Info file gcc.info, produced by Makeinfo, -*- Text -*- from input file gcc.texinfo. This file documents the use and the internals of the GNU compiler. Copyright (C) 1988, 1989 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'' 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'' 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.  File: gcc.info, Node: Installation, Next: Trouble, Prev: Options, Up: Top Installing GNU CC ***************** Here is the procedure for installing GNU CC on a Unix system. * Menu: * Other Dir:: Compiling in a separate directory (not where the source is). * Sun Install:: See below for installation on the Sun. * 3B1 Install:: See below for installation on the 3B1. * VMS Install:: See below for installation on VMS. * HPUX Install:: See below for installation on HPUX. 1. Edit `Makefile'. If you are using HPUX, or any form of system V, you must make a few changes described in comments at the beginning of the file. Genix requires changes also, and so does the Pyramid. 2. On a Sequent system, go to the Berkeley universe. 3. Choose configuration files. The easy way to do this is to run the command file `config.gcc' with a single argument, which specifies the type of machine (and in some cases which operating system). Here is a list of the possible arguments: `vax' Vaxes running BSD. `vms' Vaxes running VMS. `vax-sysv' Vaxes running system V. `i386-sysv' Intel 386 PCs running system V. `i386-sysv-gas' Intel 386 PCs running system V, using the GNU assembler and GNU linker. `sequent-i386' Sequent with Intel 386 processors. `i386-aix' Intel 386 PCs or PS/2s running AIX. `sun2' Sun 2 running system version 2 or 3. `sun3' Sun 3 running system version 2 or 3, with 68881. Note there we do not provide a configuration file to use an FPA by default, because programs that establish signal handlers for floating point traps inherently cannot work with the FPA. `sun3-nfp' Sun 3 running system version 2 or 3, without 68881. `sun4' Sun 4 running system version 2 or 3. *Note Incompatibilities::, for calling convention incompatibilities on the Sun 4 (sparc). `sun2-os4' Sun 2 running system version 4. `sun3-os4' Sun 3 running system version 4, with 68881. `sun3-nfp-os4' Sun 3 running system version 4, without 68881. `sun4-os4' Sun 4 running system version 4. *Note Incompatibilities::, for calling convention incompatibilities on the Sun 4 (sparc). `sun386' Sun 386 (``roadrunner''). `alliant' Alliant FX/8 computer. Note that the standard installed C compiler in Concentrix 5.0 has a bug which prevent it from compiling GNU CC correctly. You can patch the compiler bug as follows: cp /bin/pcc ./pcc adb -w ./pcc - << EOF 15f6?w 6610 EOF Then you must use the `-ip12' option when compiling GNU CC with the patched compiler, as shown here: make CC="./pcc -ip12" CFLAGS=-w Note also that Alliant's version of DBX does not manage to work with the output from GNU CC. `tahoe' The tahoe computer (running BSD, and using DBX). `decstation' The DEC 3100 Mips machine (``pmax''). Note that GNU CC cannot generate debugging information in the unusual format used on the Mips. `mips-sysv' The Mips computer, RS series, with the System V environment as default. Note that GNU CC cannot generate debugging information in the unusual format used on the Mips. `mips-bsd43' The Mips computer, RS series, with the BSD 4.3 environment as default. Note that GNU CC cannot generate debugging information in the unusual format used on the Mips. `mips' The Mips computer, M series. Note that GNU CC cannot generate debugging information in the unusual format used on the Mips. `iris' The Mips computer, as delivered by Iris. Note that GNU CC cannot generate debugging information in the unusual format used on the Mips. `convex-c1' Convex C1 computer. `convex-c2' Convex C2 computer. `pyramid' Pyramid computer. `hp9k320' HP 9000 series 300 using HPUX assembler. Note there is no support in GNU CC for HP's debugger; thus, `-g' is not available in this configuration. `hp9k320-gas' HP 9000 series 300 using GNU assembler, linker and debugger. This requires the HP-adapt package, which is available along with the GNU linker as part of the ``binutils'' distribution. This is on the GNU CC distribution tape. `hp9k320-old' HP 9000 series 300 using HPUX assembler, in operating system versions older than 6.5. Note there is no support in GNU CC for HP's debugger; thus, `-g' is not available in this configuration. `hp9k320-bsd' HP 9000 series 300 running BSD. `isi68' ISI 68000 or 68020 system with a 68881. `isi68-nfp' ISI 68000 or 68020 system without a 68881. `news800' Sony NEWS 68020 system. `next' NeXT system. `altos' Altos 3068. Note that you must use the GNU assembler, linker and debugger, with COFF-encapsulation. Also, you must fix a kernel bug. Details in the file `ALTOS-README'. `3b1' AT&T 3b1, a.k.a. 7300 PC. Note that special procedures are needed to compile GNU CC with this machine's standard C compiler, due to bugs in that compiler. *Note 3b1 Install::. You can bootstrap it more easily with previous versions of GNU CC if you have them. `3b1-gas' AT&T 3b1 using the GNU assembler. `sequent-ns32k' Sequent containing ns32000 processors. `encore' Encore ns32000 system. `genix' National Semiconductor ns32000 system. `88000' Motorola 88000 processor. This port is not finished. Here we spell out what files need to be set up: * Make a symbolic link named `config.h' to the top-level config file for the machine you are using (*note Config::.). This file is responsible for defining information about the host machine. It includes `tm.h'. The file is located in the subdirectory `config'. Its name should be `xm-MACHINE.h', with these exceptions: `xm-vms.h' for vaxen running VMS. `xm-vaxv.h' for vaxen running system V. `xm-i386v.h' for Intel 80386's running system V. `xm-sun386i.h' for Sun roadrunner running any version of the operating system. `xm-hp9k320.h' for the HP 9000 series 300. `xm-genix.h' for the ns32000 running Genix If your system does not support symbolic links, you might want to set up `config.h' to contain a `#include' command which refers to the appropriate file. * Make a symbolic link named `tm.h' to the machine-description macro file for your machine. It should be in the subdirectory `config' and its name should be `tm-MACHINE.h'. If your system is a 68000, don't use the file `tm-m68k.h' directly. Instead, use one of these files: `tm-sun3.h' for Sun 3 machines with 68881. `tm-sun3-nfp.h' for Sun 3 machines with no hardware floating point. `tm-sun3os3.h' for Sun 3 machines with 68881, running Sunos version 3. `tm-sun3os3nf.h' for Sun 3 machines with no hardware floating point, running Sunos version 3. `tm-sun2.h' for Sun 2 machines. `tm-3b1.h' for AT&T 3b1 (aka 7300 Unix PC). `tm-isi68.h' for Integrated Solutions systems. This file assumes you use the GNU assembler. `tm-isi68-nfp.h' for Integrated Solutions systems without a 68881. This file assumes you use the GNU assembler. `tm-news800.h' for Sony NEWS systems. `tm-hp9k320.h' for HPUX systems, if you are using GNU CC with the system's assembler and linker. `tm-hp9k320g.h' for HPUX systems, if you are using the GNU assembler, linker and other utilities. Not all of the pieces of GNU software needed for this mode of operation are as yet in distribution; full instructions will appear here in the future. For the vax, use `tm-vax.h' on BSD Unix, `tm-vaxv.h' on system V, or `tm-vms.h' on VMS. For the Motorola 88000, use `tm-m88k.h'. The support for the 88000 does not currently work; it requires extensive changes which we hope to reconcile in version 2. For the 80386, don't use `tm-i386.h' directly. Use `tm-i386v.h' if the target machine is running system V, `tm-i386gas.h' if it is running system V but you are using the GNU assembler and linker, `tm-seq386.h' for a Sequent 386 system, or `tm-compaq.h' for a Compaq, or `tm-sun386i.h' for a Sun 386 system. For the Mips computer, there are five choices: `tm-mips.h' for the M series, `tm-mips-bsd.h' for the RS series with BSD, `tm-mips-sysv.h' for the RS series with System V, `tm-iris.h' for the Iris version of the machine, and `tm-decstatn.h' for the Decstation. For the 32000, use `tm-sequent.h' if you are using a Sequent machine, or `tm-encore.h' for an Encore machine, or `tm-genix.h' if you are using Genix version 3; otherwise, perhaps `tm-ns32k.h' will work for you. Note that Genix has bugs in `alloca' and `malloc'; you must get the compiled versions of these from GNU Emacs and edit GNU CC's `Makefile' to use them. Note that Encore systems are supported only under BSD. For Sparc (Sun 4) machines, use `tm-sparc.h' with operating system version 4, and `tm-sun4os3.h' with system version 3. * Make a symbolic link named `md' to the machine description pattern file. It should be in the `config' subdirectory and its name should be `MACHINE.md'; but MACHINE is often not the same as the name used in the `tm.h' file because the `md' files are more general. * Make a symbolic link named `aux-output.c' to the output subroutine file for your machine. It should be in the `config' subdirectory and its name should be `out-MACHINE.c'. 4. Make sure the Bison parser generator is installed. (This is unnecessary if the Bison output files `c-parse.tab.c' and `cexp.c' are more recent than `c-parse.y' and `cexp.y' and you do not plan to change the `.y' files.) Bison versions older than Sept 8, 1988 will produce incorrect output for `c-parse.tab.c'. 5. Build the compiler. Just type `make' in the compiler directory. Ignore any warnings you may see about ``statement not reached'' in the `insn-emit.c'; they are normal. Any other compilation errors may represent bugs in the port to your machine or operating system, and should be investigated and reported (*note 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. 6. If you are using COFF-encapsulation, you must convert `gnulib' to a GNU-format library at this point. See the file `README-ENCAP' in the directory containing the GNU binary file utilities, for directions. 7. Move the first-stage object files and executables into a subdirectory with this command: make stage1 The files are moved into a subdirectory named `stage1'. Once installation is complete, you may wish to delete these files with `rm -r stage1'. 8. Recompile the compiler with itself, with this command: make CC=stage1/gcc CFLAGS="-g -O -Bstage1/" On a 68000 or 68020 system lacking floating point hardware, unless you have selected a `tm.h' file that expects by default that there is no such hardware, do this instead: make CC=stage1/gcc CFLAGS="-g -O -Bstage1/ -msoft-float" 9. If you wish to test the compiler by compiling it with itself one more time, do this (in C shell): make stage2 make CC=stage2/gcc CFLAGS="-g -O -Bstage2/" foreach file (*.o) cmp $file stage2/$file end Aside from the `-B' option, the options should be the same as when you made stage 2. The `foreach' command (written in C shell) will notify you if any of these stage 3 object files differs from those of stage 2. On BSD systems, 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 (*note Bugs::.). On systems that use COFF object files, bytes 5 to 8 will always be different, since it is a timestamp. On these systems, you can do the comparison as follows (in Bourne shell): for file in *.o; do echo $file tail +10 $file > foo1 tail +10 stage2/$file > foo2 cmp foo1 foo2 done 10. Install the compiler driver, the compiler's passes and run-time support. You can use the following command: make install This copies the files `cc1', `cpp' and `gnulib' to files `gcc-cc1', `gcc-cpp' and `gcc-gnulib' in directory `/usr/local/lib', which is where the compiler driver program looks for them. It also copies the driver program `gcc' into the directory `/usr/local/bin', so that it appears in typical execution search paths. *Warning: there is a bug in `alloca' in the Sun library. To avoid this bug, install the binaries of GNU CC that were compiled by GNU CC. They use `alloca' as a built-in function and never the one in the library.* *Warning: the GNU CPP may not work for `ioctl.h', `ttychars.h' and other system header files unless the `-traditional' option is used.* The bug is in the header files: at least on some machines, they rely on behavior that is incompatible with ANSI C. This behavior consists of substituting for macro argument names when they appear inside of character constants. The `-traditional' option tells GNU CC to behave the way these headers expect. Because of this problem, you might prefer to configure GNU CC to use the system's own C preprocessor. To do so, make the file `/usr/local/lib/gcc-cpp' a link to `/lib/cpp'. Alternatively, on Sun systems and 4.3BSD at least, you can correct the include files by running the shell script `fixincludes'. This installs modified, corrected copies of the files `ioctl.h', `ttychars.h' and many others, in a special directory where only GNU CC will normally look for them. This script will work on various systems because it chooses the files by searching all the system headers for the problem cases that we know about. If you cannot install the compiler's passes and run-time support in `/usr/local/lib', you can alternatively use the `-B' option to specify a prefix by which they may be found. The compiler concatenates the prefix with the names `cpp', `cc1' and `gnulib'. Thus, you can put the files in a directory `/usr/foo/gcc' and specify `-B/usr/foo/gcc/' when you run GNU CC. Also, you can specify an alternative default directory for these files by setting the Make variable `libdir' when you make GNU CC.  File: gcc.info, Node: Other Dir, Next: Sun Install, Prev: Installation, Up: 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. Go to that directory before running `config.gcc': 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. 2. Specify where to find `config.gcc' when you run it: ../gcc-1.36/config.gcc ... 3. Specify where to find the sources, as an argument to `config.gcc': ../gcc-1.36/config.gcc -srcdir=../gcc-1.36 sun3 The `-srcdir=DIR' option is not needed when the source directory is the parent of the current directory, because `config.gcc' detects that case automatically. Now, you can run `make' in that directory. You need not repeat the configuration steps shown above, when ordinary source files change. You must, however, run `config.gcc' again when the configuration files change, if your system does not support symbolic links.  File: gcc.info, Node: Sun Install, Next: 3b1 Install, Prev: Other Dir, Up: Installation Installing GNU CC on the Sun ============================ Make sure the environment variable `FLOAT_OPTION' is not set when you compile `gnulib'. If this option were set to `f68881' when `gnulib' 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 `alloca' 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 `alloca' 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"  File: gcc.info, Node: 3b1 Install, Next: VMS Install, Prev: Sun Install, Up: Installation Installing GNU CC on the 3b1 ============================ 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 `#include "config.h"' line on line 37 of `cccp.c' and do `make cpp'. This makes a preliminary version of GNU cpp. 2. Save the old `/lib/cpp' and copy the preliminary GNU cpp to that file name. 3. Undo your change in `cccp.c', or reinstall the original version, and do `make cpp' again. 4. Copy this final version of GNU cpp into `/lib/cpp'. 5. Replace every occurrence of `obstack_free' in `tree.c' with `_obstack_free'. 6. Run `make' to get the first-stage GNU CC. 7. Reinstall the original version of `/lib/cpp'. 8. Now you can compile GNU CC with itself and install it in the normal fashion. If you have installed an earlier version of GCC, you can compile the newer version with that. However, you will run into trouble compiling `gnulib', since that is normally compiled with CC. To solve the problem, uncomment this line in `Makefile': CCLIBFLAGS = -B/usr/local/lib/gcc- -tp -Wp,-traditional  File: gcc.info, Node: VMS Install, Next: HPUX Install, Prev: 3B1 Install, Up: 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 `gcc' 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 `GNU_CC' and `GNU_CC_INCLUDE' to point to the directories where the GNU CC executables (`gcc-cpp', `gcc-cc1', etc.) and the C include files are kept. This should be done with the commands: $ assign /super /system disk:[gcc.] gnu_cc $ assign /super /system 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 `GCC_INSTALL.COM' script in the `[GCC]' directory. 2. Install the `GCC' command with the command line: $ set command /table=sys$library:dcltables gnu_cc:[000000]gcc 3. To install the help file, do the following: $ lib/help sys$library:helplib.hlb gcc.hlp Now you can invoke the compiler with a command like `gcc /verbose file.c', which is equivalent to the command `gcc -v -c file.c' 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 that the sources, because we don't always have time to update them. (Use the `/verbose' option to determine the version number of the binaries and compare it with the source file `version.c' 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. Copy the file `tm-vms.h' to `tm.h', `xm-vms.h' to `config.h', `vax.md' to `md.' and `out-vax.c' to `aux-output.c'. The files to be copied are found in the subdirectory named `config'; they should be copied to the main directory of GNU CC. 2. Setup the logical names and command tables as defined above. In addition, define the vms logical name `GNU_BISON' to point at the to the directories where the Bison executable is kept. This should be done with the command: $ assign /super /system disk:[bison.] gnu_bison You may, if you choose, use the `INSTALL_BISON.COM' script in the `[BISON]' directory. 3. Install the `BISON' command with the command line: $ set command /table=sys$library:dcltables gnu_bison:[000000]bison 4. Type `@make' to do recompile everything. If you are compiling with a version of GNU CC older than 1.33, specify `/DEFINE=("inline=")' as an option in all the compilations. This requires editing all the `gcc' commands in `make-cc1.com'. (The older versions had problems supporting `inline'.) Once you have a working 1.33 or newer GNU CC, you can change this file back. There is a known problem on VMS: `const' global variables don't work compatibly with the VMS C compiler; we don't know a way to get them to the linker properly. Note that GNU CC on VMS does not generate debugging information to describe the program's symbols. It is not straightforward to implement this, and we have no time to spend on it, but we might consent to install a very modular implementation if you write it. You will probably have to modify GAS as well as GNU CC.  File: gcc.info, Node: HPUX Install, Prev: VMS Install, Up: Installation Installing GNU CC on HPUX ========================= To install GNU CC on HPUX, you must start by editing the file `Makefile'. Search for the string `HPUX' to find comments saying what to change. You need to change some variable definitions and (if you are using GAS) some lines in the rule for the target `gnulib'. To compile with the HPUX C compiler, you must specify get the file `alloca.c' from GNU Emacs. Then, when you run `make', use this argument: make ALLOCA=alloca.o When recompiling GNU CC with itself, do not define `ALLOCA'. Instead, an `-I' option needs to be added to `CFLAGS' as follows: make CC=stage1/gcc CFLAGS="-g -O -Bstage1/ -I../binutils/hp-include"  File: gcc.info, Node: Trouble, Next: Incompatibilities, Prev: Installation, Up: Top Known Causes of Trouble with GNU CC. ************************************ Here are some of the things that have caused trouble for people installing or using GNU CC. * On certain systems, defining certain environment variables such as `CC' can interfere with the functioning of `make'. * Cross compilation can run into trouble for certain machines because some target machines' assemblers require floating point numbers to be written as *integer* 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 `REAL_VALUE_TYPE'. But doing so is a substantial amount of work for each target machine. *Note Cross-compilation::. * 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. * Users often think it is a bug when GNU CC reports an error for code like this: int foo (short); int foo (x) short x; {...} The error message is correct: this code really is erroneous, because the old-style non-prototype definition passes subword integers in their promoted types. In other words, the argument is really an `int', not a `short'. The correct prototype is this: int foo (int); * 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 `struct mumble' the prototype is limited to the argument list containing it. It does not refer to the `struct mumble' defined with file scope immediately below--they are two unrelated types with similar names in different scopes. But in the definition of `foo', 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 `struct mumble' above the prototype. I don't think it's worth being incompatible for.  File: gcc.info, Node: Incompatibilities, Next: Extensions, Prev: Trouble, Up: Top Incompatibilities of GNU CC *************************** There are several noteworthy incompatibilities between GNU C and most existing (non-ANSI) versions of C. The `-traditional' option eliminates most of these incompatibilities, *but not all*, by telling GNU C to behave like older C compilers. * 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 `mktemp' with a string constant argument. The function `mktemp' always alters the string its argument points to. Another consequence is that `sscanf' does not work on some systems when passed a string constant as its format control string. This is because `sscanf' incorrectly tries to write into the string constant. Likewise `fscanf' and `scanf'. The best solution to these problems is to change the program to use `char'-array variables with initialization strings for these purposes instead of string constants. But if this is not possible, you can use the `-fwritable-strings' flag, which directs GNU CC to handle string constants the same way most C compilers do. `-traditional' also has this effect, among others. * 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 `"a"' regardless of what the argument A is. The `-traditional' option directs GNU CC to handle such cases (among others) in the old-fashioned (non-ANSI) fashion. * When you use `setjmp' and `longjmp', the only automatic variables guaranteed to remain valid are those declared `volatile'. 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 (); /* `longjmp (j)' may be occur in `fun3'. */ return a + fun3 (); } Here `a' may or may not be restored to its first value when the `longjmp' occurs. If `a' is allocated in a register, then its first value is restored; otherwise, it keeps the last value stored in it. If you use the `-W' option with the `-O' option, you will get a warning when GNU CC thinks such a problem might be possible. The `-traditional' option directs GNU C to put variables in the stack by default, rather than in registers, in functions that call `setjmp'. This results in the behavior found in traditional C compilers. * 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 `extern' declaration affects all the rest of the file even if it happens within a block. The `-traditional' option directs GNU C to treat all `extern' declarations as global, like traditional compilers. * In traditional C, you can combine `long', etc., with a typedef name, as shown here: typedef int foo; typedef long foo bar; In ANSI C, this is not allowed: `long' and other type modifiers require an explicit `int'. Because this criterion is expressed by Bison grammar rules rather than C code, the `-traditional' flag cannot alter it. * PCC allows typedef names to be used as function parameters. The difficulty described immediately above applies here too. * PCC allows whitespace in the middle of compound assignment operators such as `+='. GNU CC, following the ANSI standard, does not allow this. The difficulty described immediately above applies here too. * GNU CC will flag unterminated character constants inside of preprocessor 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 `/*...*/'. However, `-traditional' suppresses these error messages. * When compiling functions that return `float', PCC converts it to a double. GNU CC actually returns a `float'. If you are concerned with PCC compatibility, you should declare your functions to return `double'; you might as well say what you mean. * 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 in a special, fixed register. PCC usually handles all sizes of structures and unions by returning the address of a block of static storage containing the value. This method is not used in GNU CC because it is slower and nonreentrant. You can tell GNU CC to use the PCC convention with the option `-fpcc-struct-return'. * On the Sparc, GNU CC uses an incompatible calling convention for structures. It passes them by including their contents in the argument list, whereas the standard compiler passes them effectively by reference. This really ought to be fixed, but such calling conventions are not yet supported in GNU CC, so it isn't straightforward to fix it. The convention for structure returning is also incompatible, and `-fpcc-struct-return' does not help.  File: gcc.info, Node: Extensions, Next: Bugs, Prev: Incompatibilities, Up: Top GNU Extensions to the C Language ******************************** GNU C provides several language features not found in ANSI standard C. (The `-pedantic' 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 `__GNUC__', which is always defined under GNU CC. * Menu: * Statement Exprs:: Putting statements and declarations inside expressions. * Naming Types:: Giving a name to the type of some expression. * Typeof:: `typeof': referring to the type of an expression. * Lvalues:: Using `?:', `,' and casts in lvalues. * Conditionals:: Omitting the middle operand of a `?:' expression. * Zero-Length:: Zero-length arrays. * Variable-Length:: Arrays whose length is computed at run time. * Subscripting:: Any array can be subscripted, even if not an lvalue. * Pointer Arith:: Arithmetic on `void'-pointers and function pointers. * Initializers:: Non-constant initializers. * Constructors:: Constructor expressions give structures, unions or arrays as values. * Function Attributes:: Declaring that functions have no side effects, or that they can never return. * Dollar Signs:: Dollar sign is allowed in identifiers. * Alignment:: Inquiring about the alignment of a type or variable. * Inline:: Defining inline functions (as fast as macros). * Extended Asm:: Assembler instructions with C expressions as operands. (With them you can define ``built-in'' functions.) * Asm Labels:: Specifying the assembler name to use for a C symbol. * Explicit Reg Vars:: Defining variables residing in specified registers. * Alternate Keywords:: `__const__', `__asm__', etc., for header files.  File: gcc.info, Node: Statement Exprs, Next: Naming Types, Prev: Extensions, Up: Extensions Statements and Declarations inside of Expressions ================================================= A compound statement in parentheses may appear inside an expression in GNU C. This allows you to declare variables within an expression. 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 `foo ()'. 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 A or B 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 `int'), 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 `typeof' (*note Typeof::.) or type naming (*note Naming Types::.).  File: gcc.info, Node: Naming Types, Next: Typeof, Prev: Statement Exprs, Up: Extensions Naming an Expression's Type =========================== You can give a name to the type of an expression using a `typedef' declaration with an initializer. Here is how to define NAME as a type name for the type of EXP: typedef NAME = EXP; 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 `a' and `b'. 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.  File: gcc.info, Node: Typeof, Next: Lvalues, Prev: Naming Types, Up: Extensions Referring to a Type with `typeof' ================================= Another way to refer to the type of an expression is with `typeof'. The syntax of using of this keyword looks like `sizeof', but the construct acts semantically like a type name defined with `typedef'. There are two ways of writing the argument to `typeof': with an expression or with a type. Here is an example with an expression: typeof (x[0](1)) This assumes that `x' 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 `int'. If you are writing a header file that must work when included in ANSI C programs, write `__typeof__' instead of `typeof'. *Note Alternate Keywords::. A `typeof'-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 `sizeof' or `typeof'. * This declares `y' with the type of what `x' points to. typeof (*x) y; * This declares `y' as an array of such values. typeof (*x) y[4]; * This declares `y' 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 `typeof', 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, `array (pointer (char), 4)' is the type of arrays of 4 pointers to `char'.  File: gcc.info, Node: Lvalues, Next: Conditionals, Prev: Typeof, Up: 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. 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 valid. Taking the address of the cast is the same as taking the address without a cast, except for the type of the result. For example, these two expressions are equivalent (but the second may be valid when the type of `a' does not permit a cast to `int *'). &(int *)a (int **)&a 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 converter back to the specified type to become the value of the assignment. Thus, if `a' has type `char *', the following two expressions are equivalent: (int)a = 5 (int)(a = (char *)5) An assignment-with-arithmetic operation such as `+=' 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)a + 5))  File: gcc.info, Node: Conditionals, Next: Zero-Length, Prev: Lvalues, Up: Extensions Conditional Expressions with Omitted Middle-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 `x' if that is nonzero; otherwise, the value of `y'. 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.  File: gcc.info, Node: Zero-Length, Next: Variable-Length, Prev: Conditionals, Up: 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 `contents' a length of 1, which means either you waste space or complicate the argument to `malloc'.  File: gcc.info, Node: Variable-Length, Next: Subscripting, Prev: Zero-Length, Up: 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 that time 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); } You can also use variable-length arrays as arguments to functions: struct entry tester (int len, char data[len]) { ... } The length of an array is computed on entry to the brace-level where the array is declared and is remembered for the scope of the array in case you access it with `sizeof'. Jumping or breaking out of the scope of the array name will also deallocate the storage. Jumping into the scope is not allowed; you will get an error message for it. You can use the function `alloca' to get an effect much like variable-length arrays. The function `alloca' 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 `alloca' exists until the containing *function* 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 `alloca' in the same function, deallocation of a variable-length array will also deallocate anything more recently allocated with `alloca'.)  File: gcc.info, Node: Subscripting, Next: Pointer Arith, Prev: Variable-Length, Up: Extensions Non-Lvalue Arrays May Have Subscripts ===================================== Subscripting is allowed on arrays that are not lvalues, even though the unary `&' 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]; }  File: gcc.info, Node: Pointer Arith, Next: Initializers, Prev: Subscripting, Up: Extensions Arithmetic on `void'-Pointers and Function Pointers =================================================== In GNU C, addition and subtraction operations are supported on pointers to `void' and on pointers to functions. This is done by treating the size of a `void' or of a function as 1. A consequence of this is that `sizeof' is also allowed on `void' and on function types, and returns 1. The option `-Wpointer-arith' requests a warning if these extensions are used.  File: gcc.info, Node: Initializers, Next: Constructors, Prev: Pointer Arith, Up: Extensions Non-Constant Initializers ========================= 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 }; ... }  File: gcc.info, Node: Constructors, Next: Function Attributes, Prev: Initializers, Up: 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. The type must be a structure, union or array type. Assume that `struct foo' and `structure' are declared as shown: struct foo {int a; char b[2];} structure; Here is an example of constructing a `struct foo' 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 `switch' statement, while the latter does the same thing an ordinary C initializer would do. output = ((int[]) { 2, x, 28 }) [input];