimination @{" get_insns " Link "Insns"} Insns @{" get_last_insn " Link "Insns"} Insns @{" GET_MODE " Link "Machine Modes"} Machine Modes @{" GET_MODE_ALIGNMENT " Link "Machine Modes"} Machine Modes @{" GET_MODE_BITSIZE " Link "Machine Modes"} Machine Modes @{" GET_MODE_CLASS " Link "Machine Modes"} Machine Modes @{" GET_MODE_MASK " Link "Machine Modes"} Machine Modes @{" GET_MODE_NAME " Link "Machine Modes"} Machine Modes @{" GET_MODE_NUNITS " Link "Machine Modes"} Machine Modes @{" GET_MODE_SIZE " Link "Machine Modes"} Machine Modes @{" GET_MODE_UNIT_SIZE " Link "Machine Modes"} Machine Modes @{" GET_MODE_WIDER_MODE " Link "Machine Modes"} Machine Modes @{" GET_RTX_CLASS " Link "Accessors"} Accessors @{" GET_RTX_FORMAT " Link "Accessors"} Accessors @{" GET_RTX_LENGTH " Link "Accessors"} Accessors @{" geu " Link "Comparisons"} Comparisons @{" geu and attributes " Link "Expressions"} Expressions @{" global offset table " Link "Code Gen Options"} Code Gen Options @{" global register after longjmp " Link "Global Reg Vars"} Global Reg Vars @{" global register allocation " Link "Passes"} Passes @{" global register variables " Link "Global Reg Vars"} Global Reg Vars @{" GLOBALDEF " Link "Global Declarations"} Global Declarations @{" GLOBALREF " Link "Global Declarations"} Global Declarations @{" GLOBALVALUEDEF " Link "Global Declarations"} Global Declarations @{" GLOBALVALUEREF " Link "Global Declarations"} Global Declarations @{" GNU CC and portability " Link "Portability"} Portability @{" GNU CC command options " Link "Invoking GCC"} Invoking GCC @{" GO_IF_LEGITIMATE_ADDRESS " Link "Addressing Modes"} Addressing Modes @{" GO_IF_MODE_DEPENDENT_ADDRESS " Link "Addressing Modes"} Addressing Modes @{" goto in C++ " Link "Destructors and Goto"} Destructors and Goto @{" goto with computed label " Link "Labels as Values"} Labels as Values @{" gp-relative references (MIPS) " Link "MIPS Options"} MIPS Options @{" gprof " Link "Debugging Options"} Debugging Options @{" greater than " Link "Comparisons"} Comparisons @{" grouping options " Link "Invoking GCC"} Invoking GCC @{" gt " Link "Comparisons"} Comparisons @{" gt and attributes " Link "Expressions"} Expressions @{" gtu " Link "Comparisons"} Comparisons @{" gtu and attributes " Link "Expressions"} Expressions @{" H in constraint " Link "Simple Constraints"} Simple Constraints @{" HANDLE_PRAGMA " Link "Misc"} Misc @{" hard registers " Link "Regs and Memory"} Regs and Memory @{" HARD_FRAME_POINTER_REGNUM " Link "Frame Registers"} Frame Registers @{" HARD_REGNO_MODE_OK " Link "Values in Registers"} Values in Registers @{" HARD_REGNO_NREGS " Link "Values in Registers"} Values in Registers @{" hardware models and configurations, specifying " Link "Submodel Options"} Submodel Options @{" HAS_INIT_SECTION " Link "Macros for Initialization"} Macros for Initialization @{" HAVE_ATEXIT " Link "Misc"} Misc @{" HAVE_POST_DECREMENT " Link "Addressing Modes"} Addressing Modes @{" HAVE_POST_INCREMENT " Link "Addressing Modes"} Addressing Modes @{" HAVE_PRE_DECREMENT " Link "Addressing Modes"} Addressing Modes @{" HAVE_PRE_INCREMENT " Link "Addressing Modes"} Addressing Modes @{" HAVE_PUTENV " Link "Config"} Config @{" HAVE_VPRINTF " Link "Config"} Config @{" header files and VMS " Link "Include Files and VMS"} Include Files and VMS @{" high " Link "Constants"} Constants @{" HImode " Link "Machine Modes"} Machine Modes @{" HImode, in insn " Link "Insns"} Insns @{" host makefile fragment " Link "Host Fragment"} Host Fragment @{" HOST_BITS_PER_CHAR " Link "Config"} Config @{" HOST_BITS_PER_INT " Link "Config"} Config @{" HOST_BITS_PER_LONG " Link "Config"} Config @{" HOST_BITS_PER_SHORT " Link "Config"} Config @{" HOST_FLOAT_FORMAT " Link "Config"} Config @{" HOST_FLOAT_WORDS_BIG_ENDIAN " Link "Config"} Config @{" HOST_WORDS_BIG_ENDIAN " Link "Config"} Config @{" HPPA Options " Link "HPPA Options"} HPPA Options @{" I in constraint " Link "Simple Constraints"} Simple Constraints @{" i in constraint " Link "Simple Constraints"} Simple Constraints @{" i386 Options " Link "i386 Options"} i386 Options @{" IBM RS/6000 and PowerPC Options " Link "RS-6000 and PowerPC Options"} RS-6000 and PowerPC Options @{" IBM RT options " Link "RT Options"} RT Options @{" IBM RT PC " Link "Interoperation"} Interoperation @{" identifier names, dollar signs in " Link "Dollar Signs"} Dollar Signs @{" identifiers, names in assembler code " Link "Asm Labels"} Asm Labels @{" identifying source, compiler (88k) " Link "M88K Options"} M88K Options @{" IEEE_FLOAT_FORMAT " Link "Storage Layout"} Storage Layout @{" if_then_else " Link "Comparisons"} Comparisons @{" if_then_else and attributes " Link "Expressions"} Expressions @{" if_then_else usage " Link "Side Effects"} Side Effects @{" immediate_operand " Link "RTL Template"} RTL Template @{" IMMEDIATE_PREFIX " Link "Instruction Output"} Instruction Output @{" implicit argument: return value " Link "Naming Results"} Naming Results @{" IMPLICIT_FIX_EXPR " Link "Misc"} Misc @{" implied #pragma implementation " Link "C++ Interface"} C++ Interface @{" in_data " Link "Sections"} Sections @{" in_struct " Link "Flags"} Flags @{" in_struct, in code_label " Link "Flags"} Flags @{" in_struct, in insn " Link "Flags"} Flags @{" in_struct, in label_ref " Link "Flags"} Flags @{" in_struct, in mem " Link "Flags"} Flags @{" in_struct, in reg " Link "Flags"} Flags @{" in_struct, in subreg " Link "Flags"} Flags @{" in_text " Link "Sections"} Sections @{" include files and VMS " Link "Include Files and VMS"} Include Files and VMS @{" INCLUDE_DEFAULTS " Link "Driver"} Driver @{" inclusive-or, bitwise " Link "Arithmetic"} Arithmetic @{" INCOMING_REGNO " Link "Register Basics"} Register Basics @{" incompatibilities of GNU CC " Link "Incompatibilities"} Incompatibilities @{" increment operators " Link "Bug Criteria"} Bug Criteria @{" INDEX_REG_CLASS " Link "Register Classes"} Register Classes @{" indirect_jump instruction pattern " Link "Standard Names"} Standard Names @{" INIT_CUMULATIVE_ARGS " Link "Register Arguments"} Register Arguments @{" INIT_CUMULATIVE_INCOMING_ARGS " Link "Register Arguments"} Register Arguments @{" INIT_ENVIRONMENT " Link "Driver"} Driver @{" INIT_SECTION_ASM_OP <1> " Link "Sections"} Sections @{" INIT_SECTION_ASM_OP " Link "Macros for Initialization"} Macros for Initialization @{" INIT_TARGET_OPTABS " Link "Library Calls"} Library Calls @{" INITIAL_ELIMINATION_OFFSET " Link "Elimination"} Elimination @{" INITIAL_FRAME_POINTER_OFFSET " Link "Elimination"} Elimination @{" initialization routines " Link "Initialization"} Initialization @{" initializations in expressions " Link "Constructors"} Constructors @{" INITIALIZE_TRAMPOLINE " Link "Trampolines"} Trampolines @{" initializers with labeled elements " Link "Labeled Elements"} Labeled Elements @{" initializers, non-constant " Link "Initializers"} Initializers @{" inline automatic for C++ member fns " Link "Inline"} Inline @{" inline functions " Link "Inline"} Inline @{" inline functions, omission of " Link "Inline"} Inline @{" inline, automatic " Link "Passes"} Passes @{" inlining and C++ pragmas " Link "C++ Interface"} C++ Interface @{" insn " Link "Insns"} Insns @{" insn and /i " Link "Flags"} Flags @{" insn and /s " Link "Flags"} Flags @{" insn and /u " Link "Flags"} Flags @{" insn attributes " Link "Insn Attributes"} Insn Attributes @{" insn canonicalization " Link "Insn Canonicalizations"} Insn Canonicalizations @{" insn lengths, computing " Link "Insn Lengths"} Insn Lengths @{" insn splitting " Link "Insn Splitting"} Insn Splitting @{" insn-attr.h " Link "Defining Attributes"} Defining Attributes @{" INSN_ANNULLED_BRANCH_P " Link "Flags"} Flags @{" INSN_CACHE_DEPTH " Link "Trampolines"} Trampolines @{" INSN_CACHE_LINE_WIDTH " Link "Trampolines"} Trampolines @{" INSN_CACHE_SIZE " Link "Trampolines"} Trampolines @{" INSN_CLOBBERS_REGNO_P " Link "Obsolete Register Macros"} Obsolete Register Macros @{" INSN_CODE " Link "Insns"} Insns @{" INSN_DELETED_P " Link "Flags"} Flags @{" INSN_FROM_TARGET_P " Link "Flags"} Flags @{" insn_list " Link "Insns"} Insns @{" INSN_REFERENCES_ARE_DELAYED " Link "Misc"} Misc @{" INSN_SETS_ARE_DELAYED " Link "Misc"} Misc @{" INSN_UID " Link "Insns"} Insns @{" insns " Link "Insns"} Insns @{" insns, generating " Link "RTL Template"} RTL Template @{" insns, recognizing " Link "RTL Template"} RTL Template @{" INSTALL " Link "Host Fragment"} Host Fragment @{" installation trouble " Link "Trouble"} Trouble @{" installing GNU CC " Link "Installation"} Installation @{" installing GNU CC on the Sun " Link "Sun Install"} Sun Install @{" installing GNU CC on VMS " Link "VMS Install"} VMS Install @{" instruction attributes " Link "Insn Attributes"} Insn Attributes @{" instruction combination " Link "Passes"} Passes @{" instruction patterns " Link "Patterns"} Patterns @{" instruction recognizer " Link "Passes"} Passes @{" instruction scheduling " Link "Passes"} Passes @{" instruction splitting " Link "Insn Splitting"} Insn Splitting @{" insv instruction pattern " Link "Standard Names"} Standard Names @{" INT_TYPE_SIZE " Link "Type Layout"} Type Layout @{" INTEGRATE_THRESHOLD " Link "Misc"} Misc @{" integrated " Link "Flags"} Flags @{" integrated, in insn " Link "Flags"} Flags @{" integrated, in reg " Link "Flags"} Flags @{" integrating function code " Link "Inline"} Inline @{" Intel 386 Options " Link "i386 Options"} i386 Options @{" Interdependence of Patterns " Link "Dependent Patterns"} Dependent Patterns @{" interface and implementation headers, C++ " Link "C++ Interface"} C++ Interface @{" interfacing to GNU CC output " Link "Interface"} Interface @{" intermediate C version, nonexistent " Link "G++ and GCC"} G++ and GCC @{" INTIFY " Link "Library Calls"} Library Calls @{" introduction " Link "Top"} Top @{" invalid assembly code " Link "Bug Criteria"} Bug Criteria @{" invalid input " Link "Bug Criteria"} Bug Criteria @{" INVOKE__main " Link "Macros for Initialization"} Macros for Initialization @{" invoking g++ " Link "Invoking G++"} Invoking G++ @{" ior " Link "Arithmetic"} Arithmetic @{" ior and attributes " Link "Expressions"} Expressions @{" ior, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @{" iorM3 instruction pattern " Link "Standard Names"} Standard Names @{" IS_ASM_LOGICAL_LINE_SEPARATOR " Link "Data Output"} Data Output @{" isinf " Link "Cross-compilation"} Cross-compilation @{" isnan " Link "Cross-compilation"} Cross-compilation @{" jump instruction patterns " Link "Jump Patterns"} Jump Patterns @{" jump instructions and set " Link "Side Effects"} Side Effects @{" jump optimization " Link "Passes"} Passes @{" jump threading " Link "Passes"} Passes @{" jump_insn " Link "Insns"} Insns @{" JUMP_LABEL " Link "Insns"} Insns @{" JUMP_TABLES_IN_TEXT_SECTION " Link "Sections"} Sections @{" kernel and user registers (29k) " Link "AMD29K Options"} AMD29K Options @{" keywords, alternate " Link "Alternate Keywords"} Alternate Keywords @{" known causes of trouble " Link "Trouble"} Trouble @{" LABEL_NUSES " Link "Insns"} Insns @{" LABEL_OUTSIDE_LOOP_P " Link "Flags"} Flags @{" LABEL_PRESERVE_P " Link "Flags"} Flags @{" label_ref " Link "Constants"} Constants @{" label_ref and /s " Link "Flags"} Flags @{" label_ref, RTL sharing " Link "Sharing"} Sharing @{" labeled elements in initializers " Link "Labeled Elements"} Labeled Elements @{" labels as values " Link "Labels as Values"} Labels as Values @{" labs " Link "C Dialect Options"} C Dialect Options @{" language dialect options " Link "C Dialect Options"} C Dialect Options @{" large bit shifts (88k) " Link "M88K Options"} M88K Options @{" large return values " Link "Aggregate Return"} Aggregate Return @{" LAST_STACK_REG " Link "Stack Registers"} Stack Registers @{" LAST_VIRTUAL_REGISTER " Link "Regs and Memory"} Regs and Memory @{" LD_FINI_SWITCH " Link "Macros for Initialization"} Macros for Initialization @{" LD_INIT_SWITCH " Link "Macros for Initialization"} Macros for Initialization @{" LDD_SUFFIX " Link "Macros for Initialization"} Macros for Initialization @{" ldexp " Link "Cross-compilation"} Cross-compilation @{" le " Link "Comparisons"} Comparisons @{" le and attributes " Link "Expressions"} Expressions @{" leaf functions " Link "Leaf Functions"} Leaf Functions @{" leaf_function " Link "Leaf Functions"} Leaf Functions @{" leaf_function_p " Link "Standard Names"} Standard Names @{" LEAF_REG_REMAP " Link "Leaf Functions"} Leaf Functions @{" LEAF_REGISTERS " Link "Leaf Functions"} Leaf Functions @{" left rotate " Link "Arithmetic"} Arithmetic @{" left shift " Link "Arithmetic"} Arithmetic @{" LEGITIMATE_CONSTANT_P " Link "Addressing Modes"} Addressing Modes @{" LEGITIMATE_PIC_OPERAND_P " Link "PIC"} PIC @{" LEGITIMIZE_ADDRESS " Link "Addressing Modes"} Addressing Modes @{" length-zero arrays " Link "Zero Length"} Zero Length @{" less than " Link "Comparisons"} Comparisons @{" less than or equal " Link "Comparisons"} Comparisons @{" leu " Link "Comparisons"} Comparisons @{" leu and attributes " Link "Expressions"} Expressions @{" LIB2FUNCS_EXTRA " Link "Target Fragment"} Target Fragment @{" LIB_SPEC " Link "Driver"} Driver @{" LIBCALL_VALUE " Link "Scalar Return"} Scalar Return @{" libgcc.a " Link "Library Calls"} Library Calls @{" LIBGCC1 " Link "Target Fragment"} Target Fragment @{" LIBGCC2_CFLAGS " Link "Target Fragment"} Target Fragment @{" LIBGCC2_WORDS_BIG_ENDIAN " Link "Storage Layout"} Storage Layout @{" LIBGCC_NEEDS_DOUBLE " Link "Library Calls"} Library Calls @{" LIBGCC_SPEC " Link "Driver"} Driver @{" Libraries " Link "Link Options"} Link Options @{" library subroutine names " Link "Library Calls"} Library Calls @{" LIBRARY_PATH " Link "Environment Variables"} Environment Variables @{" LIMIT_RELOAD_CLASS " Link "Register Classes"} Register Classes @{" link options " Link "Link Options"} Link Options @{" LINK_LIBGCC_SPECIAL " Link "Driver"} Driver @{" LINK_LIBGCC_SPECIAL_1 " Link "Driver"} Driver @{" LINK_SPEC " Link "Driver"} Driver @{" lo_sum " Link "Arithmetic"} Arithmetic @{" load address instruction " Link "Simple Constraints"} Simple Constraints @{" LOAD_EXTEND_OP " Link "Misc"} Misc @{" load_multiple instruction pattern " Link "Standard Names"} Standard Names @{" local labels " Link "Local Labels"} Local Labels @{" local register allocation " Link "Passes"} Passes @{" local variables in macros " Link "Naming Types"} Naming Types @{" local variables, specifying registers " Link "Local Reg Vars"} Local Reg Vars @{" LOCAL_INCLUDE_DIR " Link "Driver"} Driver @{" LOCAL_LABEL_PREFIX " Link "Instruction Output"} Instruction Output @{" LOG_LINKS " Link "Insns"} Insns @{" logical-and, bitwise " Link "Arithmetic"} Arithmetic @{" long long data types " Link "Long Long"} Long Long @{" LONG_DOUBLE_TYPE_SIZE " Link "Type Layout"} Type Layout @{" LONG_LONG_TYPE_SIZE " Link "Type Layout"} Type Layout @{" LONG_TYPE_SIZE " Link "Type Layout"} Type Layout @{" longjmp " Link "Global Reg Vars"} Global Reg Vars @{" longjmp and automatic variables <1> " Link "C Dialect Options"} C Dialect Options @{" longjmp and automatic variables " Link "Interface"} Interface @{" longjmp incompatibilities " Link "Incompatibilities"} Incompatibilities @{" longjmp warnings " Link "Warning Options"} Warning Options @{" LONGJMP_RESTORE_FROM_STACK " Link "Elimination"} Elimination @{" loop optimization " Link "Passes"} Passes @{" lshiftrt " Link "Arithmetic"} Arithmetic @{" lshiftrt and attributes " Link "Expressions"} Expressions @{" lshrM3 instruction pattern " Link "Standard Names"} Standard Names @{" lt " Link "Comparisons"} Comparisons @{" lt and attributes " Link "Expressions"} Expressions @{" ltu " Link "Comparisons"} Comparisons @{" lvalues, generalized " Link "Lvalues"} Lvalues @{" m in constraint " Link "Simple Constraints"} Simple Constraints @{" M680x0 options " Link "M680x0 Options"} M680x0 Options @{" M88k options " Link "M88K Options"} M88K Options @{" machine dependent options " Link "Submodel Options"} Submodel Options @{" machine description macros " Link "Target Macros"} Target Macros @{" machine descriptions " Link "Machine Desc"} Machine Desc @{" machine mode conversions " Link "Conversions"} Conversions @{" machine modes " Link "Machine Modes"} Machine Modes @{" machine specific constraints " Link "Machine Constraints"} Machine Constraints @{" MACHINE_DEPENDENT_REORG " Link "Misc"} Misc @{" macro with variable arguments " Link "Macro Varargs"} Macro Varargs @{" macros containing asm " Link "Extended Asm"} Extended Asm @{" macros, inline alternative " Link "Inline"} Inline @{" macros, local labels " Link "Local Labels"} Local Labels @{" macros, local variables in " Link "Naming Types"} Naming Types @{" macros, statements in expressions " Link "Statement Exprs"} Statement Exprs @{" macros, target description " Link "Target Macros"} Target Macros @{" macros, types of arguments " Link "Typeof"} Typeof @{" main and the exit status " Link "VMS Misc"} VMS Misc @{" make " Link "Preprocessor Options"} Preprocessor Options @{" make_safe_from " Link "Expander Definitions"} Expander Definitions @{" makefile fragment " Link "Fragments"} Fragments @{" match_dup " Link "RTL Template"} RTL Template @{" match_dup and attributes " Link "Insn Lengths"} Insn Lengths @{" match_op_dup " Link "RTL Template"} RTL Template @{" match_operand " Link "RTL Template"} RTL Template @{" match_operand and attributes " Link "Expressions"} Expressions @{" match_operator " Link "RTL Template"} RTL Template @{" match_par_dup " Link "RTL Template"} RTL Template @{" match_parallel " Link "RTL Template"} RTL Template @{" match_scratch " Link "RTL Template"} RTL Template @{" matching constraint " Link "Simple Constraints"} Simple Constraints @{" matching operands " Link "Output Template"} Output Template @{" math libraries " Link "Interface"} Interface @{" math, in RTL " Link "Arithmetic"} Arithmetic @{" MAX_BITS_PER_WORD " Link "Storage Layout"} Storage Layout @{" MAX_CHAR_TYPE_SIZE " Link "Type Layout"} Type Layout @{" MAX_FIXED_MODE_SIZE " Link "Storage Layout"} Storage Layout @{" MAX_INT_TYPE_SIZE " Link "Type Layout"} Type Layout @{" MAX_LONG_TYPE_SIZE " Link "Type Layout"} Type Layout @{" MAX_MOVE_MAX " Link "Misc"} Misc @{" MAX_OFILE_ALIGNMENT " Link "Storage Layout"} Storage Layout @{" MAX_REGS_PER_ADDRESS " Link "Addressing Modes"} Addressing Modes @{" MAX_WCHAR_TYPE_SIZE " Link "Type Layout"} Type Layout @{" maximum operator " Link "Min and Max"} Min and Max @{" maxM3 instruction pattern " Link "Standard Names"} Standard Names @{" MAYBE_REG_PARM_STACK_SPACE " Link "Stack Arguments"} Stack Arguments @{" mcount " Link "Profiling"} Profiling @{" MD_CALL_PROTOTYPES " Link "Config"} Config @{" MD_EXEC_PREFIX " Link "Driver"} Driver @{" MD_STARTFILE_PREFIX " Link "Driver"} Driver @{" MD_STARTFILE_PREFIX_1 " Link "Driver"} Driver @{" mem " Link "Regs and Memory"} Regs and Memory @{" mem and /s " Link "Flags"} Flags @{" mem and /u " Link "Flags"} Flags @{" mem and /v " Link "Flags"} Flags @{" mem, RTL sharing " Link "Sharing"} Sharing @{" MEM_IN_STRUCT_P " Link "Flags"} Flags @{" MEM_VOLATILE_P " Link "Flags"} Flags @{" member fns, automatically inline " Link "Inline"} Inline @{" memcmp " Link "C Dialect Options"} C Dialect Options @{" memcpy " Link "C Dialect Options"} C Dialect Options @{" memcpy, implicit usage " Link "Library Calls"} Library Calls @{" memory model (29k) " Link "AMD29K Options"} AMD29K Options @{" memory reference, nonoffsettable " Link "Simple Constraints"} Simple Constraints @{" memory references in constraints " Link "Simple Constraints"} Simple Constraints @{" MEMORY_MOVE_COST " Link "Costs"} Costs @{" memset, implicit usage " Link "Library Calls"} Library Calls @{" messages, warning " Link "Warning Options"} Warning Options @{" messages, warning and error " Link "Warnings and Errors"} Warnings and Errors @{" middle-operands, omitted " Link "Conditionals"} Conditionals @{" MIN_UNITS_PER_WORD " Link "Storage Layout"} Storage Layout @{" minimum operator " Link "Min and Max"} Min and Max @{" minM3 instruction pattern " Link "Standard Names"} Standard Names @{" minus " Link "Arithmetic"} Arithmetic @{" minus and attributes " Link "Expressions"} Expressions @{" minus, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @{" MIPS options " Link "MIPS Options"} MIPS Options @{" misunderstandings in C++ " Link "C++ Misunderstandings"} C++ Misunderstandings @{" mktemp, and constant strings " Link "Incompatibilities"} Incompatibilities @{" mod " Link "Arithmetic"} Arithmetic @{" mod and attributes " Link "Expressions"} Expressions @{" MODDI3_LIBCALL " Link "Library Calls"} Library Calls @{" mode attribute " Link "Variable Attributes"} Variable Attributes @{" mode classes " Link "Machine Modes"} Machine Modes @{" MODE_CC " Link "Machine Modes"} Machine Modes @{" MODE_COMPLEX_FLOAT " Link "Machine Modes"} Machine Modes @{" MODE_COMPLEX_INT " Link "Machine Modes"} Machine Modes @{" MODE_FLOAT " Link "Machine Modes"} Machine Modes @{" MODE_FUNCTION " Link "Machine Modes"} Machine Modes @{" MODE_INT " Link "Machine Modes"} Machine Modes @{" MODE_PARTIAL_INT " Link "Machine Modes"} Machine Modes @{" MODE_RANDOM " Link "Machine Modes"} Machine Modes @{" MODES_TIEABLE_P " Link "Values in Registers"} Values in Registers @{" modifiers in constraints " Link "Modifiers"} Modifiers @{" modM3 instruction pattern " Link "Standard Names"} Standard Names @{" MODSI3_LIBCALL " Link "Library Calls"} Library Calls @{" MOVE_MAX " Link "Misc"} Misc @{" MOVE_RATIO " Link "Costs"} Costs @{" movM instruction pattern " Link "Standard Names"} Standard Names @{" movMODEcc instruction pattern " Link "Standard Names"} Standard Names @{" movstrictM instruction pattern " Link "Standard Names"} Standard Names @{" movstrM instruction pattern " Link "Standard Names"} Standard Names @{" MULDI3_LIBCALL " Link "Library Calls"} Library Calls @{" mulhisi3 instruction pattern " Link "Standard Names"} Standard Names @{" mulM3 instruction pattern " Link "Standard Names"} Standard Names @{" mulqihi3 instruction pattern " Link "Standard Names"} Standard Names @{" MULSI3_LIBCALL " Link "Library Calls"} Library Calls @{" mulsidi3 instruction pattern " Link "Standard Names"} Standard Names @{" mult " Link "Arithmetic"} Arithmetic @{" mult and attributes " Link "Expressions"} Expressions @{" mult, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @{" MULTIBYTE_CHARS " Link "Config"} Config @{" MULTILIB_DEFAULTS " Link "Driver"} Driver @{" MULTILIB_DIRNAMES " Link "Target Fragment"} Target Fragment @{" MULTILIB_MATCHES " Link "Target Fragment"} Target Fragment @{" MULTILIB_OPTIONS " Link "Target Fragment"} Target Fragment @{" multiple alternative constraints " Link "Multi-Alternative"} Multi-Alternative @{" multiplication " Link "Arithmetic"} Arithmetic @{" multiprecision arithmetic " Link "Long Long"} Long Long @{" MUST_PASS_IN_STACK, and FUNCTION_ARG " Link "Register Arguments"} Register Arguments @{" n in constraint " Link "Simple Constraints"} Simple Constraints @{" N_REG_CLASSES " Link "Register Classes"} Register Classes @{" name augmentation " Link "VMS Misc"} VMS Misc @{" named patterns and conditions " Link "Patterns"} Patterns @{" named return value in C++ " Link "Naming Results"} Naming Results @{" names used in assembler code " Link "Asm Labels"} Asm Labels @{" names, pattern " Link "Standard Names"} Standard Names @{" naming convention, implementation headers " Link "C++ Interface"} C++ Interface @{" naming types " Link "Naming Types"} Naming Types @{" ne " Link "Comparisons"} Comparisons @{" ne and attributes " Link "Expressions"} Expressions @{" neg " Link "Arithmetic"} Arithmetic @{" neg and attributes " Link "Expressions"} Expressions @{" neg, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @{" negM2 instruction pattern " Link "Standard Names"} Standard Names @{" nested functions " Link "Nested Functions"} Nested Functions @{" nested functions, trampolines for " Link "Trampolines"} Trampolines @{" newline vs string constants " Link "C Dialect Options"} C Dialect Options @{" next_cc0_user " Link "Jump Patterns"} Jump Patterns @{" NEXT_INSN " Link "Insns"} Insns @{" NEXT_OBJC_RUNTIME " Link "Library Calls"} Library Calls @{" nil " Link "RTL Objects"} RTL Objects @{" no constraints " Link "No Constraints"} No Constraints @{" no-op move instructions " Link "Passes"} Passes @{" NO_BUILTIN_PTRDIFF_TYPE " Link "Driver"} Driver @{" NO_BUILTIN_SIZE_TYPE " Link "Driver"} Driver @{" NO_DOLLAR_IN_LABEL " Link "Misc"} Misc @{" NO_DOT_IN_LABEL " Link "Misc"} Misc @{" NO_FUNCTION_CSE " Link "Costs"} Costs @{" NO_IMPLICIT_EXTERN_C " Link "Misc"} Misc @{" NO_MD_PROTOTYPES " Link "Config"} Config @{" NO_RECURSIVE_FUNCTION_CSE " Link "Costs"} Costs @{" NO_REGS " Link "Register Classes"} Register Classes @{" NO_STAB_H " Link "Config"} Config @{" NO_SYS_SIGLIST " Link "Config"} Config @{" nocommon attribute " Link "Variable Attributes"} Variable Attributes @{" non-constant initializers " Link "Initializers"} Initializers @{" non-static inline function " Link "Inline"} Inline @{" NON_SAVING_SETJMP " Link "Register Basics"} Register Basics @{" nongcc_SI_type " Link "Library Calls"} Library Calls @{" nongcc_word_type " Link "Library Calls"} Library Calls @{" nonoffsettable memory reference " Link "Simple Constraints"} Simple Constraints @{" nop instruction pattern " Link "Standard Names"} Standard Names @{" noreturn function attribute " Link "Function Attributes"} Function Attributes @{" not " Link "Arithmetic"} Arithmetic @{" not and attributes " Link "Expressions"} Expressions @{" not equal " Link "Comparisons"} Comparisons @{" not using constraints " Link "No Constraints"} No Constraints @{" not, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @{" note " Link "Insns"} Insns @{" NOTE_INSN_BLOCK_BEG " Link "Insns"} Insns @{" NOTE_INSN_BLOCK_END " Link "Insns"} Insns @{" NOTE_INSN_DELETED " Link "Insns"} Insns @{" NOTE_INSN_FUNCTION_END " Link "Insns"} Insns @{" NOTE_INSN_LOOP_BEG " Link "Insns"} Insns @{" NOTE_INSN_LOOP_CONT " Link "Insns"} Insns @{" NOTE_INSN_LOOP_END " Link "Insns"} Insns @{" NOTE_INSN_LOOP_VTOP " Link "Insns"} Insns @{" NOTE_INSN_SETJMP " Link "Insns"} Insns @{" NOTE_LINE_NUMBER " Link "Insns"} Insns @{" NOTE_SOURCE_FILE " Link "Insns"} Insns @{" NOTICE_UPDATE_CC " Link "Condition Code"} Condition Code @{" NUM_MACHINE_MODES " Link "Machine Modes"} Machine Modes @{" o in constraint " Link "Simple Constraints"} Simple Constraints @{" OBJC_GEN_METHOD_LABEL " Link "Label Output"} Label Output @{" OBJC_INCLUDE_PATH " Link "Environment Variables"} Environment Variables @{" OBJC_INT_SELECTORS " Link "Type Layout"} Type Layout @{" OBJC_PROLOGUE " Link "File Framework"} File Framework @{" OBJC_SELECTORS_WITHOUT_LABELS " Link "Type Layout"} Type Layout @{" OBJECT_FORMAT_COFF " Link "Macros for Initialization"} Macros for Initialization @{" OBJECT_FORMAT_ROSE " Link "Macros for Initialization"} Macros for Initialization @{" OBJECT_SUFFIX " Link "Config"} Config @{" Objective C " Link "G++ and GCC"} G++ and GCC @{" OBSTACK_CHUNK_ALLOC " Link "Config"} Config @{" OBSTACK_CHUNK_FREE " Link "Config"} Config @{" OBSTACK_CHUNK_SIZE " Link "Config"} Config @{" obstack_free " Link "Configurations"} Configurations @{" OCS (88k) " Link "M88K Options"} M88K Options @{" offsettable address " Link "Simple Constraints"} Simple Constraints @{" old-style function definitions " Link "Function Prototypes"} Function Prototypes @{" OLDAR " Link "Host Fragment"} Host Fragment @{" OLDCC " Link "Host Fragment"} Host Fragment @{" omitted middle-operands " Link "Conditionals"} Conditionals @{" one_cmplM2 instruction pattern " Link "Standard Names"} Standard Names @{" ONLY_INT_FIELDS " Link "Config"} Config @{" open coding " Link "Inline"} Inline @{" operand access " Link "Accessors"} Accessors @{" operand constraints " Link "Constraints"} Constraints @{" operand substitution " Link "Output Template"} Output Template @{" operands " Link "Patterns"} Patterns @{" OPTIMIZATION_OPTIONS " Link "Run-time Target"} Run-time Target @{" optimize options " Link "Optimize Options"} Optimize Options @{" optional hardware or system features " Link "Run-time Target"} Run-time Target @{" options to control warnings " Link "Warning Options"} Warning Options @{" options, C++ " Link "C++ Dialect Options"} C++ Dialect Options @{" options, code generation " Link "Code Gen Options"} Code Gen Options @{" options, debugging " Link "Debugging Options"} Debugging Options @{" options, dialect " Link "C Dialect Options"} C Dialect Options @{" options, directory search " Link "Directory Options"} Directory Options @{" options, GNU CC command " Link "Invoking GCC"} Invoking GCC @{" options, grouping " Link "Invoking GCC"} Invoking GCC @{" options, linking " Link "Link Options"} Link Options @{" options, optimization " Link "Optimize Options"} Optimize Options @{" options, order " Link "Invoking GCC"} Invoking GCC @{" options, preprocessor " Link "Preprocessor Options"} Preprocessor Options @{" order of evaluation, side effects " Link "Non-bugs"} Non-bugs @{" order of options " Link "Invoking GCC"} Invoking GCC @{" order of register allocation " Link "Allocation Order"} Allocation Order @{" ORDER_REGS_FOR_LOCAL_ALLOC " Link "Allocation Order"} Allocation Order @{" Ordering of Patterns " Link "Pattern Ordering"} Pattern Ordering @{" other directory, compilation in " Link "Other Dir"} Other Dir @{" OUTGOING_REG_PARM_STACK_SPACE " Link "Stack Arguments"} Stack Arguments @{" OUTGOING_REGNO " Link "Register Basics"} Register Basics @{" output file option " Link "Overall Options"} Overall Options @{" output of assembler code " Link "File Framework"} File Framework @{" output statements " Link "Output Statement"} Output Statement @{" output templates " Link "Output Template"} Output Template @{" output_addr_const " Link "Data Output"} Data Output @{" output_asm_insn " Link "Output Statement"} Output Statement @{" overflow while constant folding " Link "Cross-compilation"} Cross-compilation @{" OVERLAPPING_REGNO_P " Link "Obsolete Register Macros"} Obsolete Register Macros @{" overloaded virtual fn, warning " Link "Warning Options"} Warning Options @{" OVERRIDE_OPTIONS " Link "Run-time Target"} Run-time Target @{" p in constraint " Link "Simple Constraints"} Simple Constraints @{" packed attribute " Link "Variable Attributes"} Variable Attributes @{" parallel " Link "Side Effects"} Side Effects @{" parameter forward declaration " Link "Variable Length"} Variable Length @{" parameters, miscellaneous " Link "Misc"} Misc @{" PARM_BOUNDARY " Link "Storage Layout"} Storage Layout @{" PARSE_LDD_OUTPUT " Link "Macros for Initialization"} Macros for Initialization @{" parser generator, Bison " Link "Installation"} Installation @{" parsing pass " Link "Passes"} Passes @{" passes and files of the compiler " Link "Passes"} Passes @{" passing arguments " Link "Interface"} Interface @{" PATH_SEPARATOR " Link "Config"} Config @{" PATTERN " Link "Insns"} Insns @{" pattern conditions " Link "Patterns"} Patterns @{" pattern names " Link "Standard Names"} Standard Names @{" Pattern Ordering " Link "Pattern Ordering"} Pattern Ordering @{" patterns " Link "Patterns"} Patterns @{" pc " Link "Regs and Memory"} Regs and Memory @{" pc and attributes " Link "Insn Lengths"} Insn Lengths @{" pc, RTL sharing " Link "Sharing"} Sharing @{" pc_rtx " Link "Regs and Memory"} Regs and Memory @{" PCC_BITFIELD_TYPE_MATTERS " Link "Storage Layout"} Storage Layout @{" PCC_STATIC_STRUCT_RETURN " Link "Aggregate Return"} Aggregate Return @{" PDImode " Link "Machine Modes"} Machine Modes @{" peephole optimization " Link "Passes"} Passes @{" peephole optimization, RTL representation " Link "Side Effects"} Side Effects @{" peephole optimizer definitions " Link "Peephole Definitions"} Peephole Definitions @{" percent sign " Link "Output Template"} Output Template @{" perform_... " Link "Library Calls"} Library Calls @{" PIC <1> " Link "Code Gen Options"} Code Gen Options @{" PIC " Link "PIC"} PIC @{" PIC_OFFSET_TABLE_REG_CALL_CLOBBERED " Link "PIC"} PIC @{" PIC_OFFSET_TABLE_REGNUM " Link "PIC"} PIC @{" plus " Link "Arithmetic"} Arithmetic @{" plus and attributes " Link "Expressions"} Expressions @{" plus, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @{" Pmode " Link "Misc"} Misc @{" pointer arguments " Link "Function Attributes"} Function Attributes @{" POINTER_SIZE " Link "Storage Layout"} Storage Layout @{" POINTERS_EXTEND_UNSIGNED " Link "Storage Layout"} Storage Layout @{" portability " Link "Portability"} Portability @{" portions of temporary objects, pointers to " Link "Temporaries"} Temporaries @{" position independent code " Link "PIC"} PIC @{" POSIX " Link "Config"} Config @{" post_dec " Link "Incdec"} Incdec @{" post_inc " Link "Incdec"} Incdec @{" pragma " Link "Misc"} Misc @{" pragma, reason for not using " Link "Function Attributes"} Function Attributes @{" pragmas in C++, effect on inlining " Link "C++ Interface"} C++ Interface @{" pragmas, interface and implementation " Link "C++ Interface"} C++ Interface @{" pre_dec " Link "Incdec"} Incdec @{" pre_inc " Link "Incdec"} Incdec @{" predefined macros " Link "Run-time Target"} Run-time Target @{" PREDICATE_CODES " Link "Misc"} Misc @{" PREFERRED_DEBUGGING_TYPE " Link "All Debuggers"} All Debuggers @{" PREFERRED_OUTPUT_RELOAD_CLASS " Link "Register Classes"} Register Classes @{" PREFERRED_RELOAD_CLASS " Link "Register Classes"} Register Classes @{" preprocessing numbers " Link "Incompatibilities"} Incompatibilities @{" preprocessing tokens " Link "Incompatibilities"} Incompatibilities @{" preprocessor options " Link "Preprocessor Options"} Preprocessor Options @{" PRESERVE_DEATH_INFO_REGNO_P " Link "Obsolete Register Macros"} Obsolete Register Macros @{" prev_active_insn " Link "Peephole Definitions"} Peephole Definitions @{" prev_cc0_setter " Link "Jump Patterns"} Jump Patterns @{" PREV_INSN " Link "Insns"} Insns @{" PRINT_OPERAND " Link "Instruction Output"} Instruction Output @{" PRINT_OPERAND_ADDRESS " Link "Instruction Output"} Instruction Output @{" PRINT_OPERAND_PUNCT_VALID_P " Link "Instruction Output"} Instruction Output @{" processor selection (29k) " Link "AMD29K Options"} AMD29K Options @{" product " Link "Arithmetic"} Arithmetic @{" prof " Link "Debugging Options"} Debugging Options @{" PROFILE_BEFORE_PROLOGUE " Link "Profiling"} Profiling @{" profiling, code generation " Link "Profiling"} Profiling @{" program counter " Link "Regs and Memory"} Regs and Memory @{" prologue " Link "Function Entry"} Function Entry @{" PROMOTE_FOR_CALL_ONLY " Link "Storage Layout"} Storage Layout @{" PROMOTE_FUNCTION_ARGS " Link "Storage Layout"} Storage Layout @{" PROMOTE_FUNCTION_RETURN " Link "Storage Layout"} Storage Layout @{" PROMOTE_MODE " Link "Storage Layout"} Storage Layout @{" PROMOTE_PROTOTYPES " Link "Stack Arguments"} Stack Arguments @{" promotion of formal parameters " Link "Function Prototypes"} Function Prototypes @{" pseudo registers " Link "Regs and Memory"} Regs and Memory @{" PSImode " Link "Machine Modes"} Machine Modes @{" PTRDIFF_TYPE " Link "Type Layout"} Type Layout @{" push address instruction " Link "Simple Constraints"} Simple Constraints @{" PUSH_ROUNDING " Link "Stack Arguments"} Stack Arguments @{" PUSH_ROUNDING, interaction with STACK_BOUNDARY " Link "Storage Layout"} Storage Layout @{" PUT_CODE " Link "RTL Objects"} RTL Objects @{" PUT_MODE " Link "Machine Modes"} Machine Modes @{" PUT_REG_NOTE_KIND " Link "Insns"} Insns @{" PUT_SDB_... " Link "SDB and DWARF"} SDB and DWARF @{" putenv " Link "Config"} Config @{" Q, in constraint " Link "Simple Constraints"} Simple Constraints @{" QImode " Link "Machine Modes"} Machine Modes @{" QImode, in insn " Link "Insns"} Insns @{" qsort, and global register variables " Link "Global Reg Vars"} Global Reg Vars @{" question mark " Link "Multi-Alternative"} Multi-Alternative @{" quotient " Link "Arithmetic"} Arithmetic @{" r in constraint " Link "Simple Constraints"} Simple Constraints @{" r0-relative references (88k) " Link "M88K Options"} M88K Options @{" ranges in case statements " Link "Case Ranges"} Case Ranges @{" read-only strings " Link "Incompatibilities"} Incompatibilities @{" READONLY_DATA_SECTION " Link "Sections"} Sections @{" REAL_ARITHMETIC " Link "Cross-compilation"} Cross-compilation @{" REAL_INFINITY " Link "Cross-compilation"} Cross-compilation @{" REAL_NM_FILE_NAME " Link "Macros for Initialization"} Macros for Initialization @{" REAL_VALUE_ATOF " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_FIX " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_FROM_INT " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_ISINF " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_ISNAN " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_LDEXP " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_NEGATE " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_RNDZINT " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_TO_DECIMAL " Link "Data Output"} Data Output @{" REAL_VALUE_TO_INT " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_TO_TARGET_DOUBLE " Link "Data Output"} Data Output @{" REAL_VALUE_TO_TARGET_LONG_DOUBLE " Link "Data Output"} Data Output @{" REAL_VALUE_TO_TARGET_SINGLE " Link "Data Output"} Data Output @{" REAL_VALUE_TRUNCATE " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_TYPE " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_UNSIGNED_FIX " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUE_UNSIGNED_RNDZINT " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUES_EQUAL " Link "Cross-compilation"} Cross-compilation @{" REAL_VALUES_LESS " Link "Cross-compilation"} Cross-compilation @{" recog_operand " Link "Instruction Output"} Instruction Output @{" recognizing insns " Link "RTL Template"} RTL Template @{" reg " Link "Regs and Memory"} Regs and Memory @{" reg and /i " Link "Flags"} Flags @{" reg and /s " Link "Flags"} Flags @{" reg and /u " Link "Flags"} Flags @{" reg and /v " Link "Flags"} Flags @{" reg, RTL sharing " Link "Sharing"} Sharing @{" REG_ALLOC_ORDER " Link "Allocation Order"} Allocation Order @{" REG_CC_SETTER " Link "Insns"} Insns @{" REG_CC_USER " Link "Insns"} Insns @{" REG_CLASS_CONTENTS " Link "Register Classes"} Register Classes @{" REG_CLASS_FROM_LETTER " Link "Register Classes"} Register Classes @{" REG_CLASS_NAMES " Link "Register Classes"} Register Classes @{" REG_DEAD " Link "Insns"} Insns @{" REG_DEP_ANTI " Link "Insns"} Insns @{" REG_DEP_OUTPUT " Link "Insns"} Insns @{" REG_EQUAL " Link "Insns"} Insns @{" REG_EQUIV " Link "Insns"} Insns @{" REG_FUNCTION_VALUE_P " Link "Flags"} Flags @{" REG_INC " Link "Insns"} Insns @{" REG_LABEL " Link "Insns"} Insns @{" REG_LIBCALL " Link "Insns"} Insns @{" REG_LOOP_TEST_P " Link "Flags"} Flags @{" reg_names " Link "Instruction Output"} Instruction Output @{" REG_NO_CONFLICT " Link "Insns"} Insns @{" REG_NONNEG " Link "Insns"} Insns @{" REG_NOTE_KIND " Link "Insns"} Insns @{" REG_NOTES " Link "Insns"} Insns @{" REG_OK_FOR_BASE_P " Link "Addressing Modes"} Addressing Modes @{" REG_OK_FOR_INDEX_P " Link "Addressing Modes"} Addressing Modes @{" REG_OK_STRICT " Link "Addressing Modes"} Addressing Modes @{" REG_PARM_STACK_SPACE " Link "Stack Arguments"} Stack Arguments @{" REG_PARM_STACK_SPACE, and FUNCTION_ARG " Link "Register Arguments"} Register Arguments @{" REG_RETVAL " Link "Insns"} Insns @{" REG_UNUSED " Link "Insns"} Insns @{" REG_USERVAR_P " Link "Flags"} Flags @{" REG_WAS_0 " Link "Insns"} Insns @{" register allocation " Link "Passes"} Passes @{" register allocation order " Link "Allocation Order"} Allocation Order @{" register allocation, stupid " Link "Passes"} Passes @{" register class definitions " Link "Register Classes"} Register Classes @{" register class preference constraints " Link "Class Preferences"} Class Preferences @{" register class preference pass " Link "Passes"} Passes @{" register pairs " Link "Values in Registers"} Values in Registers @{" register positions in frame (88k) " Link "M88K Options"} M88K Options @{" Register Transfer Language (RTL) " Link "RTL"} RTL @{" register usage " Link "Registers"} Registers @{" register use analysis " Link "Passes"} Passes @{" register variable after longjmp " Link "Global Reg Vars"} Global Reg Vars @{" register-to-stack conversion " Link "Passes"} Passes @{" REGISTER_MOVE_COST " Link "Costs"} Costs @{" REGISTER_NAMES " Link "Instruction Output"} Instruction Output @{" register_operand " Link "RTL Template"} RTL Template @{" REGISTER_PREFIX " Link "Instruction Output"} Instruction Output @{" registers " Link "Extended Asm"} Extended Asm @{" registers arguments " Link "Register Arguments"} Register Arguments @{" registers for local variables " Link "Local Reg Vars"} Local Reg Vars @{" registers in constraints " Link "Simple Constraints"} Simple Constraints @{" registers, global allocation " Link "Explicit Reg Vars"} Explicit Reg Vars @{" registers, global variables in " Link "Global Reg Vars"} Global Reg Vars @{" REGNO_OK_FOR_BASE_P " Link "Register Classes"} Register Classes @{" REGNO_OK_FOR_INDEX_P " Link "Register Classes"} Register Classes @{" REGNO_REG_CLASS " Link "Register Classes"} Register Classes @{" regs_ever_live " Link "Function Entry"} Function Entry @{" relative costs " Link "Costs"} Costs @{" RELATIVE_PREFIX_NOT_LINKDIR " Link "Driver"} Driver @{" reload pass " Link "Regs and Memory"} Regs and Memory @{" reload_completed " Link "Standard Names"} Standard Names @{" reload_in instruction pattern " Link "Standard Names"} Standard Names @{" reload_in_progress " Link "Standard Names"} Standard Names @{" reload_out instruction pattern " Link "Standard Names"} Standard Names @{" reloading " Link "Passes"} Passes @{" remainder " Link "Arithmetic"} Arithmetic @{" reordering, warning " Link "Warning Options"} Warning Options @{" reporting bugs " Link "Bugs"} Bugs @{" representation of RTL " Link "RTL"} RTL @{" rest argument (in macro) " Link "Macro Varargs"} Macro Varargs @{" rest_of_compilation " Link "Passes"} Passes @{" rest_of_decl_compilation " Link "Passes"} Passes @{" restore_stack_block instruction pattern " Link "Standard Names"} Standard Names @{" restore_stack_function instruction pattern " Link "Standard Names"} Standard Names @{" restore_stack_nonlocal instruction pattern " Link "Standard Names"} Standard Names @{" return " Link "Side Effects"} Side Effects @{" return instruction pattern " Link "Standard Names"} Standard Names @{" return value of main " Link "VMS Misc"} VMS Misc @{" return value, named, in C++ " Link "Naming Results"} Naming Results @{" return values in registers " Link "Scalar Return"} Scalar Return @{" return, in C++ function header " Link "Naming Results"} Naming Results @{" RETURN_ADDR_IN_PREVIOUS_FRAME " Link "Frame Layout"} Frame Layout @{" RETURN_ADDR_RTX " Link "Frame Layout"} Frame Layout @{" RETURN_IN_MEMORY " Link "Aggregate Return"} Aggregate Return @{" RETURN_POPS_ARGS " Link "Stack Arguments"} Stack Arguments @{" returning aggregate values " Link "Aggregate Return"} Aggregate Return @{" returning structures and unions " Link "Interface"} Interface @{" REVERSIBLE_CC_MODE " Link "Condition Code"} Condition Code @{" right rotate " Link "Arithmetic"} Arithmetic @{" right shift " Link "Arithmetic"} Arithmetic @{" rotate " Link "Arithmetic"} Arithmetic @{" rotatert " Link "Arithmetic"} Arithmetic @{" rotlM3 instruction pattern " Link "Standard Names"} Standard Names @{" rotrM3 instruction pattern " Link "Standard Names"} Standard Names @{" ROUND_TYPE_ALIGN " Link "Storage Layout"} Storage Layout @{" ROUND_TYPE_SIZE " Link "Storage Layout"} Storage Layout @{" RS/6000 and PowerPC Options " Link "RS-6000 and PowerPC Options"} RS-6000 and PowerPC Options @{" RT options " Link "RT Options"} RT Options @{" RT PC " Link "Interoperation"} Interoperation @{" RTL addition " Link "Arithmetic"} Arithmetic @{" RTL comparison " Link "Arithmetic"} Arithmetic @{" RTL comparison operations " Link "Comparisons"} Comparisons @{" RTL constant expression types " Link "Constants"} Constants @{" RTL constants " Link "Constants"} Constants @{" RTL declarations " Link "RTL Declarations"} RTL Declarations @{" RTL difference " Link "Arithmetic"} Arithmetic @{" RTL expression " Link "RTL Objects"} RTL Objects @{" RTL expressions for arithmetic " Link "Arithmetic"} Arithmetic @{" RTL format " Link "Accessors"} Accessors @{" RTL format characters " Link "Accessors"} Accessors @{" RTL function-call insns " Link "Calls"} Calls @{" RTL generation " Link "Passes"} Passes @{" RTL insn template " Link "RTL Template"} RTL Template @{" RTL integers " Link "RTL Objects"} RTL Objects @{" RTL memory expressions " Link "Regs and Memory"} Regs and Memory @{" RTL object types " Link "RTL Objects"} RTL Objects @{" RTL postdecrement " Link "Incdec"} Incdec @{" RTL postincrement " Link "Incdec"} Incdec @{" RTL predecrement " Link "Incdec"} Incdec @{" RTL preincrement " Link "Incdec"} Incdec @{" RTL register expressions " Link "Regs and Memory"} Regs and Memory @{" RTL representation " Link "RTL"} RTL @{" RTL side effect expressions " Link "Side Effects"} Side Effects @{" RTL strings " Link "RTL Objects"} RTL Objects @{" RTL structure sharing assumptions " Link "Sharing"} Sharing @{" RTL subtraction " Link "Arithmetic"} Arithmetic @{" RTL sum " Link "Arithmetic"} Arithmetic @{" RTL vectors " Link "RTL Objects"} RTL Objects @{" RTX (See RTL) " Link "RTL Objects"} RTL Objects @{" RTX_COSTS " Link "Costs"} Costs @{" RTX_INTEGRATED_P " Link "Flags"} Flags @{" RTX_UNCHANGING_P " Link "Flags"} Flags @{" run-time conventions " Link "Interface"} Interface @{" run-time options " Link "Code Gen Options"} Code Gen Options @{" run-time target specification " Link "Run-time Target"} Run-time Target @{" s in constraint " Link "Simple Constraints"} Simple Constraints @{" save_stack_block instruction pattern " Link "Standard Names"} Standard Names @{" save_stack_function instruction pattern " Link "Standard Names"} Standard Names @{" save_stack_nonlocal instruction pattern " Link "Standard Names"} Standard Names @{" saveable_obstack " Link "Addressing Modes"} Addressing Modes @{" scalars, returned as values " Link "Scalar Return"} Scalar Return @{" scanf, and constant strings " Link "Incompatibilities"} Incompatibilities @{" SCCS_DIRECTIVE " Link "Misc"} Misc @{" SCHED_GROUP_P " Link "Flags"} Flags @{" scheduling, delayed branch " Link "Passes"} Passes @{" scheduling, instruction " Link "Passes"} Passes @{" SCmode " Link "Machine Modes"} Machine Modes @{" sCOND instruction pattern " Link "Standard Names"} Standard Names @{" scope of a variable length array " Link "Variable Length"} Variable Length @{" scope of declaration " Link "Disappointments"} Disappointments @{" scope of external declarations " Link "Incompatibilities"} Incompatibilities @{" scratch " Link "Regs and Memory"} Regs and Memory @{" scratch operands " Link "Regs and Memory"} Regs and Memory @{" scratch, RTL sharing " Link "Sharing"} Sharing @{" SDB_ALLOW_FORWARD_REFERENCES " Link "SDB and DWARF"} SDB and DWARF @{" SDB_ALLOW_UNKNOWN_REFERENCES " Link "SDB and DWARF"} SDB and DWARF @{" SDB_DEBUGGING_INFO " Link "SDB and DWARF"} SDB and DWARF @{" SDB_DELIM " Link "SDB and DWARF"} SDB and DWARF @{" SDB_GENERATE_FAKE " Link "SDB and DWARF"} SDB and DWARF @{" search path " Link "Directory Options"} Directory Options @{" second include path " Link "Preprocessor Options"} Preprocessor Options @{" SECONDARY_INPUT_RELOAD_CLASS " Link "Register Classes"} Register Classes @{" SECONDARY_MEMORY_NEEDED " Link "Register Classes"} Register Classes @{" SECONDARY_MEMORY_NEEDED_MODE " Link "Register Classes"} Register Classes @{" SECONDARY_MEMORY_NEEDED_RTX " Link "Register Classes"} Register Classes @{" SECONDARY_OUTPUT_RELOAD_CLASS " Link "Register Classes"} Register Classes @{" SECONDARY_RELOAD_CLASS " Link "Register Classes"} Register Classes @{" section function attribute " Link "Function Attributes"} Function Attributes @{" section variable attribute " Link "Variable Attributes"} Variable Attributes @{" SELECT_CC_MODE " Link "Condition Code"} Condition Code @{" SELECT_RTX_SECTION " Link "Sections"} Sections @{" SELECT_SECTION " Link "Sections"} Sections @{" separate directory, compilation in " Link "Other Dir"} Other Dir @{" sequence " Link "Side Effects"} Side Effects @{" sequential consistency on 88k " Link "M88K Options"} M88K Options @{" set " Link "Side Effects"} Side Effects @{" set_attr " Link "Tagging Insns"} Tagging Insns @{" set_attr_alternative " Link "Tagging Insns"} Tagging Insns @{" SET_DEFAULT_TYPE_ATTRIBUTES " Link "Misc"} Misc @{" SET_DEST " Link "Side Effects"} Side Effects @{" SET_SRC " Link "Side Effects"} Side Effects @{" setjmp " Link "Global Reg Vars"} Global Reg Vars @{" setjmp incompatibilities " Link "Incompatibilities"} Incompatibilities @{" SETUP_FRAME_ADDRESSES " Link "Frame Layout"} Frame Layout @{" SETUP_INCOMING_VARARGS " Link "Varargs"} Varargs @{" SFmode " Link "Machine Modes"} Machine Modes @{" shared strings " Link "Incompatibilities"} Incompatibilities @{" shared VMS run time system " Link "VMS Misc"} VMS Misc @{" SHARED_SECTION_ASM_OP " Link "Sections"} Sections @{" sharing of RTL components " Link "Sharing"} Sharing @{" shift " Link "Arithmetic"} Arithmetic @{" SHIFT_COUNT_TRUNCATED " Link "Misc"} Misc @{" SHORT_TYPE_SIZE " Link "Type Layout"} Type Layout @{" side effect in ?: " Link "Conditionals"} Conditionals @{" side effects, macro argument " Link "Statement Exprs"} Statement Exprs @{" side effects, order of evaluation " Link "Non-bugs"} Non-bugs @{" sign_extend " Link "Conversions"} Conversions @{" sign_extract " Link "Bit Fields"} Bit Fields @{" sign_extract, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @{" signature " Link "C++ Signatures"} C++ Signatures @{" signature in C++, advantages " Link "C++ Signatures"} C++ Signatures @{" signature member function default implementation " Link "C++ Signatures"} C++ Signatures @{" signatures, C++ " Link "C++ Signatures"} C++ Signatures @{" signed division " Link "Arithmetic"} Arithmetic @{" signed maximum " Link "Arithmetic"} Arithmetic @{" signed minimum " Link "Arithmetic"} Arithmetic @{" SIGNED_CHAR_SPEC " Link "Driver"} Driver @{" SImode " Link "Machine Modes"} Machine Modes @{" simple constraints " Link "Simple Constraints"} Simple Constraints @{" simplifications, arithmetic " Link "Passes"} Passes @{" sin " Link "C Dialect Options"} C Dialect Options @{" SIZE_TYPE " Link "Type Layout"} Type Layout @{" sizeof " Link "Typeof"} Typeof @{" SLOW_BYTE_ACCESS " Link "Costs"} Costs @{" SLOW_UNALIGNED_ACCESS " Link "Costs"} Costs @{" SLOW_ZERO_EXTEND " Link "Costs"} Costs @{" SMALL_REGISTER_CLASSES " Link "Register Classes"} Register Classes @{" smaller data references (88k) " Link "M88K Options"} M88K Options @{" smaller data references (MIPS) " Link "MIPS Options"} MIPS Options @{" smax " Link "Arithmetic"} Arithmetic @{" smin " Link "Arithmetic"} Arithmetic @{" smulM3_highpart instruction pattern " Link "Standard Names"} Standard Names @{" SPARC options " Link "SPARC Options"} SPARC Options @{" specified registers " Link "Explicit Reg Vars"} Explicit Reg Vars @{" specifying compiler version and target machine " Link "Target Options"} Target Options @{" specifying hardware config " Link "Submodel Options"} Submodel Options @{" specifying machine version " Link "Target Options"} Target Options @{" specifying registers for local variables " Link "Local Reg Vars"} Local Reg Vars @{" speed of instructions " Link "Costs"} Costs @{" splitting instructions " Link "Insn Splitting"} Insn Splitting @{" sqrt <1> " Link "Arithmetic"} Arithmetic @{" sqrt " Link "C Dialect Options"} C Dialect Options @{" sqrtM2 instruction pattern " Link "Standard Names"} Standard Names @{" square root " Link "Arithmetic"} Arithmetic @{" sscanf, and constant strings " Link "Incompatibilities"} Incompatibilities @{" stack arguments " Link "Stack Arguments"} Stack Arguments @{" stack checks (29k) " Link "AMD29K Options"} AMD29K Options @{" stack frame layout " Link "Frame Layout"} Frame Layout @{" STACK_BOUNDARY " Link "Storage Layout"} Storage Layout @{" STACK_DYNAMIC_OFFSET " Link "Frame Layout"} Frame Layout @{" STACK_DYNAMIC_OFFSET and virtual registers " Link "Regs and Memory"} Regs and Memory @{" STACK_GROWS_DOWNWARD " Link "Frame Layout"} Frame Layout @{" STACK_PARMS_IN_REG_PARM_AREA " Link "Stack Arguments"} Stack Arguments @{" STACK_POINTER_OFFSET " Link "Frame Layout"} Frame Layout @{" STACK_POINTER_OFFSET and virtual registers " Link "Regs and Memory"} Regs and Memory @{" STACK_POINTER_REGNUM " Link "Frame Registers"} Frame Registers @{" STACK_POINTER_REGNUM and virtual registers " Link "Regs and Memory"} Regs and Memory @{" stack_pointer_rtx " Link "Frame Registers"} Frame Registers @{" STACK_REGS " Link "Stack Registers"} Stack Registers @{" stage1 " Link "Installation"} Installation @{" standard pattern names " Link "Standard Names"} Standard Names @{" STANDARD_EXEC_PREFIX " Link "Driver"} Driver @{" STANDARD_INCLUDE_DIR " Link "Driver"} Driver @{" STANDARD_STARTFILE_PREFIX " Link "Driver"} Driver @{" start files " Link "Tools and Libraries"} Tools and Libraries @{" STARTFILE_SPEC " Link "Driver"} Driver @{" STARTING_FRAME_OFFSET " Link "Frame Layout"} Frame Layout @{" STARTING_FRAME_OFFSET and virtual registers " Link "Regs and Memory"} Regs and Memory @{" statements inside expressions " Link "Statement Exprs"} Statement Exprs @{" static data in C++, declaring and defining " Link "Static Definitions"} Static Definitions @{" STATIC_CHAIN " Link "Frame Registers"} Frame Registers @{" STATIC_CHAIN_INCOMING " Link "Frame Registers"} Frame Registers @{" STATIC_CHAIN_INCOMING_REGNUM " Link "Frame Registers"} Frame Registers @{" STATIC_CHAIN_REGNUM " Link "Frame Registers"} Frame Registers @{" stdarg.h and register arguments " Link "Register Arguments"} Register Arguments @{" stdarg.h and RT PC " Link "RT Options"} RT Options @{" storage layout " Link "Storage Layout"} Storage Layout @{" STORE_FLAG_VALUE " Link "Misc"} Misc @{" store_multiple instruction pattern " Link "Standard Names"} Standard Names @{" storem bug (29k) " Link "AMD29K Options"} AMD29K Options @{" strcmp " Link "C Dialect Options"} C Dialect Options @{" strcpy <1> " Link "C Dialect Options"} C Dialect Options @{" strcpy " Link "Storage Layout"} Storage Layout @{" strength-reduction " Link "Passes"} Passes @{" STRICT_ALIGNMENT " Link "Storage Layout"} Storage Layout @{" STRICT_ARGUMENT_NAMING " Link "Varargs"} Varargs @{" strict_low_part " Link "RTL Declarations"} RTL Declarations @{" string constants " Link "Incompatibilities"} Incompatibilities @{" string constants vs newline " Link "C Dialect Options"} C Dialect Options @{" STRIP_NAME_ENCODING " Link "Sections"} Sections @{" strlen " Link "C Dialect Options"} C Dialect Options @{" strlenM instruction pattern " Link "Standard Names"} Standard Names @{" STRUCT_VALUE " Link "Aggregate Return"} Aggregate Return @{" STRUCT_VALUE_INCOMING " Link "Aggregate Return"} Aggregate Return @{" STRUCT_VALUE_INCOMING_REGNUM " Link "Aggregate Return"} Aggregate Return @{" STRUCT_VALUE_REGNUM " Link "Aggregate Return"} Aggregate Return @{" structure passing (88k) " Link "M88K Options"} M88K Options @{" structure value address " Link "Aggregate Return"} Aggregate Return @{" STRUCTURE_SIZE_BOUNDARY " Link "Storage Layout"} Storage Layout @{" structures " Link "Incompatibilities"} Incompatibilities @{" structures, constructor expression " Link "Constructors"} Constructors @{" structures, returning " Link "Interface"} Interface @{" stupid register allocation " Link "Passes"} Passes @{" subM3 instruction pattern " Link "Standard Names"} Standard Names @{" submodel options " Link "Submodel Options"} Submodel Options @{" subreg " Link "Regs and Memory"} Regs and Memory @{" subreg and /s " Link "Flags"} Flags @{" subreg and /u " Link "Flags"} Flags @{" subreg, in strict_low_part " Link "RTL Declarations"} RTL Declarations @{" subreg, special reload handling " Link "Regs and Memory"} Regs and Memory @{" SUBREG_PROMOTED_UNSIGNED_P " Link "Flags"} Flags @{" SUBREG_PROMOTED_VAR_P " Link "Flags"} Flags @{" SUBREG_REG " Link "Regs and Memory"} Regs and Memory @{" SUBREG_WORD " Link "Regs and Memory"} Regs and Memory @{" subscripting " Link "Subscripting"} Subscripting @{" subscripting and function values " Link "Subscripting"} Subscripting @{" subtype polymorphism, C++ " Link "C++ Signatures"} C++ Signatures @{" SUCCESS_EXIT_CODE " Link "Config"} Config @{" suffixes for C++ source " Link "Invoking G++"} Invoking G++ @{" Sun installation " Link "Sun Install"} Sun Install @{" SUPPORTS_WEAK " Link "Label Output"} Label Output @{" suppressing warnings " Link "Warning Options"} Warning Options @{" surprises in C++ " Link "C++ Misunderstandings"} C++ Misunderstandings @{" SVr4 " Link "M88K Options"} M88K Options @{" SWITCH_TAKES_ARG " Link "Driver"} Driver @{" SWITCHES_NEED_SPACES " Link "Driver"} Driver @{" symbol_ref " Link "Constants"} Constants @{" symbol_ref and /u " Link "Flags"} Flags @{" symbol_ref and /v " Link "Flags"} Flags @{" symbol_ref, RTL sharing " Link "Sharing"} Sharing @{" SYMBOL_REF_FLAG " Link "Flags"} Flags @{" SYMBOL_REF_FLAG, in ENCODE_SECTION_INFO " Link "Sections"} Sections @{" SYMBOL_REF_USED " Link "Flags"} Flags @{" symbolic label " Link "Sharing"} Sharing @{" syntax checking " Link "Warning Options"} Warning Options @{" synthesized methods, warning " Link "Warning Options"} Warning Options @{" sys_siglist " Link "Config"} Config @{" SYSTEM_INCLUDE_DIR " Link "Driver"} Driver @{" t-TARGET " Link "Target Fragment"} Target Fragment @{" tablejump instruction pattern " Link "Standard Names"} Standard Names @{" tagging insns " Link "Tagging Insns"} Tagging Insns @{" tail recursion optimization " Link "Passes"} Passes @{" target description macros " Link "Target Macros"} Target Macros @{" target machine, specifying " Link "Target Options"} Target Options @{" target makefile fragment " Link "Target Fragment"} Target Fragment @{" target options " Link "Target Options"} Target Options @{" target specifications " Link "Run-time Target"} Run-time Target @{" target-parameter-dependent code " Link "Passes"} Passes @{" TARGET_BELL " Link "Type Layout"} Type Layout @{" TARGET_BS " Link "Type Layout"} Type Layout @{" TARGET_CR " Link "Type Layout"} Type Layout @{" TARGET_EDOM " Link "Library Calls"} Library Calls @{" TARGET_FF " Link "Type Layout"} Type Layout @{" TARGET_FLOAT_FORMAT " Link "Storage Layout"} Storage Layout @{" TARGET_MEM_FUNCTIONS " Link "Library Calls"} Library Calls @{" TARGET_NEWLINE " Link "Type Layout"} Type Layout @{" TARGET_OPTIONS " Link "Run-time Target"} Run-time Target @{" TARGET_SWITCHES " Link "Run-time Target"} Run-time Target @{" TARGET_TAB " Link "Type Layout"} Type Layout @{" TARGET_VERSION " Link "Run-time Target"} Run-time Target @{" TARGET_VT " Link "Type Layout"} Type Layout @{" TCmode " Link "Machine Modes"} Machine Modes @{" tcov " Link "Debugging Options"} Debugging Options @{" template debugging " Link "Warning Options"} Warning Options @{" template instantiation " Link "Template Instantiation"} Template Instantiation @{" temporaries, lifetime of " Link "Temporaries"} Temporaries @{" termination routines " Link "Initialization"} Initialization @{" text_section " Link "Sections"} Sections @{" TEXT_SECTION_ASM_OP " Link "Sections"} Sections @{" TFmode " Link "Machine Modes"} Machine Modes @{" thunks " Link "Nested Functions"} Nested Functions @{" TImode " Link "Machine Modes"} Machine Modes @{" tm.h macros " Link "Target Macros"} Target Macros @{" TMPDIR " Link "Environment Variables"} Environment Variables @{" top level of compiler " Link "Passes"} Passes @{" traditional C language " Link "C Dialect Options"} C Dialect Options @{" TRADITIONAL_RETURN_FLOAT " Link "Scalar Return"} Scalar Return @{" TRAMPOLINE_ALIGNMENT " Link "Trampolines"} Trampolines @{" TRAMPOLINE_SECTION " Link "Trampolines"} Trampolines @{" TRAMPOLINE_SIZE " Link "Trampolines"} Trampolines @{" TRAMPOLINE_TEMPLATE " Link "Trampolines"} Trampolines @{" trampolines for nested functions " Link "Trampolines"} Trampolines @{" TRANSFER_FROM_TRAMPOLINE " Link "Trampolines"} Trampolines @{" TRULY_NOOP_TRUNCATION " Link "Misc"} Misc @{" truncate " Link "Conversions"} Conversions @{" truncMN instruction pattern " Link "Standard Names"} Standard Names @{" tstM instruction pattern " Link "Standard Names"} Standard Names @{" type abstraction, C++ " Link "C++ Signatures"} C++ Signatures @{" type alignment " Link "Alignment"} Alignment @{" type attributes " Link "Type Attributes"} Type Attributes @{" typedef names as function parameters " Link "Incompatibilities"} Incompatibilities @{" typeof " Link "Typeof"} Typeof @{" udiv " Link "Arithmetic"} Arithmetic @{" UDIVDI3_LIBCALL " Link "Library Calls"} Library Calls @{" udivM3 instruction pattern " Link "Standard Names"} Standard Names @{" udivmodM4 instruction pattern " Link "Standard Names"} Standard Names @{" UDIVSI3_LIBCALL " Link "Library Calls"} Library Calls @{" Ultrix calling convention " Link "Interoperation"} Interoperation @{" umax " Link "Arithmetic"} Arithmetic @{" umaxM3 instruction pattern " Link "Standard Names"} Standard Names @{" umin " Link "Arithmetic"} Arithmetic @{" uminM3 instruction pattern " Link "Standard Names"} Standard Names @{" umod " Link "Arithmetic"} Arithmetic @{" UMODDI3_LIBCALL " Link "Library Calls"} Library Calls @{" umodM3 instruction pattern " Link "Standard Names"} Standard Names @{" UMODSI3_LIBCALL " Link "Library Calls"} Library Calls @{" umulhisi3 instruction pattern " Link "Standard Names"} Standard Names @{" umulM3_highpart instruction pattern " Link "Standard Names"} Standard Names @{" umulqihi3 instruction pattern " Link "Standard Names"} Standard Names @{" umulsidi3 instruction pattern " Link "Standard Names"} Standard Names @{" unchanging " Link "Flags"} Flags @{" unchanging, in call_insn " Link "Flags"} Flags @{" unchanging, in insn " Link "Flags"} Flags @{" unchanging, in reg and mem " Link "Flags"} Flags @{" unchanging, in subreg " Link "Flags"} Flags @{" unchanging, in symbol_ref " Link "Flags"} Flags @{" undefined behavior " Link "Bug Criteria"} Bug Criteria @{" undefined function value " Link "Bug Criteria"} Bug Criteria @{" underscores in variables in macros " Link "Naming Types"} Naming Types @{" underscores, avoiding (88k) " Link "M88K Options"} M88K Options @{" union, casting to a " Link "Cast to Union"} Cast to Union @{" unions " Link "Incompatibilities"} Incompatibilities @{" unions, returning " Link "Interface"} Interface @{" UNITS_PER_WORD " Link "Storage Layout"} Storage Layout @{" UNKNOWN_FLOAT_FORMAT " Link "Storage Layout"} Storage Layout @{" unreachable code " Link "Passes"} Passes @{" unresolved references and -nodefaultlibs " Link "Link Options"} Link Options @{" unresolved references and -nostdlib " Link "Link Options"} Link Options @{" unshare_all_rtl " Link "Sharing"} Sharing @{" unsigned division " Link "Arithmetic"} Arithmetic @{" unsigned greater than " Link "Comparisons"} Comparisons @{" unsigned less than " Link "Comparisons"} Comparisons @{" unsigned minimum and maximum " Link "Arithmetic"} Arithmetic @{" unsigned_fix " Link "Conversions"} Conversions @{" unsigned_float " Link "Conversions"} Conversions @{" unspec " Link "Side Effects"} Side Effects @{" unspec_volatile " Link "Side Effects"} Side Effects @{" untyped_call instruction pattern " Link "Standard Names"} Standard Names @{" untyped_return instruction pattern " Link "Standard Names"} Standard Names @{" use " Link "Side Effects"} Side Effects @{" USE_C_ALLOCA " Link "Config"} Config @{" USE_PROTOTYPES " Link "Config"} Config @{" used " Link "Flags"} Flags @{" used, in symbol_ref " Link "Flags"} Flags @{" USER_LABEL_PREFIX " Link "Instruction Output"} Instruction Output @{" USG " Link "Config"} Config @{" V in constraint " Link "Simple Constraints"} Simple Constraints @{" VALID_MACHINE_DECL_ATTRIBUTE " Link "Misc"} Misc @{" VALID_MACHINE_TYPE_ATTRIBUTE " Link "Misc"} Misc @{" value after longjmp " Link "Global Reg Vars"} Global Reg Vars @{" values, returned by functions " Link "Scalar Return"} Scalar Return @{" varargs implementation " Link "Varargs"} Varargs @{" varargs.h and RT PC " Link "RT Options"} RT Options @{" variable alignment " Link "Alignment"} Alignment @{" variable attributes " Link "Variable Attributes"} Variable Attributes @{" variable number of arguments " Link "Macro Varargs"} Macro Varargs @{" variable-length array scope " Link "Variable Length"} Variable Length @{" variable-length arrays " Link "Variable Length"} Variable Length @{" variables in specified registers " Link "Explicit Reg Vars"} Explicit Reg Vars @{" variables, local, in macros " Link "Naming Types"} Naming Types @{" Vax calling convention " Link "Interoperation"} Interoperation @{" VAX options " Link "VAX Options"} VAX Options @{" VAX_FLOAT_FORMAT " Link "Storage Layout"} Storage Layout @{" VAXCRTL " Link "VMS Misc"} VMS Misc @{" VIRTUAL_INCOMING_ARGS_REGNUM " Link "Regs and Memory"} Regs and Memory @{" VIRTUAL_OUTGOING_ARGS_REGNUM " Link "Regs and Memory"} Regs and Memory @{" VIRTUAL_STACK_DYNAMIC_REGNUM " Link "Regs and Memory"} Regs and Memory @{" VIRTUAL_STACK_VARS_REGNUM " Link "Regs and Memory"} Regs and Memory @{" VMS " Link "Config"} Config @{" VMS and case sensitivity " Link "VMS Misc"} VMS Misc @{" VMS and include files " Link "Include Files and VMS"} Include Files and VMS @{" VMS installation " Link "VMS Install"} VMS Install @{" void pointers, arithmetic " Link "Pointer Arith"} Pointer Arith @{" void, size of pointer to " Link "Pointer Arith"} Pointer Arith @{" VOIDmode " Link "Machine Modes"} Machine Modes @{" volatil " Link "Flags"} Flags @{" volatil, in insn " Link "Flags"} Flags @{" volatil, in mem " Link "Flags"} Flags @{" volatil, in reg " Link "Flags"} Flags @{" volatil, in symbol_ref " Link "Flags"} Flags @{" volatile applied to function " Link "Function Attributes"} Function Attributes @{" volatile memory references " Link "Flags"} Flags @{" voting between constraint alternatives " Link "Class Preferences"} Class Preferences @{" vprintf " Link "Config"} Config @{" warning for enumeration conversions " Link "Warning Options"} Warning Options @{" warning for overloaded virtual fn " Link "Warning Options"} Warning Options @{" warning for reordering of member initializers " Link "Warning Options"} Warning Options @{" warning for synthesized methods " Link "Warning Options"} Warning Options @{" warning messages " Link "Warning Options"} Warning Options @{" warnings vs errors " Link "Warnings and Errors"} Warnings and Errors @{" WCHAR_TYPE " Link "Type Layout"} Type Layout @{" WCHAR_TYPE_SIZE " Link "Type Layout"} Type Layout @{" weak attribute " Link "Function Attributes"} Function Attributes @{" which_alternative " Link "Output Statement"} Output Statement @{" whitespace " Link "Incompatibilities"} Incompatibilities @{" word_mode " Link "Machine Modes"} Machine Modes @{" WORD_REGISTER_OPERATIONS " Link "Misc"} Misc @{" WORD_SWITCH_TAKES_ARG " Link "Driver"} Driver @{" WORDS_BIG_ENDIAN " Link "Storage Layout"} Storage Layout @{" WORDS_BIG_ENDIAN, effect on subreg " Link "Regs and Memory"} Regs and Memory @{" X in constraint " Link "Simple Constraints"} Simple Constraints @{" x-HOST " Link "Host Fragment"} Host Fragment @{" XCmode " Link "Machine Modes"} Machine Modes @{" XCOFF_DEBUGGING_INFO " Link "DBX Options"} DBX Options @{" XEXP " Link "Accessors"} Accessors @{" XFmode " Link "Machine Modes"} Machine Modes @{" XINT " Link "Accessors"} Accessors @{" xm-MACHINE.h " Link "Config"} Config @{" xor " Link "Arithmetic"} Arithmetic @{" xor, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @{" xorM3 instruction pattern " Link "Standard Names"} Standard Names @{" XSTR " Link "Accessors"} Accessors @{" XVEC " Link "Accessors"} Accessors @{" XVECEXP " Link "Accessors"} Accessors @{" XVECLEN " Link "Accessors"} Accessors @{" XWINT " Link "Accessors"} Accessors @{" zero division on 88k " Link "M88K Options"} M88K Options @{" zero-length arrays " Link "Zero Length"} Zero Length @{" zero_extend " Link "Conversions"} Conversions @{" zero_extendMN instruction pattern " Link "Standard Names"} Standard Names @{" zero_extract " Link "Bit Fields"} Bit Fields @{" zero_extract, canonicalization of " Link "Insn Canonicalizations"} Insn Canonicalizations @EndNode