1 @c Copyright (C) 1988,89,92,93,94,95,96,1997 Free Software Foundation, Inc.
2 @c This is part of the GCC manual.
3 @c For copying conditions, see the file gcc.texi.
6 @chapter GNU CC Command Options
7 @cindex GNU CC command options
8 @cindex command options
9 @cindex options, GNU CC command
11 When you invoke GNU CC, it normally does preprocessing, compilation,
12 assembly and linking. The ``overall options'' allow you to stop this
13 process at an intermediate stage. For example, the @samp{-c} option
14 says not to run the linker. Then the output consists of object files
15 output by the assembler.
17 Other options are passed on to one stage of processing. Some options
18 control the preprocessor and others the compiler itself. Yet other
19 options control the assembler and linker; most of these are not
20 documented here, since you rarely need to use any of them.
22 @cindex C compilation options
23 Most of the command line options that you can use with GNU CC are useful
24 for C programs; when an option is only useful with another language
25 (usually C++), the explanation says so explicitly. If the description
26 for a particular option does not mention a source language, you can use
27 that option with all supported languages.
29 @cindex C++ compilation options
30 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
31 options for compiling C++ programs.
33 @cindex grouping options
34 @cindex options, grouping
35 The @code{gcc} program accepts options and file names as operands. Many
36 options have multiletter names; therefore multiple single-letter options
37 may @emph{not} be grouped: @samp{-dr} is very different from @w{@samp{-d
40 @cindex order of options
41 @cindex options, order
42 You can mix options and other arguments. For the most part, the order
43 you use doesn't matter. Order does matter when you use several options
44 of the same kind; for example, if you specify @samp{-L} more than once,
45 the directories are searched in the order specified.
47 Many options have long names starting with @samp{-f} or with
48 @samp{-W}---for example, @samp{-fforce-mem},
49 @samp{-fstrength-reduce}, @samp{-Wformat} and so on. Most of
50 these have both positive and negative forms; the negative form of
51 @samp{-ffoo} would be @samp{-fno-foo}. This manual documents
52 only one of these two forms, whichever one is not the default.
55 * Option Summary:: Brief list of all options, without explanations.
56 * Overall Options:: Controlling the kind of output:
57 an executable, object files, assembler files,
58 or preprocessed source.
59 * Invoking G++:: Compiling C++ programs.
60 * C Dialect Options:: Controlling the variant of C language compiled.
61 * C++ Dialect Options:: Variations on C++.
62 * Warning Options:: How picky should the compiler be?
63 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
64 * Optimize Options:: How much optimization?
65 * Preprocessor Options:: Controlling header files and macro definitions.
66 Also, getting dependency information for Make.
67 * Assembler Options:: Passing options to the assembler.
68 * Link Options:: Specifying libraries and so on.
69 * Directory Options:: Where to find header files and libraries.
70 Where to find the compiler executable files.
71 * Target Options:: Running a cross-compiler, or an old version of GNU CC.
72 * Submodel Options:: Specifying minor hardware or convention variations,
73 such as 68010 vs 68020.
74 * Code Gen Options:: Specifying conventions for function calls, data layout
76 * Environment Variables:: Env vars that affect GNU CC.
77 * Running Protoize:: Automatically adding or removing function prototypes.
81 @section Option Summary
83 Here is a summary of all the options, grouped by type. Explanations are
84 in the following sections.
88 @xref{Overall Options,,Options Controlling the Kind of Output}.
90 -c -S -E -o @var{file} -pipe -v -x @var{language}
93 @item C Language Options
94 @xref{C Dialect Options,,Options Controlling C Dialect}.
96 -ansi -fallow-single-precision -fcond-mismatch -fno-asm
97 -fno-builtin -ffreestanding - fhosted -fsigned-bitfields -fsigned-char
98 -funsigned-bitfields -funsigned-char -fwritable-strings
99 -traditional -traditional-cpp -trigraphs
102 @item C++ Language Options
103 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
105 -fall-virtual -fdollars-in-identifiers -felide-constructors
106 -fenum-int-equiv -fexternal-templates -ffor-scope -fno-for-scope
107 -fhandle-signatures -fmemoize-lookups -fno-default-inline -fno-gnu-keywords
108 -fnonnull-objects -foperator-names -fstrict-prototype -fthis-is-variable
109 -ftemplate-depth-@var{n} -nostdinc++ -traditional +e@var{n}
112 @item Warning Options
113 @xref{Warning Options,,Options to Request or Suppress Warnings}.
115 -fsyntax-only -pedantic -pedantic-errors
116 -w -W -Wall -Waggregate-return -Wbad-function-cast
117 -Wcast-align -Wcast-qual -Wchar-subscript -Wcomment
118 -Wconversion -Werror -Wformat
119 -Wid-clash-@var{len} -Wimplicit -Wimport -Winline
120 -Wlarger-than-@var{len} -Wmain -Wmissing-declarations
121 -Wmissing-prototypes -Wnested-externs
122 -Wno-import -Woverloaded-virtual -Wparentheses
123 -Wpointer-arith -Wredundant-decls -Wreorder -Wreturn-type -Wshadow
124 -Wsign-compare -Wstrict-prototypes -Wswitch -Wsynth
125 -Wtemplate-debugging -Wtraditional -Wtrigraphs
126 -Wundef -Wuninitialized -Wunused -Wwrite-strings
129 @item Debugging Options
130 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
132 -a -ax -d@var{letters} -fpretend-float
133 -fprofile-arcs -ftest-coverage
134 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2
135 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+
136 -p -pg -print-file-name=@var{library} -print-libgcc-file-name
137 -print-prog-name=@var{program} -print-search-dirs -save-temps
140 @item Optimization Options
141 @xref{Optimize Options,,Options that Control Optimization}.
143 -fbranch-probabilities
144 -fcaller-saves -fcse-follow-jumps -fcse-skip-blocks
145 -fdelayed-branch -fexpensive-optimizations
146 -ffast-math -ffloat-store -fforce-addr -fforce-mem
147 -ffunction-sections -finline-functions
148 -fkeep-inline-functions -fno-default-inline
149 -fno-defer-pop -fno-function-cse
150 -fno-inline -fno-peephole -fomit-frame-pointer -fregmove
151 -frerun-cse-after-loop -fschedule-insns
152 -fschedule-insns2 -fstrength-reduce -fthread-jumps
153 -funroll-all-loops -funroll-loops
154 -fmove-all-movables -freduce-all-givs
158 @item Preprocessor Options
159 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
161 -A@var{question}(@var{answer}) -C -dD -dM -dN
162 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H
164 -include @var{file} -imacros @var{file}
165 -iprefix @var{file} -iwithprefix @var{dir}
166 -iwithprefixbefore @var{dir} -isystem @var{dir}
167 -M -MD -MM -MMD -MG -nostdinc -P -trigraphs
168 -undef -U@var{macro} -Wp,@var{option}
171 @item Assembler Option
172 @xref{Assembler Options,,Passing Options to the Assembler}.
178 @xref{Link Options,,Options for Linking}.
180 @var{object-file-name} -l@var{library}
181 -nostartfiles -nodefaultlibs -nostdlib
182 -s -static -shared -symbolic
183 -Wl,@var{option} -Xlinker @var{option}
187 @item Directory Options
188 @xref{Directory Options,,Options for Directory Search}.
190 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}
194 @c I wrote this xref this way to avoid overfull hbox. -- rms
195 @xref{Target Options}.
197 -b @var{machine} -V @var{version}
200 @item Machine Dependent Options
201 @xref{Submodel Options,,Hardware Models and Configurations}.
203 @emph{M680x0 Options}
204 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040
205 -m68060 -m5200 -m68881 -mbitfield -mc68000 -mc68020 -mfpa
206 -mnobitfield -mrtd -mshort -msoft-float -malign-int
213 -mtune=@var{cpu type}
214 -mapp-regs -mbroken-saverestore -mcypress -mepilogue
215 -mflat -mfpu -mfullany -mhard-float -mhard-quad-float
216 -mimpure-text -mint32 -mint64 -mlive-g0
217 -mlong32 -mlong64 -mmedlow -mmedany
218 -mno-app-regs -mno-epilogue
219 -mno-flat -mno-fpu -mno-impure-text
220 -mno-stack-bias -mno-unaligned-doubles
221 -msoft-float -msoft-quad-float -msparclite -mstack-bias
222 -msupersparc -munaligned-doubles -mv8
224 @emph{Convex Options}
225 -mc1 -mc2 -mc32 -mc34 -mc38
226 -margcount -mnoargcount
228 -mvolatile-cache -mvolatile-nocache
230 @emph{AMD29K Options}
231 -m29000 -m29050 -mbw -mnbw -mdw -mndw
232 -mlarge -mnormal -msmall
233 -mkernel-registers -mno-reuse-arg-regs
234 -mno-stack-check -mno-storem-bug
235 -mreuse-arg-regs -msoft-float -mstack-check
236 -mstorem-bug -muser-registers
239 -mapcs-frame -mapcs-26 -mapcs-32
240 -mlittle-endian -mbig-endian -mwords-little-endian
241 -mshort-load-bytes -mno-short-load-bytes
242 -msoft-float -mhard-float
243 -mbsd -mxopen -mno-symrename
245 @emph{M32R/D Options}
246 -mcode-model=@var{model type} -msdata=@var{sdata type}
250 -m88000 -m88100 -m88110 -mbig-pic
251 -mcheck-zero-division -mhandle-large-shift
252 -midentify-revision -mno-check-zero-division
253 -mno-ocs-debug-info -mno-ocs-frame-position
254 -mno-optimize-arg-area -mno-serialize-volatile
255 -mno-underscores -mocs-debug-info
256 -mocs-frame-position -moptimize-arg-area
257 -mserialize-volatile -mshort-data-@var{num} -msvr3
258 -msvr4 -mtrap-large-shift -muse-div-instruction
259 -mversion-03.00 -mwarn-passed-structs
261 @emph{RS/6000 and PowerPC Options}
263 -mtune=@var{cpu type}
264 -mpower -mno-power -mpower2 -mno-power2
265 -mpowerpc -mno-powerpc
266 -mpowerpc-gpopt -mno-powerpc-gpopt
267 -mpowerpc-gfxopt -mno-powerpc-gfxopt
268 -mnew-mnemonics -mno-new-mnemonics
269 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc
270 -mxl-call -mno-xl-call -mthreads -mpe
271 -msoft-float -mhard-float -mmultiple -mno-multiple
272 -mstring -mno-string -mupdate -mno-update
273 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align
274 -mstrict-align -mno-strict-align -mrelocatable
275 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib
276 -mtoc -mno-toc -mtraceback -mno-traceback
277 -mlittle -mlittle-endian -mbig -mbig-endian
278 -mcall-aix -mcall-sysv -mprototype -mno-prototype
279 -msim -mmvme -memb -msdata -msdata=@var{opt} -G @var{num}
282 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs
283 -mfull-fp-blocks -mhc-struct-return -min-line-mul
284 -mminimum-fp-blocks -mnohc-struct-return
287 -mabicalls -mcpu=@var{cpu type} -membedded-data
288 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64
289 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1
290 -mips2 -mips3 -mlong64 -mlong-calls -mmemcpy
291 -mmips-as -mmips-tfile -mno-abicalls
292 -mno-embedded-data -mno-embedded-pic
293 -mno-gpopt -mno-long-calls
294 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats
295 -mrnames -msoft-float
296 -m4650 -msingle-float -mmad
297 -mstats -EL -EB -G @var{num} -nocpp
300 -m486 -m386 -mieee-fp -mno-fancy-math-387
301 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib
302 -mno-wide-multiply -mrtd -malign-double
303 -mreg-alloc=@var{list} -mregparm=@var{num}
304 -malign-jumps=@var{num} -malign-loops=@var{num}
305 -malign-functions=@var{num}
308 -mbig-switch -mdisable-fpregs -mdisable-indexing -mfast-indirect-calls
309 -mgas -mjump-in-delay -mlong-load-store -mno-big-switch -mno-disable-fpregs
310 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas
313 -mno-portable-runtime -mno-soft-float -mno-space -mno-space-regs -msoft-float
314 -mpa-risc-1-0 -mpa-risc-1-1 -mportable-runtime
315 -mschedule=@var{list} -mspace -mspace-regs
317 @emph{Intel 960 Options}
318 -m@var{cpu type} -masm-compat -mclean-linkage
319 -mcode-align -mcomplex-addr -mleaf-procedures
320 -mic-compat -mic2.0-compat -mic3.0-compat
321 -mintel-asm -mno-clean-linkage -mno-code-align
322 -mno-complex-addr -mno-leaf-procedures
323 -mno-old-align -mno-strict-align -mno-tail-call
324 -mnumerics -mold-align -msoft-float -mstrict-align
327 @emph{DEC Alpha Options}
328 -mfp-regs -mno-fp-regs
329 -mno-soft-float -msoft-float
330 -mieee -mieee-with-inexact -mieee-conformant
331 -mfp-trap-mode -mfp-rounding-mode -mtrap-precision
334 @emph{Clipper Options}
337 @emph{H8/300 Options}
338 -mrelax -mh -mint32 -malign-300
341 -m1 -m2 -m3 -m3e -mb -ml -mrelax
343 @emph{System V Options}
344 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}
347 -mlong-calls -mno-long-calls -mep -mno-ep
348 -mprolog-function -mno-prolog-function -mspace
349 -mtda=@var{n} -msda=@var{n} -mzda=@var{n}
352 @item Code Generation Options
353 @xref{Code Gen Options,,Options for Code Generation Conventions}.
355 -fcall-saved-@var{reg} -fcall-used-@var{reg}
356 -ffixed-@var{reg} -finhibit-size-directive
357 -fcheck-memory-usage -fprefix-function-name
358 -fno-common -fno-ident -fno-gnu-linker
359 -fpcc-struct-return -fpic -fPIC
360 -freg-struct-return -fshared-data -fshort-enums
361 -fshort-double -fvolatile -fvolatile-global
362 -fverbose-asm -fpack-struct -fstack-check +e0 +e1
363 -fargument-alias -fargument-noalias
364 -fargument-noalias-global
369 * Overall Options:: Controlling the kind of output:
370 an executable, object files, assembler files,
371 or preprocessed source.
372 * C Dialect Options:: Controlling the variant of C language compiled.
373 * C++ Dialect Options:: Variations on C++.
374 * Warning Options:: How picky should the compiler be?
375 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
376 * Optimize Options:: How much optimization?
377 * Preprocessor Options:: Controlling header files and macro definitions.
378 Also, getting dependency information for Make.
379 * Assembler Options:: Passing options to the assembler.
380 * Link Options:: Specifying libraries and so on.
381 * Directory Options:: Where to find header files and libraries.
382 Where to find the compiler executable files.
383 * Target Options:: Running a cross-compiler, or an old version of GNU CC.
386 @node Overall Options
387 @section Options Controlling the Kind of Output
389 Compilation can involve up to four stages: preprocessing, compilation
390 proper, assembly and linking, always in that order. The first three
391 stages apply to an individual source file, and end by producing an
392 object file; linking combines all the object files (those newly
393 compiled, and those specified as input) into an executable file.
395 @cindex file name suffix
396 For any given input file, the file name suffix determines what kind of
401 C source code which must be preprocessed.
404 C source code which should not be preprocessed.
407 C++ source code which should not be preprocessed.
410 Objective-C source code. Note that you must link with the library
411 @file{libobjc.a} to make an Objective-C program work.
414 C header file (not to be compiled or linked).
417 @itemx @var{file}.cxx
418 @itemx @var{file}.cpp
420 C++ source code which must be preprocessed. Note that in @samp{.cxx},
421 the last two letters must both be literally @samp{x}. Likewise,
422 @samp{.C} refers to a literal capital C.
428 Assembler code which must be preprocessed.
431 An object file to be fed straight into linking.
432 Any file name with no recognized suffix is treated this way.
435 You can specify the input language explicitly with the @samp{-x} option:
438 @item -x @var{language}
439 Specify explicitly the @var{language} for the following input files
440 (rather than letting the compiler choose a default based on the file
441 name suffix). This option applies to all following input files until
442 the next @samp{-x} option. Possible values for @var{language} are:
445 c-header cpp-output c++-cpp-output
446 assembler assembler-with-cpp
450 Turn off any specification of a language, so that subsequent files are
451 handled according to their file name suffixes (as they are if @samp{-x}
452 has not been used at all).
455 If you only want some of the stages of compilation, you can use
456 @samp{-x} (or filename suffixes) to tell @code{gcc} where to start, and
457 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
458 @code{gcc} is to stop. Note that some combinations (for example,
459 @samp{-x cpp-output -E} instruct @code{gcc} to do nothing at all.
463 Compile or assemble the source files, but do not link. The linking
464 stage simply is not done. The ultimate output is in the form of an
465 object file for each source file.
467 By default, the object file name for a source file is made by replacing
468 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
470 Unrecognized input files, not requiring compilation or assembly, are
474 Stop after the stage of compilation proper; do not assemble. The output
475 is in the form of an assembler code file for each non-assembler input
478 By default, the assembler file name for a source file is made by
479 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
481 Input files that don't require compilation are ignored.
484 Stop after the preprocessing stage; do not run the compiler proper. The
485 output is in the form of preprocessed source code, which is sent to the
488 Input files which don't require preprocessing are ignored.
490 @cindex output file option
492 Place output in file @var{file}. This applies regardless to whatever
493 sort of output is being produced, whether it be an executable file,
494 an object file, an assembler file or preprocessed C code.
496 Since only one output file can be specified, it does not make sense to
497 use @samp{-o} when compiling more than one input file, unless you are
498 producing an executable file as output.
500 If @samp{-o} is not specified, the default is to put an executable file
501 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
502 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
503 all preprocessed C source on standard output.@refill
506 Print (on standard error output) the commands executed to run the stages
507 of compilation. Also print the version number of the compiler driver
508 program and of the preprocessor and the compiler proper.
511 Use pipes rather than temporary files for communication between the
512 various stages of compilation. This fails to work on some systems where
513 the assembler is unable to read from a pipe; but the GNU assembler has
518 @section Compiling C++ Programs
520 @cindex suffixes for C++ source
521 @cindex C++ source file suffixes
522 C++ source files conventionally use one of the suffixes @samp{.C},
523 @samp{.cc}, @samp{cpp}, or @samp{.cxx}; preprocessed C++ files use the
524 suffix @samp{.ii}. GNU CC recognizes files with these names and
525 compiles them as C++ programs even if you call the compiler the same way
526 as for compiling C programs (usually with the name @code{gcc}).
530 However, C++ programs often require class libraries as well as a
531 compiler that understands the C++ language---and under some
532 circumstances, you might want to compile programs from standard input,
533 or otherwise without a suffix that flags them as C++ programs.
534 @code{g++} is a program that calls GNU CC with the default language
535 set to C++, and automatically specifies linking against the GNU class
537 @cindex @code{g++ 1.@var{xx}}
538 @cindex @code{g++}, separate compiler
539 @cindex @code{g++} older version
540 @footnote{Prior to release 2 of the compiler,
541 there was a separate @code{g++} compiler. That version was based on GNU
542 CC, but not integrated with it. Versions of @code{g++} with a
543 @samp{1.@var{xx}} version number---for example, @code{g++} version 1.37
544 or 1.42---are much less reliable than the versions integrated with GCC
545 2. Moreover, combining G++ @samp{1.@var{xx}} with a version 2 GCC will
546 simply not work.} On many systems, the script @code{g++} is also
547 installed with the name @code{c++}.
549 @cindex invoking @code{g++}
550 When you compile C++ programs, you may specify many of the same
551 command-line options that you use for compiling programs in any
552 language; or command-line options meaningful for C and related
553 languages; or options that are meaningful only for C++ programs.
554 @xref{C Dialect Options,,Options Controlling C Dialect}, for
555 explanations of options for languages related to C.
556 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
557 explanations of options that are meaningful only for C++ programs.
559 @node C Dialect Options
560 @section Options Controlling C Dialect
561 @cindex dialect options
562 @cindex language dialect options
563 @cindex options, dialect
565 The following options control the dialect of C (or languages derived
566 from C, such as C++ and Objective C) that the compiler accepts:
571 Support all ANSI standard C programs.
573 This turns off certain features of GNU C that are incompatible with ANSI
574 C, such as the @code{asm}, @code{inline} and @code{typeof} keywords, and
575 predefined macros such as @code{unix} and @code{vax} that identify the
576 type of system you are using. It also enables the undesirable and
577 rarely used ANSI trigraph feature, and it disables recognition of C++
578 style @samp{//} comments.
580 The alternate keywords @code{__asm__}, @code{__extension__},
581 @code{__inline__} and @code{__typeof__} continue to work despite
582 @samp{-ansi}. You would not want to use them in an ANSI C program, of
583 course, but it is useful to put them in header files that might be included
584 in compilations done with @samp{-ansi}. Alternate predefined macros
585 such as @code{__unix__} and @code{__vax__} are also available, with or
586 without @samp{-ansi}.
588 The @samp{-ansi} option does not cause non-ANSI programs to be
589 rejected gratuitously. For that, @samp{-pedantic} is required in
590 addition to @samp{-ansi}. @xref{Warning Options}.
592 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
593 option is used. Some header files may notice this macro and refrain
594 from declaring certain functions or defining certain macros that the
595 ANSI standard doesn't call for; this is to avoid interfering with any
596 programs that might use these names for other things.
598 The functions @code{alloca}, @code{abort}, @code{exit}, and
599 @code{_exit} are not builtin functions when @samp{-ansi} is used.
602 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
603 keyword, so that code can use these words as identifiers. You can use
604 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
605 instead. @samp{-ansi} implies @samp{-fno-asm}.
607 In C++, this switch only affects the @code{typeof} keyword, since
608 @code{asm} and @code{inline} are standard keywords. You may want to
609 use the @samp{-fno-gnu-keywords} flag instead, as it also disables the
610 other, C++-specific, extension keywords such as @code{headof}.
613 @cindex builtin functions
629 Don't recognize builtin functions that do not begin with two leading
630 underscores. Currently, the functions affected include @code{abort},
631 @code{abs}, @code{alloca}, @code{cos}, @code{exit}, @code{fabs},
632 @code{ffs}, @code{labs}, @code{memcmp}, @code{memcpy}, @code{sin},
633 @code{sqrt}, @code{strcmp}, @code{strcpy}, and @code{strlen}.
635 GCC normally generates special code to handle certain builtin functions
636 more efficiently; for instance, calls to @code{alloca} may become single
637 instructions that adjust the stack directly, and calls to @code{memcpy}
638 may become inline copy loops. The resulting code is often both smaller
639 and faster, but since the function calls no longer appear as such, you
640 cannot set a breakpoint on those calls, nor can you change the behavior
641 of the functions by linking with a different library.
643 The @samp{-ansi} option prevents @code{alloca} and @code{ffs} from being
644 builtin functions, since these functions do not have an ANSI standard
648 @cindex hosted environment
650 Assert that compilation takes place in a hosted environment. This implies
651 @samp{-fbuiltin}. A hosted environment is one in which the
652 entire standard library is available, and in which @code{main} has a return
653 type of @code{int}. Examples are nearly everything except a kernel.
654 This is equivalent to @samp{-fno-freestanding}.
657 @cindex hosted environment
659 Assert that compilation takes place in a freestanding environment. This
660 implies @samp{-fno-builtin}. A freestanding environment
661 is one in which the standard library may not exist, and program startup may
662 not necessarily be at @code{main}. The most obvious example is an OS kernel.
663 This is equivalent to @samp{-fno-hosted}.
666 Support ANSI C trigraphs. You don't want to know about this
667 brain-damage. The @samp{-ansi} option implies @samp{-trigraphs}.
669 @cindex traditional C language
670 @cindex C language, traditional
672 Attempt to support some aspects of traditional C compilers.
677 All @code{extern} declarations take effect globally even if they
678 are written inside of a function definition. This includes implicit
679 declarations of functions.
682 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
683 and @code{volatile} are not recognized. (You can still use the
684 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
688 Comparisons between pointers and integers are always allowed.
691 Integer types @code{unsigned short} and @code{unsigned char} promote
692 to @code{unsigned int}.
695 Out-of-range floating point literals are not an error.
698 Certain constructs which ANSI regards as a single invalid preprocessing
699 number, such as @samp{0xe-0xd}, are treated as expressions instead.
702 String ``constants'' are not necessarily constant; they are stored in
703 writable space, and identical looking constants are allocated
704 separately. (This is the same as the effect of
705 @samp{-fwritable-strings}.)
707 @cindex @code{longjmp} and automatic variables
709 All automatic variables not declared @code{register} are preserved by
710 @code{longjmp}. Ordinarily, GNU C follows ANSI C: automatic variables
711 not declared @code{volatile} may be clobbered.
716 @cindex escape sequences, traditional
717 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
718 literal characters @samp{x} and @samp{a} respectively. Without
719 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
720 representation of a character, and @samp{\a} produces a bell.
723 In C++ programs, assignment to @code{this} is permitted with
724 @samp{-traditional}. (The option @samp{-fthis-is-variable} also has
728 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
729 if your program uses names that are normally GNU C builtin functions for
730 other purposes of its own.
732 You cannot use @samp{-traditional} if you include any header files that
733 rely on ANSI C features. Some vendors are starting to ship systems with
734 ANSI C header files and you cannot use @samp{-traditional} on such
735 systems to compile files that include any system headers.
737 The @samp{-traditional} option also enables the @samp{-traditional-cpp}
738 option, which is described next.
740 @item -traditional-cpp
741 Attempt to support some aspects of traditional C preprocessors.
746 Comments convert to nothing at all, rather than to a space. This allows
747 traditional token concatenation.
750 In a preprocessing directive, the @samp{#} symbol must appear as the first
754 Macro arguments are recognized within string constants in a macro
755 definition (and their values are stringified, though without additional
756 quote marks, when they appear in such a context). The preprocessor
757 always considers a string constant to end at a newline.
760 @cindex detecting @w{@samp{-traditional}}
761 The predefined macro @code{__STDC__} is not defined when you use
762 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
763 which @code{__GNUC__} indicates are not affected by
764 @samp{-traditional}). If you need to write header files that work
765 differently depending on whether @samp{-traditional} is in use, by
766 testing both of these predefined macros you can distinguish four
767 situations: GNU C, traditional GNU C, other ANSI C compilers, and other
768 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
769 not defined when you use @samp{-traditional}. @xref{Standard
770 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
771 for more discussion of these and other predefined macros.
774 @cindex string constants vs newline
775 @cindex newline vs string constants
776 The preprocessor considers a string constant to end at a newline (unless
777 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
778 string constants can contain the newline character as typed.)
781 @item -fcond-mismatch
782 Allow conditional expressions with mismatched types in the second and
783 third arguments. The value of such an expression is void.
785 @item -funsigned-char
786 Let the type @code{char} be unsigned, like @code{unsigned char}.
788 Each kind of machine has a default for what @code{char} should
789 be. It is either like @code{unsigned char} by default or like
790 @code{signed char} by default.
792 Ideally, a portable program should always use @code{signed char} or
793 @code{unsigned char} when it depends on the signedness of an object.
794 But many programs have been written to use plain @code{char} and
795 expect it to be signed, or expect it to be unsigned, depending on the
796 machines they were written for. This option, and its inverse, let you
797 make such a program work with the opposite default.
799 The type @code{char} is always a distinct type from each of
800 @code{signed char} or @code{unsigned char}, even though its behavior
801 is always just like one of those two.
804 Let the type @code{char} be signed, like @code{signed char}.
806 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
807 the negative form of @samp{-funsigned-char}. Likewise, the option
808 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
810 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
811 if your program uses names that are normally GNU C builtin functions for
812 other purposes of its own.
814 You cannot use @samp{-traditional} if you include any header files that
815 rely on ANSI C features. Some vendors are starting to ship systems with
816 ANSI C header files and you cannot use @samp{-traditional} on such
817 systems to compile files that include any system headers.
819 @item -fsigned-bitfields
820 @itemx -funsigned-bitfields
821 @itemx -fno-signed-bitfields
822 @itemx -fno-unsigned-bitfields
823 These options control whether a bitfield is signed or unsigned, when the
824 declaration does not use either @code{signed} or @code{unsigned}. By
825 default, such a bitfield is signed, because this is consistent: the
826 basic integer types such as @code{int} are signed types.
828 However, when @samp{-traditional} is used, bitfields are all unsigned
831 @item -fwritable-strings
832 Store string constants in the writable data segment and don't uniquize
833 them. This is for compatibility with old programs which assume they can
834 write into string constants. The option @samp{-traditional} also has
837 Writing into string constants is a very bad idea; ``constants'' should
840 @item -fallow-single-precision
841 Do not promote single precision math operations to double precision,
842 even when compiling with @samp{-traditional}.
844 Traditional K&R C promotes all floating point operations to double
845 precision, regardless of the sizes of the operands. On the
846 architecture for which you are compiling, single precision may be faster
847 than double precision. If you must use @samp{-traditional}, but want
848 to use single precision operations when the operands are single
849 precision, use this option. This option has no effect when compiling
850 with ANSI or GNU C conventions (the default).
854 @node C++ Dialect Options
855 @section Options Controlling C++ Dialect
857 @cindex compiler options, C++
858 @cindex C++ options, command line
860 This section describes the command-line options that are only meaningful
861 for C++ programs; but you can also use most of the GNU compiler options
862 regardless of what language your program is in. For example, you
863 might compile a file @code{firstClass.C} like this:
866 g++ -g -felide-constructors -O -c firstClass.C
870 In this example, only @samp{-felide-constructors} is an option meant
871 only for C++ programs; you can use the other options with any
872 language supported by GNU CC.
874 Here is a list of options that are @emph{only} for compiling C++ programs:
877 @item -fno-access-control
878 Turn off all access checking. This switch is mainly useful for working
879 around bugs in the access control code.
882 Treat all possible member functions as virtual, implicitly.
883 All member functions (except for constructor functions and @code{new} or
884 @code{delete} member operators) are treated as virtual functions of the
885 class where they appear.
887 This does not mean that all calls to these member functions will be made
888 through the internal table of virtual functions. Under some
889 circumstances, the compiler can determine that a call to a given virtual
890 function can be made directly; in these cases the calls are direct in
894 Check that the pointer returned by @code{operator new} is non-null
895 before attempting to modify the storage allocated. The current Working
896 Paper requires that @code{operator new} never return a null pointer, so
897 this check is normally unnecessary.
899 @item -fconserve-space
900 Put uninitialized or runtime-initialized global variables into the
901 common segment, as C does. This saves space in the executable at the
902 cost of not diagnosing duplicate definitions. If you compile with this
903 flag and your program mysteriously crashes after @code{main()} has
904 completed, you may have an object that is being destroyed twice because
905 two definitions were merged.
907 @item -fdollars-in-identifiers
908 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
909 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
910 @samp{$} by default on most target systems, but there are a few exceptions.)
911 Traditional C allowed the character @samp{$} to form part of
912 identifiers. However, ANSI C and C++ forbid @samp{$} in identifiers.
914 @item -fenum-int-equiv
915 Anachronistically permit implicit conversion of @code{int} to
916 enumeration types. Current C++ allows conversion of @code{enum} to
917 @code{int}, but not the other way around.
919 @item -fexternal-templates
920 Cause template instantiations to obey @samp{#pragma interface} and
921 @samp{implementation}; template instances are emitted or not according
922 to the location of the template definition. @xref{Template
923 Instantiation}, for more information.
925 @item -falt-external-templates
926 Similar to -fexternal-templates, but template instances are emitted or
927 not according to the place where they are first instantiated.
928 @xref{Template Instantiation}, for more information.
932 If -ffor-scope is specified, the scope of variables declared in
933 a @i{for-init-statement} is limited to the @samp{for} loop itself,
934 as specified by the draft C++ standard.
935 If -fno-for-scope is specified, the scope of variables declared in
936 a @i{for-init-statement} extends to the end of the enclosing scope,
937 as was the case in old versions of gcc, and other (traditional)
938 implementations of C++.
940 The default if neither flag is given to follow the standard,
941 but to allow and give a warning for old-style code that would
942 otherwise be invalid, or have different behavior.
944 @item -fno-gnu-keywords
945 Do not recognize @code{classof}, @code{headof}, @code{signature},
946 @code{sigof} or @code{typeof} as a keyword, so that code can use these
947 words as identifiers. You can use the keywords @code{__classof__},
948 @code{__headof__}, @code{__signature__}, @code{__sigof__}, and
949 @code{__typeof__} instead. @samp{-ansi} implies
950 @samp{-fno-gnu-keywords}.
952 @item -fno-implicit-templates
953 Never emit code for templates which are instantiated implicitly (i.e. by
954 use); only emit code for explicit instantiations. @xref{Template
955 Instantiation}, for more information.
957 @item -fhandle-signatures
958 Recognize the @code{signature} and @code{sigof} keywords for specifying
959 abstract types. The default (@samp{-fno-handle-signatures}) is not to
960 recognize them. @xref{C++ Signatures, Type Abstraction using
964 Support virtual function calls for objects that exceed the size
965 representable by a @samp{short int}. Users should not use this flag by
966 default; if you need to use it, the compiler will tell you so. If you
967 compile any of your code with this flag, you must compile @emph{all} of
968 your code with this flag (including libg++, if you use it).
970 This flag is not useful when compiling with -fvtable-thunks.
972 @item -fno-implement-inlines
973 To save space, do not emit out-of-line copies of inline functions
974 controlled by @samp{#pragma implementation}. This will cause linker
975 errors if these functions are not inlined everywhere they are called.
977 @item -fmemoize-lookups
978 @itemx -fsave-memoized
979 Use heuristics to compile faster. These heuristics are not enabled by
980 default, since they are only effective for certain input files. Other
981 input files compile more slowly.
983 The first time the compiler must build a call to a member function (or
984 reference to a data member), it must (1) determine whether the class
985 implements member functions of that name; (2) resolve which member
986 function to call (which involves figuring out what sorts of type
987 conversions need to be made); and (3) check the visibility of the member
988 function to the caller. All of this adds up to slower compilation.
989 Normally, the second time a call is made to that member function (or
990 reference to that data member), it must go through the same lengthy
991 process again. This means that code like this:
994 cout << "This " << p << " has " << n << " legs.\n";
998 makes six passes through all three steps. By using a software cache, a
999 ``hit'' significantly reduces this cost. Unfortunately, using the cache
1000 introduces another layer of mechanisms which must be implemented, and so
1001 incurs its own overhead. @samp{-fmemoize-lookups} enables the software
1004 Because access privileges (visibility) to members and member functions
1005 may differ from one function context to the next, G++ may need to flush
1006 the cache. With the @samp{-fmemoize-lookups} flag, the cache is flushed
1007 after every function that is compiled. The @samp{-fsave-memoized} flag
1008 enables the same software cache, but when the compiler determines that
1009 the context of the last function compiled would yield the same access
1010 privileges of the next function to compile, it preserves the cache.
1011 This is most helpful when defining many member functions for the same
1012 class: with the exception of member functions which are friends of other
1013 classes, each member function has exactly the same access privileges as
1014 every other, and the cache need not be flushed.
1016 The code that implements these flags has rotted; you should probably
1019 @item -fstrict-prototype
1020 Within an @samp{extern "C"} linkage specification, treat a function
1021 declaration with no arguments, such as @samp{int foo ();}, as declaring
1022 the function to take no arguments. Normally, such a declaration means
1023 that the function @code{foo} can take any combination of arguments, as
1024 in C. @samp{-pedantic} implies @samp{-fstrict-prototype} unless
1025 overridden with @samp{-fno-strict-prototype}.
1027 This flag no longer affects declarations with C++ linkage.
1029 @item -fno-nonnull-objects
1030 Don't assume that a reference is initialized to refer to a valid object.
1031 Although the current C++ Working Paper prohibits null references, some
1032 old code may rely on them, and you can use @samp{-fno-nonnull-objects}
1033 to turn on checking.
1035 At the moment, the compiler only does this checking for conversions to
1036 virtual base classes.
1038 @item -foperator-names
1039 Recognize the operator name keywords @code{and}, @code{bitand},
1040 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1041 synonyms for the symbols they refer to. @samp{-ansi} implies
1042 @samp{-foperator-names}.
1044 @item -fthis-is-variable
1045 Permit assignment to @code{this}. The incorporation of user-defined
1046 free store management into C++ has made assignment to @samp{this} an
1047 anachronism. Therefore, by default it is invalid to assign to
1048 @code{this} within a class member function; that is, GNU C++ treats
1049 @samp{this} in a member function of class @code{X} as a non-lvalue of
1050 type @samp{X *}. However, for backwards compatibility, you can make it
1051 valid with @samp{-fthis-is-variable}.
1053 @item -fvtable-thunks
1054 Use @samp{thunks} to implement the virtual function dispatch table
1055 (@samp{vtable}). The traditional (cfront-style) approach to
1056 implementing vtables was to store a pointer to the function and two
1057 offsets for adjusting the @samp{this} pointer at the call site. Newer
1058 implementations store a single pointer to a @samp{thunk} function which
1059 does any necessary adjustment and then calls the target function.
1061 This option also enables a heuristic for controlling emission of
1062 vtables; if a class has any non-inline virtual functions, the vtable
1063 will be emitted in the translation unit containing the first one of
1066 @item -ftemplate-depth-@var{n}
1067 Set the maximum instantiation depth for template classes to @var{n}.
1068 A limit on the template instantiation depth is needed to detect
1069 endless recursions during template class instantiation. ANSI/ISO C++
1070 conforming programs must not rely on a maximum depth greater than 17.
1073 Do not search for header files in the standard directories specific to
1074 C++, but do still search the other standard directories. (This option
1075 is used when building libg++.)
1078 For C++ programs (in addition to the effects that apply to both C and
1079 C++), this has the same effect as @samp{-fthis-is-variable}.
1080 @xref{C Dialect Options,, Options Controlling C Dialect}.
1083 In addition, these optimization, warning, and code generation options
1084 have meanings only for C++ programs:
1087 @item -fno-default-inline
1088 Do not assume @samp{inline} for functions defined inside a class scope.
1089 @xref{Optimize Options,,Options That Control Optimization}.
1091 @item -Woverloaded-virtual
1092 @itemx -Wtemplate-debugging
1093 Warnings that apply only to C++ programs. @xref{Warning
1094 Options,,Options to Request or Suppress Warnings}.
1097 Warn about violation of some style rules from Effective C++ by Scott Myers.
1100 Control how virtual function definitions are used, in a fashion
1101 compatible with @code{cfront} 1.x. @xref{Code Gen Options,,Options for
1102 Code Generation Conventions}.
1105 @node Warning Options
1106 @section Options to Request or Suppress Warnings
1107 @cindex options to control warnings
1108 @cindex warning messages
1109 @cindex messages, warning
1110 @cindex suppressing warnings
1112 Warnings are diagnostic messages that report constructions which
1113 are not inherently erroneous but which are risky or suggest there
1114 may have been an error.
1116 You can request many specific warnings with options beginning @samp{-W},
1117 for example @samp{-Wimplicit} to request warnings on implicit
1118 declarations. Each of these specific warning options also has a
1119 negative form beginning @samp{-Wno-} to turn off warnings;
1120 for example, @samp{-Wno-implicit}. This manual lists only one of the
1121 two forms, whichever is not the default.
1123 These options control the amount and kinds of warnings produced by GNU
1127 @cindex syntax checking
1129 Check the code for syntax errors, but don't do anything beyond that.
1132 Issue all the warnings demanded by strict ANSI standard C; reject
1133 all programs that use forbidden extensions.
1135 Valid ANSI standard C programs should compile properly with or without
1136 this option (though a rare few will require @samp{-ansi}). However,
1137 without this option, certain GNU extensions and traditional C features
1138 are supported as well. With this option, they are rejected.
1140 @samp{-pedantic} does not cause warning messages for use of the
1141 alternate keywords whose names begin and end with @samp{__}. Pedantic
1142 warnings are also disabled in the expression that follows
1143 @code{__extension__}. However, only system header files should use
1144 these escape routes; application programs should avoid them.
1145 @xref{Alternate Keywords}.
1147 This option is not intended to be @i{useful}; it exists only to satisfy
1148 pedants who would otherwise claim that GNU CC fails to support the ANSI
1151 Some users try to use @samp{-pedantic} to check programs for strict ANSI
1152 C conformance. They soon find that it does not do quite what they want:
1153 it finds some non-ANSI practices, but not all---only those for which
1154 ANSI C @emph{requires} a diagnostic.
1156 A feature to report any failure to conform to ANSI C might be useful in
1157 some instances, but would require considerable additional work and would
1158 be quite different from @samp{-pedantic}. We recommend, rather, that
1159 users take advantage of the extensions of GNU C and disregard the
1160 limitations of other compilers. Aside from certain supercomputers and
1161 obsolete small machines, there is less and less reason ever to use any
1162 other C compiler other than for bootstrapping GNU CC.
1164 @item -pedantic-errors
1165 Like @samp{-pedantic}, except that errors are produced rather than
1169 Inhibit all warning messages.
1172 Inhibit warning messages about the use of @samp{#import}.
1174 @item -Wchar-subscripts
1175 Warn if an array subscript has type @code{char}. This is a common cause
1176 of error, as programmers often forget that this type is signed on some
1180 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1181 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1184 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1185 the arguments supplied have types appropriate to the format string
1189 Warn whenever a function or parameter is implicitly declared,
1190 or when a type implicitly defaults to @code{int}.
1193 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1194 function with external linkage, returning int, taking either zero
1195 arguments, two, or three arguments of appropriate types.
1198 Warn if parentheses are omitted in certain contexts, such
1199 as when there is an assignment in a context where a truth value
1200 is expected, or when operators are nested whose precedence people
1201 often get confused about.
1204 Warn whenever a function is defined with a return-type that defaults
1205 to @code{int}. Also warn about any @code{return} statement with no
1206 return-value in a function whose return-type is not @code{void}.
1209 Warn whenever a @code{switch} statement has an index of enumeral type
1210 and lacks a @code{case} for one or more of the named codes of that
1211 enumeration. (The presence of a @code{default} label prevents this
1212 warning.) @code{case} labels outside the enumeration range also
1213 provoke warnings when this option is used.
1216 Warn if any trigraphs are encountered (assuming they are enabled).
1219 Warn whenever a variable is unused aside from its declaration,
1220 whenever a function is declared static but never defined, whenever a
1221 label is declared but not used, and whenever a statement computes a
1222 result that is explicitly not used.
1224 To suppress this warning for an expression, simply cast it to void. For
1225 unused variables and parameters, use the @samp{unused} attribute
1226 (@pxref{Variable Attributes}).
1228 @item -Wuninitialized
1229 An automatic variable is used without first being initialized.
1231 These warnings are possible only in optimizing compilation,
1232 because they require data flow information that is computed only
1233 when optimizing. If you don't specify @samp{-O}, you simply won't
1236 These warnings occur only for variables that are candidates for
1237 register allocation. Therefore, they do not occur for a variable that
1238 is declared @code{volatile}, or whose address is taken, or whose size
1239 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1240 structures, unions or arrays, even when they are in registers.
1242 Note that there may be no warning about a variable that is used only
1243 to compute a value that itself is never used, because such
1244 computations may be deleted by data flow analysis before the warnings
1247 These warnings are made optional because GNU CC is not smart
1248 enough to see all the reasons why the code might be correct
1249 despite appearing to have an error. Here is one example of how
1268 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1269 always initialized, but GNU CC doesn't know this. Here is
1270 another common case:
1275 if (change_y) save_y = y, y = new_y;
1277 if (change_y) y = save_y;
1282 This has no bug because @code{save_y} is used only if it is set.
1284 Some spurious warnings can be avoided if you declare all the functions
1285 you use that never return as @code{noreturn}. @xref{Function
1288 @item -Wreorder (C++ only)
1289 @cindex reordering, warning
1290 @cindex warning for reordering of member initializers
1291 Warn when the order of member initializers given in the code does not
1292 match the order in which they must be executed. For instance:
1298 A(): j (0), i (1) @{ @}
1302 Here the compiler will warn that the member initializers for @samp{i}
1303 and @samp{j} will be rearranged to match the declaration order of the
1306 @item -Wsign-compare
1307 @cindex warning for comparison of signed and unsigned values
1308 @cindex comparison of signed and unsigned values, warning
1309 @cindex signed and unsigned values, comparison warning
1310 Warn when a comparison between signed and unsigned values could produce
1311 an incorrect result when the signed value is converted to unsigned.
1313 @item -Wtemplate-debugging
1314 @cindex template debugging
1315 When using templates in a C++ program, warn if debugging is not yet
1316 fully available (C++ only).
1319 All of the above @samp{-W} options combined. This enables all the
1320 warnings about constructions that some users consider questionable, and
1321 that are easy to avoid (or modify to prevent the warning), even in
1322 conjunction with macros.
1325 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1326 Some of them warn about constructions that users generally do not
1327 consider questionable, but which occasionally you might wish to check
1328 for; others warn about constructions that are necessary or hard to avoid
1329 in some cases, and there is no simple way to modify the code to suppress
1334 Print extra warning messages for these events:
1337 @cindex @code{longjmp} warnings
1339 A nonvolatile automatic variable might be changed by a call to
1340 @code{longjmp}. These warnings as well are possible only in
1341 optimizing compilation.
1343 The compiler sees only the calls to @code{setjmp}. It cannot know
1344 where @code{longjmp} will be called; in fact, a signal handler could
1345 call it at any point in the code. As a result, you may get a warning
1346 even when there is in fact no problem because @code{longjmp} cannot
1347 in fact be called at the place which would cause a problem.
1350 A function can return either with or without a value. (Falling
1351 off the end of the function body is considered returning without
1352 a value.) For example, this function would evoke such a
1366 An expression-statement or the left-hand side of a comma expression
1367 contains no side effects.
1368 To suppress the warning, cast the unused expression to void.
1369 For example, an expression such as @samp{x[i,j]} will cause a warning,
1370 but @samp{x[(void)i,j]} will not.
1373 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
1376 A comparison like @samp{x<=y<=z} appears; this is equivalent to
1377 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
1378 that of ordinary mathematical notation.
1381 Storage-class specifiers like @code{static} are not the first things in
1382 a declaration. According to the C Standard, this usage is obsolescent.
1385 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
1389 An aggregate has a partly bracketed initializer.
1390 For example, the following code would evoke such a warning,
1391 because braces are missing around the initializer for @code{x.h}:
1394 struct s @{ int f, g; @};
1395 struct t @{ struct s h; int i; @};
1396 struct t x = @{ 1, 2, 3 @};
1401 Warn about certain constructs that behave differently in traditional and
1406 Macro arguments occurring within string constants in the macro body.
1407 These would substitute the argument in traditional C, but are part of
1408 the constant in ANSI C.
1411 A function declared external in one block and then used after the end of
1415 A @code{switch} statement has an operand of type @code{long}.
1419 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
1422 Warn whenever a local variable shadows another local variable.
1424 @item -Wid-clash-@var{len}
1425 Warn whenever two distinct identifiers match in the first @var{len}
1426 characters. This may help you prepare a program that will compile
1427 with certain obsolete, brain-damaged compilers.
1429 @item -Wlarger-than-@var{len}
1430 Warn whenever an object of larger than @var{len} bytes is defined.
1432 @item -Wpointer-arith
1433 Warn about anything that depends on the ``size of'' a function type or
1434 of @code{void}. GNU C assigns these types a size of 1, for
1435 convenience in calculations with @code{void *} pointers and pointers
1438 @item -Wbad-function-cast
1439 Warn whenever a function call is cast to a non-matching type.
1440 For example, warn if @code{int malloc()} is cast to @code{anything *}.
1443 Warn whenever a pointer is cast so as to remove a type qualifier from
1444 the target type. For example, warn if a @code{const char *} is cast
1445 to an ordinary @code{char *}.
1448 Warn whenever a pointer is cast such that the required alignment of the
1449 target is increased. For example, warn if a @code{char *} is cast to
1450 an @code{int *} on machines where integers can only be accessed at
1451 two- or four-byte boundaries.
1453 @item -Wwrite-strings
1454 Give string constants the type @code{const char[@var{length}]} so that
1455 copying the address of one into a non-@code{const} @code{char *}
1456 pointer will get a warning. These warnings will help you find at
1457 compile time code that can try to write into a string constant, but
1458 only if you have been very careful about using @code{const} in
1459 declarations and prototypes. Otherwise, it will just be a nuisance;
1460 this is why we did not make @samp{-Wall} request these warnings.
1463 Warn if a prototype causes a type conversion that is different from what
1464 would happen to the same argument in the absence of a prototype. This
1465 includes conversions of fixed point to floating and vice versa, and
1466 conversions changing the width or signedness of a fixed point argument
1467 except when the same as the default promotion.
1469 Also, warn if a negative integer constant expression is implicitly
1470 converted to an unsigned type. For example, warn about the assignment
1471 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
1472 casts like @code{(unsigned) -1}.
1474 @item -Waggregate-return
1475 Warn if any functions that return structures or unions are defined or
1476 called. (In languages where you can return an array, this also elicits
1479 @item -Wstrict-prototypes
1480 Warn if a function is declared or defined without specifying the
1481 argument types. (An old-style function definition is permitted without
1482 a warning if preceded by a declaration which specifies the argument
1485 @item -Wmissing-prototypes
1486 Warn if a global function is defined without a previous prototype
1487 declaration. This warning is issued even if the definition itself
1488 provides a prototype. The aim is to detect global functions that fail
1489 to be declared in header files.
1491 @item -Wmissing-declarations
1492 Warn if a global function is defined without a previous declaration.
1493 Do so even if the definition itself provides a prototype.
1494 Use this option to detect global functions that are not declared in
1497 @item -Wredundant-decls
1498 Warn if anything is declared more than once in the same scope, even in
1499 cases where multiple declaration is valid and changes nothing.
1501 @item -Wnested-externs
1502 Warn if an @code{extern} declaration is encountered within an function.
1505 Warn if a function can not be inlined, and either it was declared as inline,
1506 or else the @samp{-finline-functions} option was given.
1508 @item -Woverloaded-virtual
1509 @cindex overloaded virtual fn, warning
1510 @cindex warning for overloaded virtual fn
1511 Warn when a derived class function declaration may be an error in
1512 defining a virtual function (C++ only). In a derived class, the
1513 definitions of virtual functions must match the type signature of a
1514 virtual function declared in the base class. With this option, the
1515 compiler warns when you define a function with the same name as a
1516 virtual function, but with a type signature that does not match any
1517 declarations from the base class.
1519 @item -Wsynth (C++ only)
1520 @cindex warning for synthesized methods
1521 @cindex synthesized methods, warning
1522 Warn when g++'s synthesis behavior does not match that of cfront. For
1528 A& operator = (int);
1538 In this example, g++ will synthesize a default @samp{A& operator =
1539 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1542 Make all warnings into errors.
1545 @node Debugging Options
1546 @section Options for Debugging Your Program or GNU CC
1547 @cindex options, debugging
1548 @cindex debugging information options
1550 GNU CC has various special options that are used for debugging
1551 either your program or GCC:
1555 Produce debugging information in the operating system's native format
1556 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
1559 On most systems that use stabs format, @samp{-g} enables use of extra
1560 debugging information that only GDB can use; this extra information
1561 makes debugging work better in GDB but will probably make other debuggers
1563 refuse to read the program. If you want to control for certain whether
1564 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
1565 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
1568 Unlike most other C compilers, GNU CC allows you to use @samp{-g} with
1569 @samp{-O}. The shortcuts taken by optimized code may occasionally
1570 produce surprising results: some variables you declared may not exist
1571 at all; flow of control may briefly move where you did not expect it;
1572 some statements may not be executed because they compute constant
1573 results or their values were already at hand; some statements may
1574 execute in different places because they were moved out of loops.
1576 Nevertheless it proves possible to debug optimized output. This makes
1577 it reasonable to use the optimizer for programs that might have bugs.
1579 The following options are useful when GNU CC is generated with the
1580 capability for more than one debugging format.
1583 Produce debugging information for use by GDB. This means to use the
1584 most expressive format available (DWARF 2, stabs, or the native format
1585 if neither of those are supported), including GDB extensions if at all
1589 Produce debugging information in stabs format (if that is supported),
1590 without GDB extensions. This is the format used by DBX on most BSD
1591 systems. On MIPS, Alpha and System V Release 4 systems this option
1592 produces stabs debugging output which is not understood by DBX or SDB.
1593 On System V Release 4 systems this option requires the GNU assembler.
1596 Produce debugging information in stabs format (if that is supported),
1597 using GNU extensions understood only by the GNU debugger (GDB). The
1598 use of these extensions is likely to make other debuggers crash or
1599 refuse to read the program.
1602 Produce debugging information in COFF format (if that is supported).
1603 This is the format used by SDB on most System V systems prior to
1607 Produce debugging information in XCOFF format (if that is supported).
1608 This is the format used by the DBX debugger on IBM RS/6000 systems.
1611 Produce debugging information in XCOFF format (if that is supported),
1612 using GNU extensions understood only by the GNU debugger (GDB). The
1613 use of these extensions is likely to make other debuggers crash or
1614 refuse to read the program, and may cause assemblers other than the GNU
1615 assembler (GAS) to fail with an error.
1618 Produce debugging information in DWARF version 1 format (if that is
1619 supported). This is the format used by SDB on most System V Release 4
1623 Produce debugging information in DWARF version 1 format (if that is
1624 supported), using GNU extensions understood only by the GNU debugger
1625 (GDB). The use of these extensions is likely to make other debuggers
1626 crash or refuse to read the program.
1629 Produce debugging information in DWARF version 2 format (if that is
1630 supported). This is the format used by DBX on IRIX 6.
1633 @itemx -ggdb@var{level}
1634 @itemx -gstabs@var{level}
1635 @itemx -gcoff@var{level}
1636 @itemx -gxcoff@var{level}
1637 @itemx -gdwarf@var{level}
1638 @itemx -gdwarf-2@var{level}
1639 Request debugging information and also use @var{level} to specify how
1640 much information. The default level is 2.
1642 Level 1 produces minimal information, enough for making backtraces in
1643 parts of the program that you don't plan to debug. This includes
1644 descriptions of functions and external variables, but no information
1645 about local variables and no line numbers.
1647 Level 3 includes extra information, such as all the macro definitions
1648 present in the program. Some debuggers support macro expansion when
1653 Generate extra code to write profile information suitable for the
1654 analysis program @code{prof}. You must use this option when compiling
1655 the source files you want data about, and you must also use it when
1658 @cindex @code{gprof}
1660 Generate extra code to write profile information suitable for the
1661 analysis program @code{gprof}. You must use this option when compiling
1662 the source files you want data about, and you must also use it when
1667 Generate extra code to write profile information for basic blocks, which will
1668 record the number of times each basic block is executed, the basic block start
1669 address, and the function name containing the basic block. If @samp{-g} is
1670 used, the line number and filename of the start of the basic block will also be
1671 recorded. If not overridden by the machine description, the default action is
1672 to append to the text file @file{bb.out}.
1674 This data could be analyzed by a program like @code{tcov}. Note,
1675 however, that the format of the data is not what @code{tcov} expects.
1676 Eventually GNU @code{gprof} should be extended to process this data.
1679 Generate extra code to profile basic blocks. Your executable will
1680 produce output that is a superset of that produced when @samp{-a} is
1681 used. Additional output is the source and target address of the basic
1682 blocks where a jump takes place, the number of times a jump is executed,
1683 and (optionally) the complete sequence of basic blocks being executed.
1684 The output is appended to file @file{bb.out}.
1686 You can examine different profiling aspects without recompilation. Your
1687 execuable will read a list of function names from file @file{bb.in}.
1688 Profiling starts when a function on the list is entered and stops when
1689 that invocation is exited. To exclude a function from profiling, prefix
1690 its name with `-'. If a function name is not unique, you can
1691 disambiguate it by writing it in the form
1692 @samp{/path/filename.d:functionname}. Your executable will write the
1693 available paths and filenames in file @file{bb.out}.
1695 Several function names have a special meaning:
1698 Write source, target and frequency of jumps to file @file{bb.out}.
1699 @item __bb_hidecall__
1700 Exclude function calls from frequency count.
1701 @item __bb_showret__
1702 Include function returns in frequency count.
1704 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
1705 The file will be compressed using the program @samp{gzip}, which must
1706 exist in your @code{PATH}. On systems without the @samp{popen}
1707 function, the file will be named @file{bbtrace} and will not be
1708 compressed. @strong{Profiling for even a few seconds on these systems
1709 will produce a very large file.} Note: @code{__bb_hidecall__} and
1710 @code{__bb_showret__} will not affect the sequence written to
1714 Here's a short example using different profiling parameters
1715 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
1716 1 and 2 and is called twice from block 3 of function @code{main}. After
1717 the calls, block 3 transfers control to block 4 of @code{main}.
1719 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
1720 the following sequence of blocks is written to file @file{bbtrace.gz}:
1721 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
1722 the return is to a point inside the block and not to the top. The
1723 block address 0 always indicates, that control is transferred
1724 to the trace from somewhere outside the observed functions. With
1725 @samp{-foo} added to @file{bb.in}, the blocks of function
1726 @code{foo} are removed from the trace, so only 0 3 4 remains.
1728 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
1729 jump frequencies will be written to file @file{bb.out}. The
1730 frequencies are obtained by constructing a trace of blocks
1731 and incrementing a counter for every neighbouring pair of blocks
1732 in the trace. The trace 0 3 1 2 1 2 4 displays the following
1736 Jump from block 0x0 to block 0x3 executed 1 time(s)
1737 Jump from block 0x3 to block 0x1 executed 1 time(s)
1738 Jump from block 0x1 to block 0x2 executed 2 time(s)
1739 Jump from block 0x2 to block 0x1 executed 1 time(s)
1740 Jump from block 0x2 to block 0x4 executed 1 time(s)
1743 With @code{__bb_hidecall__}, control transfer due to call instructions
1744 is removed from the trace, that is the trace is cut into three parts: 0
1745 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
1746 to return instructions is added to the trace. The trace becomes: 0 3 1
1747 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
1748 written to @file{bbtrace.gz}. It is solely used for counting jump
1751 @item -fprofile-arcs
1752 Instrument @dfn{arcs} during compilation. For each function of your
1753 program, GNU CC creates a program flow graph, then finds a spanning tree
1754 for the graph. Only arcs that are not on the spanning tree have to be
1755 instrumented: the compiler adds code to count the number of times that these
1756 arcs are executed. When an arc is the only exit or only entrance to a
1757 block, the instrumentation code can be added to the block; otherwise, a
1758 new basic block must be created to hold the instrumentation code.
1760 Since not every arc in the program must be instrumented, programs
1761 compiled with this option run faster than programs compiled with
1762 @samp{-a}, which adds instrumentation code to every basic block in the
1763 program. The tradeoff: since @code{gcov} does not have
1764 execution counts for all branches, it must start with the execution
1765 counts for the instrumented branches, and then iterate over the program
1766 flow graph until the entire graph has been solved. Hence, @code{gcov}
1767 runs a little more slowly than a program which uses information from
1770 @samp{-fprofile-arcs} also makes it possible to estimate branch
1771 probabilities, and to calculate basic block execution counts. In
1772 general, basic block execution counts do not give enough information to
1773 estimate all branch probabilities. When the compiled program exits, it
1774 saves the arc execution counts to a file called
1775 @file{@var{sourcename}.da}. Use the compiler option
1776 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
1777 Control Optimization}) when recompiling, to optimize using estimated
1778 branch probabilities.
1781 @item -ftest-coverage
1782 Create data files for the @code{gcov} code-coverage utility
1783 (@pxref{Gcov,, @code{gcov}: a GNU CC Test Coverage Program}).
1784 The data file names begin with the name of your source file:
1787 @item @var{sourcename}.bb
1788 A mapping from basic blocks to line numbers, which @code{gcov} uses to
1789 associate basic block execution counts with line numbers.
1791 @item @var{sourcename}.bbg
1792 A list of all arcs in the program flow graph. This allows @code{gcov}
1793 to reconstruct the program flow graph, so that it can compute all basic
1794 block and arc execution counts from the information in the
1795 @code{@var{sourcename}.da} file (this last file is the output from
1796 @samp{-fprofile-arcs}).
1799 @item -d@var{letters}
1800 Says to make debugging dumps during compilation at times specified by
1801 @var{letters}. This is used for debugging the compiler. The file names
1802 for most of the dumps are made by appending a word to the source file
1803 name (e.g. @file{foo.c.rtl} or @file{foo.c.jump}). Here are the
1804 possible letters for use in @var{letters}, and their meanings:
1808 Dump all macro definitions, at the end of preprocessing, and write no
1811 Dump all macro names, at the end of preprocessing.
1813 Dump all macro definitions, at the end of preprocessing, in addition to
1816 Dump debugging information during parsing, to standard error.
1818 Dump after RTL generation, to @file{@var{file}.rtl}.
1820 Just generate RTL for a function instead of compiling it. Usually used
1823 Dump after first jump optimization, to @file{@var{file}.jump}.
1825 Dump after CSE (including the jump optimization that sometimes
1826 follows CSE), to @file{@var{file}.cse}.
1828 Dump after loop optimization, to @file{@var{file}.loop}.
1830 Dump after the second CSE pass (including the jump optimization that
1831 sometimes follows CSE), to @file{@var{file}.cse2}.
1833 Dump after flow analysis, to @file{@var{file}.flow}.
1835 Dump after instruction combination, to the file
1836 @file{@var{file}.combine}.
1838 Dump after the first instruction scheduling pass, to
1839 @file{@var{file}.sched}.
1841 Dump after local register allocation, to
1842 @file{@var{file}.lreg}.
1844 Dump after global register allocation, to
1845 @file{@var{file}.greg}.
1847 Dump after the second instruction scheduling pass, to
1848 @file{@var{file}.sched2}.
1850 Dump after last jump optimization, to @file{@var{file}.jump2}.
1852 Dump after delayed branch scheduling, to @file{@var{file}.dbr}.
1854 Dump after conversion from registers to stack, to @file{@var{file}.stack}.
1856 Produce all the dumps listed above.
1858 Print statistics on memory usage, at the end of the run, to
1861 Annotate the assembler output with a comment indicating which
1862 pattern and alternative was used.
1864 Annotate the assembler output with miscellaneous debugging information.
1867 @item -fpretend-float
1868 When running a cross-compiler, pretend that the target machine uses the
1869 same floating point format as the host machine. This causes incorrect
1870 output of the actual floating constants, but the actual instruction
1871 sequence will probably be the same as GNU CC would make when running on
1875 Store the usual ``temporary'' intermediate files permanently; place them
1876 in the current directory and name them based on the source file. Thus,
1877 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
1878 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}.
1880 @item -print-file-name=@var{library}
1881 Print the full absolute name of the library file @var{library} that
1882 would be used when linking---and don't do anything else. With this
1883 option, GNU CC does not compile or link anything; it just prints the
1886 @item -print-prog-name=@var{program}
1887 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
1889 @item -print-libgcc-file-name
1890 Same as @samp{-print-file-name=libgcc.a}.
1892 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
1893 but you do want to link with @file{libgcc.a}. You can do
1896 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
1899 @item -print-search-dirs
1900 Print the name of the configured installation directory and a list of
1901 program and library directories gcc will search---and don't do anything else.
1903 This is useful when gcc prints the error message
1904 @samp{installation problem, cannot exec cpp: No such file or directory}.
1905 To resolve this you either need to put @file{cpp} and the other compiler
1906 components where gcc expects to find them, or you can set the environment
1907 variable @code{GCC_EXEC_PREFIX} to the directory where you installed them.
1908 Don't forget the trailing '/'.
1909 @xref{Environment Variables}.
1912 @node Optimize Options
1913 @section Options That Control Optimization
1914 @cindex optimize options
1915 @cindex options, optimization
1917 These options control various sorts of optimizations:
1922 Optimize. Optimizing compilation takes somewhat more time, and a lot
1923 more memory for a large function.
1925 Without @samp{-O}, the compiler's goal is to reduce the cost of
1926 compilation and to make debugging produce the expected results.
1927 Statements are independent: if you stop the program with a breakpoint
1928 between statements, you can then assign a new value to any variable or
1929 change the program counter to any other statement in the function and
1930 get exactly the results you would expect from the source code.
1932 Without @samp{-O}, the compiler only allocates variables declared
1933 @code{register} in registers. The resulting compiled code is a little
1934 worse than produced by PCC without @samp{-O}.
1936 With @samp{-O}, the compiler tries to reduce code size and execution
1939 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
1940 and @samp{-fdefer-pop} on all machines. The compiler turns on
1941 @samp{-fdelayed-branch} on machines that have delay slots, and
1942 @samp{-fomit-frame-pointer} on machines that can support debugging even
1943 without a frame pointer. On some machines the compiler also turns
1944 on other flags.@refill
1947 Optimize even more. GNU CC performs nearly all supported optimizations
1948 that do not involve a space-speed tradeoff. The compiler does not
1949 perform loop unrolling or function inlining when you specify @samp{-O2}.
1950 As compared to @samp{-O}, this option increases both compilation time
1951 and the performance of the generated code.
1953 @samp{-O2} turns on all optional optimizations except for loop unrolling
1954 and function inlining. It also turns on the @samp{-fforce-mem} option
1955 on all machines and frame pointer elimination on machines where doing so
1956 does not interfere with debugging.
1959 Optimize yet more. @samp{-O3} turns on all optimizations specified by
1960 @samp{-O2} and also turns on the @samp{inline-functions} option.
1965 If you use multiple @samp{-O} options, with or without level numbers,
1966 the last such option is the one that is effective.
1969 Options of the form @samp{-f@var{flag}} specify machine-independent
1970 flags. Most flags have both positive and negative forms; the negative
1971 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
1972 only one of the forms is listed---the one which is not the default.
1973 You can figure out the other form by either removing @samp{no-} or
1978 Do not store floating point variables in registers, and inhibit other
1979 options that might change whether a floating point value is taken from a
1982 @cindex floating point precision
1983 This option prevents undesirable excess precision on machines such as
1984 the 68000 where the floating registers (of the 68881) keep more
1985 precision than a @code{double} is supposed to have. Similarly for the
1986 x86 architecture. For most programs, the excess precision does only
1987 good, but a few programs rely on the precise definition of IEEE floating
1988 point. Use @samp{-ffloat-store} for such programs.
1990 @item -fno-default-inline
1991 Do not make member functions inline by default merely because they are
1992 defined inside the class scope (C++ only). Otherwise, when you specify
1993 @w{@samp{-O}}, member functions defined inside class scope are compiled
1994 inline by default; i.e., you don't need to add @samp{inline} in front of
1995 the member function name.
1997 @item -fno-defer-pop
1998 Always pop the arguments to each function call as soon as that function
1999 returns. For machines which must pop arguments after a function call,
2000 the compiler normally lets arguments accumulate on the stack for several
2001 function calls and pops them all at once.
2004 Force memory operands to be copied into registers before doing
2005 arithmetic on them. This produces better code by making all memory
2006 references potential common subexpressions. When they are not common
2007 subexpressions, instruction combination should eliminate the separate
2008 register-load. The @samp{-O2} option turns on this option.
2011 Force memory address constants to be copied into registers before
2012 doing arithmetic on them. This may produce better code just as
2013 @samp{-fforce-mem} may.
2015 @item -fomit-frame-pointer
2016 Don't keep the frame pointer in a register for functions that
2017 don't need one. This avoids the instructions to save, set up and
2018 restore frame pointers; it also makes an extra register available
2019 in many functions. @strong{It also makes debugging impossible on
2023 On some machines, such as the Vax, this flag has no effect, because
2024 the standard calling sequence automatically handles the frame pointer
2025 and nothing is saved by pretending it doesn't exist. The
2026 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2027 whether a target machine supports this flag. @xref{Registers}.@refill
2030 On some machines, such as the Vax, this flag has no effect, because
2031 the standard calling sequence automatically handles the frame pointer
2032 and nothing is saved by pretending it doesn't exist. The
2033 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2034 whether a target machine supports this flag. @xref{Registers,,Register
2035 Usage, gcc.info, Using and Porting GCC}.@refill
2039 Don't pay attention to the @code{inline} keyword. Normally this option
2040 is used to keep the compiler from expanding any functions inline.
2041 Note that if you are not optimizing, no functions can be expanded inline.
2043 @item -finline-functions
2044 Integrate all simple functions into their callers. The compiler
2045 heuristically decides which functions are simple enough to be worth
2046 integrating in this way.
2048 If all calls to a given function are integrated, and the function is
2049 declared @code{static}, then the function is normally not output as
2050 assembler code in its own right.
2052 @item -fkeep-inline-functions
2053 Even if all calls to a given function are integrated, and the function
2054 is declared @code{static}, nevertheless output a separate run-time
2055 callable version of the function. This switch does not affect
2056 @code{extern inline} functions.
2058 @item -fkeep-static-consts
2059 Emit variables declared @code{static const} when optimization isn't turned
2060 on, even if the variables aren't referenced.
2062 GNU CC enables this option by default. If you want to force the compiler to
2063 check if the variable was referenced, regardless of whether or not
2064 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2066 @item -fno-function-cse
2067 Do not put function addresses in registers; make each instruction that
2068 calls a constant function contain the function's address explicitly.
2070 This option results in less efficient code, but some strange hacks
2071 that alter the assembler output may be confused by the optimizations
2072 performed when this option is not used.
2075 This option allows GCC to violate some ANSI or IEEE rules and/or
2076 specifications in the interest of optimizing code for speed. For
2077 example, it allows the compiler to assume arguments to the @code{sqrt}
2078 function are non-negative numbers and that no floating-point values
2081 This option should never be turned on by any @samp{-O} option since
2082 it can result in incorrect output for programs which depend on
2083 an exact implementation of IEEE or ANSI rules/specifications for
2087 @c following causes underfulls.. they don't look great, but we deal.
2089 The following options control specific optimizations. The @samp{-O2}
2090 option turns on all of these optimizations except @samp{-funroll-loops}
2091 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
2092 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
2093 but specific machines may handle it differently.
2095 You can use the following flags in the rare cases when ``fine-tuning''
2096 of optimizations to be performed is desired.
2099 @item -fstrength-reduce
2100 Perform the optimizations of loop strength reduction and
2101 elimination of iteration variables.
2103 @item -fthread-jumps
2104 Perform optimizations where we check to see if a jump branches to a
2105 location where another comparison subsumed by the first is found. If
2106 so, the first branch is redirected to either the destination of the
2107 second branch or a point immediately following it, depending on whether
2108 the condition is known to be true or false.
2110 @item -fcse-follow-jumps
2111 In common subexpression elimination, scan through jump instructions
2112 when the target of the jump is not reached by any other path. For
2113 example, when CSE encounters an @code{if} statement with an
2114 @code{else} clause, CSE will follow the jump when the condition
2117 @item -fcse-skip-blocks
2118 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
2119 follow jumps which conditionally skip over blocks. When CSE
2120 encounters a simple @code{if} statement with no else clause,
2121 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
2122 body of the @code{if}.
2124 @item -frerun-cse-after-loop
2125 Re-run common subexpression elimination after loop optimizations has been
2128 @item -fexpensive-optimizations
2129 Perform a number of minor optimizations that are relatively expensive.
2131 @item -fdelayed-branch
2132 If supported for the target machine, attempt to reorder instructions
2133 to exploit instruction slots available after delayed branch
2136 @item -fschedule-insns
2137 If supported for the target machine, attempt to reorder instructions to
2138 eliminate execution stalls due to required data being unavailable. This
2139 helps machines that have slow floating point or memory load instructions
2140 by allowing other instructions to be issued until the result of the load
2141 or floating point instruction is required.
2143 @item -fschedule-insns2
2144 Similar to @samp{-fschedule-insns}, but requests an additional pass of
2145 instruction scheduling after register allocation has been done. This is
2146 especially useful on machines with a relatively small number of
2147 registers and where memory load instructions take more than one cycle.
2149 @item -ffunction-sections
2150 Place each function into its own section in the output file if the
2151 target supports arbitrary sections. The function's name determines
2152 the section's name in the output file.
2154 Use this option on systems where the linker can perform optimizations
2155 to improve locality of reference in the instruction space. HPPA
2156 processors running HP-UX and Sparc processors running Solaris 2 have
2157 linkers with such optimizations. Other systems using the ELF object format
2158 as well as AIX may have these optimizations in the future.
2160 Only use this option when there are significant benefits from doing
2161 so. When you specify this option, the assembler and linker will
2162 create larger object and executable files and will also be slower.
2163 You will not be able to use @code{gprof} on all systems if you
2164 specify this option and you may have problems with debugging if
2165 you specify both this option and @samp{-g}.
2167 @item -fcaller-saves
2168 Enable values to be allocated in registers that will be clobbered by
2169 function calls, by emitting extra instructions to save and restore the
2170 registers around such calls. Such allocation is done only when it
2171 seems to result in better code than would otherwise be produced.
2173 This option is enabled by default on certain machines, usually those
2174 which have no call-preserved registers to use instead.
2176 @item -funroll-loops
2177 Perform the optimization of loop unrolling. This is only done for loops
2178 whose number of iterations can be determined at compile time or run time.
2179 @samp{-funroll-loop} implies both @samp{-fstrength-reduce} and
2180 @samp{-frerun-cse-after-loop}.
2182 @item -funroll-all-loops
2183 Perform the optimization of loop unrolling. This is done for all loops
2184 and usually makes programs run more slowly. @samp{-funroll-all-loops}
2185 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
2187 @item -fmove-all-movables
2188 Forces all invariant computations in loops to be moved
2191 @item -freduce-all-givs
2192 Forces all general-induction variables in loops to be
2195 @emph{Note:} When compiling programs written in Fortran,
2196 @samp{-fmove-all-moveables} and @samp{-freduce-all-givs} are enabled
2197 by default when you use the optimizer.
2199 These options may generate better or worse code; results are highly
2200 dependent on the structure of loops within the source code.
2202 These two options are intended to be removed someday, once
2203 they have helped determine the efficacy of various
2204 approaches to improving loop optimizations.
2206 Please let us (@code{egcs@@cygnus.com} and @code{fortran@@gnu.ai.mit.edu})
2207 know how use of these options affects
2208 the performance of your production code.
2209 We're very interested in code that runs @emph{slower}
2210 when these options are @emph{enabled}.
2213 Disable any machine-specific peephole optimizations.
2215 @item -fbranch-probabilities
2216 After running a program compiled with @samp{-fprofile-arcs}
2217 (@pxref{Debugging Options,, Options for Debugging Your Program or
2218 @code{gcc}}), you can compile it a second time using
2219 @samp{-fbranch-probabilities}, to improve optimizations based on
2220 guessing the path a branch might take.
2223 With @samp{-fbranch-probabilities}, GNU CC puts a @samp{REG_EXEC_COUNT}
2224 note on the first instruction of each basic block, and a
2225 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
2226 These can be used to improve optimization. Currently, they are only
2227 used in one place: in @file{reorg.c}, instead of guessing which path a
2228 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
2229 exactly determine which path is taken more often.
2233 Some machines only support 2 operands per instruction. On such
2234 machines, GNU CC might have to do extra copies. The @samp{-fregmove}
2235 option overrides the default for the machine to do the copy before
2236 register allocation.
2239 @node Preprocessor Options
2240 @section Options Controlling the Preprocessor
2241 @cindex preprocessor options
2242 @cindex options, preprocessor
2244 These options control the C preprocessor, which is run on each C source
2245 file before actual compilation.
2247 If you use the @samp{-E} option, nothing is done except preprocessing.
2248 Some of these options make sense only together with @samp{-E} because
2249 they cause the preprocessor output to be unsuitable for actual
2253 @item -include @var{file}
2254 Process @var{file} as input before processing the regular input file.
2255 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
2256 and @samp{-U} options on the command line are always processed before
2257 @samp{-include @var{file}}, regardless of the order in which they are
2258 written. All the @samp{-include} and @samp{-imacros} options are
2259 processed in the order in which they are written.
2261 @item -imacros @var{file}
2262 Process @var{file} as input, discarding the resulting output, before
2263 processing the regular input file. Because the output generated from
2264 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
2265 is to make the macros defined in @var{file} available for use in the
2268 Any @samp{-D} and @samp{-U} options on the command line are always
2269 processed before @samp{-imacros @var{file}}, regardless of the order in
2270 which they are written. All the @samp{-include} and @samp{-imacros}
2271 options are processed in the order in which they are written.
2273 @item -idirafter @var{dir}
2274 @cindex second include path
2275 Add the directory @var{dir} to the second include path. The directories
2276 on the second include path are searched when a header file is not found
2277 in any of the directories in the main include path (the one that
2280 @item -iprefix @var{prefix}
2281 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
2284 @item -iwithprefix @var{dir}
2285 Add a directory to the second include path. The directory's name is
2286 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
2287 specified previously with @samp{-iprefix}. If you have not specified a
2288 prefix yet, the directory containing the installed passes of the
2289 compiler is used as the default.
2291 @item -iwithprefixbefore @var{dir}
2292 Add a directory to the main include path. The directory's name is made
2293 by concatenating @var{prefix} and @var{dir}, as in the case of
2294 @samp{-iwithprefix}.
2296 @item -isystem @var{dir}
2297 Add a directory to the beginning of the second include path, marking it
2298 as a system directory, so that it gets the same special treatment as
2299 is applied to the standard system directories.
2302 Do not search the standard system directories for header files. Only
2303 the directories you have specified with @samp{-I} options (and the
2304 current directory, if appropriate) are searched. @xref{Directory
2305 Options}, for information on @samp{-I}.
2307 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
2308 search path to only those directories you specify explicitly.
2311 Do not predefine any nonstandard macros. (Including architecture flags).
2314 Run only the C preprocessor. Preprocess all the C source files
2315 specified and output the results to standard output or to the
2316 specified output file.
2319 Tell the preprocessor not to discard comments. Used with the
2323 Tell the preprocessor not to generate @samp{#line} directives.
2324 Used with the @samp{-E} option.
2327 @cindex dependencies, make
2329 Tell the preprocessor to output a rule suitable for @code{make}
2330 describing the dependencies of each object file. For each source file,
2331 the preprocessor outputs one @code{make}-rule whose target is the object
2332 file name for that source file and whose dependencies are all the
2333 @code{#include} header files it uses. This rule may be a single line or
2334 may be continued with @samp{\}-newline if it is long. The list of rules
2335 is printed on standard output instead of the preprocessed C program.
2337 @samp{-M} implies @samp{-E}.
2339 Another way to specify output of a @code{make} rule is by setting
2340 the environment variable @code{DEPENDENCIES_OUTPUT} (@pxref{Environment
2344 Like @samp{-M} but the output mentions only the user header files
2345 included with @samp{#include "@var{file}"}. System header files
2346 included with @samp{#include <@var{file}>} are omitted.
2349 Like @samp{-M} but the dependency information is written to a file made by
2350 replacing ".c" with ".d" at the end of the input file names.
2351 This is in addition to compiling the file as specified---@samp{-MD} does
2352 not inhibit ordinary compilation the way @samp{-M} does.
2354 In Mach, you can use the utility @code{md} to merge multiple dependency
2355 files into a single dependency file suitable for using with the @samp{make}
2359 Like @samp{-MD} except mention only user header files, not system
2363 Treat missing header files as generated files and assume they live in the
2364 same directory as the source file. If you specify @samp{-MG}, you
2365 must also specify either @samp{-M} or @samp{-MM}. @samp{-MG} is not
2366 supported with @samp{-MD} or @samp{-MMD}.
2369 Print the name of each header file used, in addition to other normal
2372 @item -A@var{question}(@var{answer})
2373 Assert the answer @var{answer} for @var{question}, in case it is tested
2374 with a preprocessing conditional such as @samp{#if
2375 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
2376 assertions that normally describe the target machine.
2379 Define macro @var{macro} with the string @samp{1} as its definition.
2381 @item -D@var{macro}=@var{defn}
2382 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
2383 the command line are processed before any @samp{-U} options.
2386 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
2387 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
2391 Tell the preprocessor to output only a list of the macro definitions
2392 that are in effect at the end of preprocessing. Used with the @samp{-E}
2396 Tell the preprocessing to pass all macro definitions into the output, in
2397 their proper sequence in the rest of the output.
2400 Like @samp{-dD} except that the macro arguments and contents are omitted.
2401 Only @samp{#define @var{name}} is included in the output.
2404 Support ANSI C trigraphs. The @samp{-ansi} option also has this effect.
2406 @item -Wp,@var{option}
2407 Pass @var{option} as an option to the preprocessor. If @var{option}
2408 contains commas, it is split into multiple options at the commas.
2411 @node Assembler Options
2412 @section Passing Options to the Assembler
2414 @c prevent bad page break with this line
2415 You can pass options to the assembler.
2418 @item -Wa,@var{option}
2419 Pass @var{option} as an option to the assembler. If @var{option}
2420 contains commas, it is split into multiple options at the commas.
2424 @section Options for Linking
2425 @cindex link options
2426 @cindex options, linking
2428 These options come into play when the compiler links object files into
2429 an executable output file. They are meaningless if the compiler is
2430 not doing a link step.
2434 @item @var{object-file-name}
2435 A file name that does not end in a special recognized suffix is
2436 considered to name an object file or library. (Object files are
2437 distinguished from libraries by the linker according to the file
2438 contents.) If linking is done, these object files are used as input
2444 If any of these options is used, then the linker is not run, and
2445 object file names should not be used as arguments. @xref{Overall
2449 @item -l@var{library}
2450 Search the library named @var{library} when linking.
2452 It makes a difference where in the command you write this option; the
2453 linker searches processes libraries and object files in the order they
2454 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
2455 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
2456 to functions in @samp{z}, those functions may not be loaded.
2458 The linker searches a standard list of directories for the library,
2459 which is actually a file named @file{lib@var{library}.a}. The linker
2460 then uses this file as if it had been specified precisely by name.
2462 The directories searched include several standard system directories
2463 plus any that you specify with @samp{-L}.
2465 Normally the files found this way are library files---archive files
2466 whose members are object files. The linker handles an archive file by
2467 scanning through it for members which define symbols that have so far
2468 been referenced but not defined. But if the file that is found is an
2469 ordinary object file, it is linked in the usual fashion. The only
2470 difference between using an @samp{-l} option and specifying a file name
2471 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
2472 and searches several directories.
2475 You need this special case of the @samp{-l} option in order to
2476 link an Objective C program.
2479 Do not use the standard system startup files when linking.
2480 The standard system libraries are used normally, unless @code{-nostdlib}
2481 or @code{-nodefaultlibs} is used.
2483 @item -nodefaultlibs
2484 Do not use the standard system libraries when linking.
2485 Only the libraries you specify will be passed to the linker.
2486 The standard startup files are used normally, unless @code{-nostartfiles}
2490 Do not use the standard system startup files or libraries when linking.
2491 No startup files and only the libraries you specify will be passed to
2494 @cindex @code{-lgcc}, use with @code{-nostdlib}
2495 @cindex @code{-nostdlib} and unresolved references
2496 @cindex unresolved references and @code{-nostdlib}
2497 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
2498 @cindex @code{-nodefaultlibs} and unresolved references
2499 @cindex unresolved references and @code{-nodefaultlibs}
2500 One of the standard libraries bypassed by @samp{-nostdlib} and
2501 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
2502 that GNU CC uses to overcome shortcomings of particular machines, or special
2503 needs for some languages.
2505 (@xref{Interface,,Interfacing to GNU CC Output}, for more discussion of
2509 (@xref{Interface,,Interfacing to GNU CC Output,gcc.info,Porting GNU CC},
2510 for more discussion of @file{libgcc.a}.)
2512 In most cases, you need @file{libgcc.a} even when you want to avoid
2513 other standard libraries. In other words, when you specify @samp{-nostdlib}
2514 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
2515 This ensures that you have no unresolved references to internal GNU CC
2516 library subroutines. (For example, @samp{__main}, used to ensure C++
2517 constructors will be called; @pxref{Collect2,,@code{collect2}}.)
2520 Remove all symbol table and relocation information from the executable.
2523 On systems that support dynamic linking, this prevents linking with the shared
2524 libraries. On other systems, this option has no effect.
2527 Produce a shared object which can then be linked with other objects to
2528 form an executable. Not all systems support this option. You must
2529 also specify @samp{-fpic} or @samp{-fPIC} on some systems when
2530 you specify this option.
2533 Bind references to global symbols when building a shared object. Warn
2534 about any unresolved references (unless overridden by the link editor
2535 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
2538 @item -Xlinker @var{option}
2539 Pass @var{option} as an option to the linker. You can use this to
2540 supply system-specific linker options which GNU CC does not know how to
2543 If you want to pass an option that takes an argument, you must use
2544 @samp{-Xlinker} twice, once for the option and once for the argument.
2545 For example, to pass @samp{-assert definitions}, you must write
2546 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
2547 @samp{-Xlinker "-assert definitions"}, because this passes the entire
2548 string as a single argument, which is not what the linker expects.
2550 @item -Wl,@var{option}
2551 Pass @var{option} as an option to the linker. If @var{option} contains
2552 commas, it is split into multiple options at the commas.
2554 @item -u @var{symbol}
2555 Pretend the symbol @var{symbol} is undefined, to force linking of
2556 library modules to define it. You can use @samp{-u} multiple times with
2557 different symbols to force loading of additional library modules.
2560 @node Directory Options
2561 @section Options for Directory Search
2562 @cindex directory options
2563 @cindex options, directory search
2566 These options specify directories to search for header files, for
2567 libraries and for parts of the compiler:
2571 Add the directory @var{dir} to the head of the list of directories to be
2572 searched for header files. This can be used to override a system header
2573 file, substituting your own version, since these directories are
2574 searched before the system header file directories. If you use more
2575 than one @samp{-I} option, the directories are scanned in left-to-right
2576 order; the standard system directories come after.
2579 Any directories you specify with @samp{-I} options before the @samp{-I-}
2580 option are searched only for the case of @samp{#include "@var{file}"};
2581 they are not searched for @samp{#include <@var{file}>}.
2583 If additional directories are specified with @samp{-I} options after
2584 the @samp{-I-}, these directories are searched for all @samp{#include}
2585 directives. (Ordinarily @emph{all} @samp{-I} directories are used
2588 In addition, the @samp{-I-} option inhibits the use of the current
2589 directory (where the current input file came from) as the first search
2590 directory for @samp{#include "@var{file}"}. There is no way to
2591 override this effect of @samp{-I-}. With @samp{-I.} you can specify
2592 searching the directory which was current when the compiler was
2593 invoked. That is not exactly the same as what the preprocessor does
2594 by default, but it is often satisfactory.
2596 @samp{-I-} does not inhibit the use of the standard system directories
2597 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
2601 Add directory @var{dir} to the list of directories to be searched
2604 @item -B@var{prefix}
2605 This option specifies where to find the executables, libraries,
2606 include files, and data files of the compiler itself.
2608 The compiler driver program runs one or more of the subprograms
2609 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
2610 @var{prefix} as a prefix for each program it tries to run, both with and
2611 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
2613 For each subprogram to be run, the compiler driver first tries the
2614 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
2615 was not specified, the driver tries two standard prefixes, which are
2616 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
2617 those results in a file name that is found, the unmodified program
2618 name is searched for using the directories specified in your
2619 @samp{PATH} environment variable.
2621 @samp{-B} prefixes that effectively specify directory names also apply
2622 to libraries in the linker, because the compiler translates these
2623 options into @samp{-L} options for the linker. They also apply to
2624 includes files in the preprocessor, because the compiler translates these
2625 options into @samp{-isystem} options for the preprocessor. In this case,
2626 the compiler appends @samp{include} to the prefix.
2628 The run-time support file @file{libgcc.a} can also be searched for using
2629 the @samp{-B} prefix, if needed. If it is not found there, the two
2630 standard prefixes above are tried, and that is all. The file is left
2631 out of the link if it is not found by those means.
2633 Another way to specify a prefix much like the @samp{-B} prefix is to use
2634 the environment variable @code{GCC_EXEC_PREFIX}. @xref{Environment
2637 @item -specs=@var{file}
2638 Process @var{file} after the compiler reads in the standard @file{specs}
2639 file, in order to override the defaults that the @file{gcc} driver
2640 program uses when determining what switches to pass to @file{cc1},
2641 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
2642 @samp{-specs=}@var{file} can be specified on the command line, and they
2643 are processed in order, from left to right.
2646 @node Target Options
2647 @section Specifying Target Machine and Compiler Version
2648 @cindex target options
2649 @cindex cross compiling
2650 @cindex specifying machine version
2651 @cindex specifying compiler version and target machine
2652 @cindex compiler version, specifying
2653 @cindex target machine, specifying
2655 By default, GNU CC compiles code for the same type of machine that you
2656 are using. However, it can also be installed as a cross-compiler, to
2657 compile for some other type of machine. In fact, several different
2658 configurations of GNU CC, for different target machines, can be
2659 installed side by side. Then you specify which one to use with the
2662 In addition, older and newer versions of GNU CC can be installed side
2663 by side. One of them (probably the newest) will be the default, but
2664 you may sometimes wish to use another.
2667 @item -b @var{machine}
2668 The argument @var{machine} specifies the target machine for compilation.
2669 This is useful when you have installed GNU CC as a cross-compiler.
2671 The value to use for @var{machine} is the same as was specified as the
2672 machine type when configuring GNU CC as a cross-compiler. For
2673 example, if a cross-compiler was configured with @samp{configure
2674 i386v}, meaning to compile for an 80386 running System V, then you
2675 would specify @samp{-b i386v} to run that cross compiler.
2677 When you do not specify @samp{-b}, it normally means to compile for
2678 the same type of machine that you are using.
2680 @item -V @var{version}
2681 The argument @var{version} specifies which version of GNU CC to run.
2682 This is useful when multiple versions are installed. For example,
2683 @var{version} might be @samp{2.0}, meaning to run GNU CC version 2.0.
2685 The default version, when you do not specify @samp{-V}, is the last
2686 version of GNU CC that you installed.
2689 The @samp{-b} and @samp{-V} options actually work by controlling part of
2690 the file name used for the executable files and libraries used for
2691 compilation. A given version of GNU CC, for a given target machine, is
2692 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
2694 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
2695 changing the names of these directories or adding alternate names (or
2696 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
2697 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
2698 80386} becomes an alias for @samp{-b i386v}.
2700 In one respect, the @samp{-b} or @samp{-V} do not completely change
2701 to a different compiler: the top-level driver program @code{gcc}
2702 that you originally invoked continues to run and invoke the other
2703 executables (preprocessor, compiler per se, assembler and linker)
2704 that do the real work. However, since no real work is done in the
2705 driver program, it usually does not matter that the driver program
2706 in use is not the one for the specified target and version.
2708 The only way that the driver program depends on the target machine is
2709 in the parsing and handling of special machine-specific options.
2710 However, this is controlled by a file which is found, along with the
2711 other executables, in the directory for the specified version and
2712 target machine. As a result, a single installed driver program adapts
2713 to any specified target machine and compiler version.
2715 The driver program executable does control one significant thing,
2716 however: the default version and target machine. Therefore, you can
2717 install different instances of the driver program, compiled for
2718 different targets or versions, under different names.
2720 For example, if the driver for version 2.0 is installed as @code{ogcc}
2721 and that for version 2.1 is installed as @code{gcc}, then the command
2722 @code{gcc} will use version 2.1 by default, while @code{ogcc} will use
2723 2.0 by default. However, you can choose either version with either
2724 command with the @samp{-V} option.
2726 @node Submodel Options
2727 @section Hardware Models and Configurations
2728 @cindex submodel options
2729 @cindex specifying hardware config
2730 @cindex hardware models and configurations, specifying
2731 @cindex machine dependent options
2733 Earlier we discussed the standard option @samp{-b} which chooses among
2734 different installed compilers for completely different target
2735 machines, such as Vax vs. 68000 vs. 80386.
2737 In addition, each of these target machine types can have its own
2738 special options, starting with @samp{-m}, to choose among various
2739 hardware models or configurations---for example, 68010 vs 68020,
2740 floating coprocessor or none. A single installed version of the
2741 compiler can compile for any model or configuration, according to the
2744 Some configurations of the compiler also support additional special
2745 options, usually for compatibility with other compilers on the same
2749 These options are defined by the macro @code{TARGET_SWITCHES} in the
2750 machine description. The default for the options is also defined by
2751 that macro, which enables you to change the defaults.
2763 * RS/6000 and PowerPC Options::
2768 * Intel 960 Options::
2769 * DEC Alpha Options::
2773 * System V Options::
2777 @node M680x0 Options
2778 @subsection M680x0 Options
2779 @cindex M680x0 options
2781 These are the @samp{-m} options defined for the 68000 series. The default
2782 values for these options depends on which style of 68000 was selected when
2783 the compiler was configured; the defaults for the most common choices are
2789 Generate output for a 68000. This is the default
2790 when the compiler is configured for 68000-based systems.
2794 Generate output for a 68020. This is the default
2795 when the compiler is configured for 68020-based systems.
2798 Generate output containing 68881 instructions for floating point.
2799 This is the default for most 68020 systems unless @samp{-nfp} was
2800 specified when the compiler was configured.
2803 Generate output for a 68030. This is the default when the compiler is
2804 configured for 68030-based systems.
2807 Generate output for a 68040. This is the default when the compiler is
2808 configured for 68040-based systems.
2810 This option inhibits the use of 68881/68882 instructions that have to be
2811 emulated by software on the 68040. If your 68040 does not have code to
2812 emulate those instructions, use @samp{-m68040}.
2815 Generate output for a 68060. This is the default when the compiler is
2816 configured for 68060-based systems.
2818 This option inhibits the use of 68020 and 68881/68882 instructions that
2819 have to be emulated by software on the 68060. If your 68060 does not
2820 have code to emulate those instructions, use @samp{-m68060}.
2823 Generate output for a 520X "coldfire" family cpu. This is the default
2824 when the compiler is configured for 520X-based systems.
2828 Generate output for a 68040, without using any of the new instructions.
2829 This results in code which can run relatively efficiently on either a
2830 68020/68881 or a 68030 or a 68040. The generated code does use the
2831 68881 instructions that are emulated on the 68040.
2834 Generate output for a 68060, without using any of the new instructions.
2835 This results in code which can run relatively efficiently on either a
2836 68020/68881 or a 68030 or a 68040. The generated code does use the
2837 68881 instructions that are emulated on the 68060.
2840 Generate output containing Sun FPA instructions for floating point.
2843 Generate output containing library calls for floating point.
2844 @strong{Warning:} the requisite libraries are not available for all m68k
2845 targets. Normally the facilities of the machine's usual C compiler are
2846 used, but this can't be done directly in cross-compilation. You must
2847 make your own arrangements to provide suitable library functions for
2848 cross-compilation. The embedded targets @samp{m68k-*-aout} and
2849 @samp{m68k-*-coff} do provide software floating point support.
2852 Consider type @code{int} to be 16 bits wide, like @code{short int}.
2855 Do not use the bit-field instructions. The @samp{-m68000} option
2856 implies @w{@samp{-mnobitfield}}.
2859 Do use the bit-field instructions. The @samp{-m68020} option implies
2860 @samp{-mbitfield}. This is the default if you use a configuration
2861 designed for a 68020.
2864 Use a different function-calling convention, in which functions
2865 that take a fixed number of arguments return with the @code{rtd}
2866 instruction, which pops their arguments while returning. This
2867 saves one instruction in the caller since there is no need to pop
2868 the arguments there.
2870 This calling convention is incompatible with the one normally
2871 used on Unix, so you cannot use it if you need to call libraries
2872 compiled with the Unix compiler.
2874 Also, you must provide function prototypes for all functions that
2875 take variable numbers of arguments (including @code{printf});
2876 otherwise incorrect code will be generated for calls to those
2879 In addition, seriously incorrect code will result if you call a
2880 function with too many arguments. (Normally, extra arguments are
2881 harmlessly ignored.)
2883 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
2884 68040, and 68060 processors, but not by the 68000 or 5200.
2887 @itemx -mno-align-int
2888 Control whether GNU CC aligns @code{int}, @code{long}, @code{long long},
2889 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
2890 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
2891 Aligning variables on 32-bit boundaries produces code that runs somewhat
2892 faster on processors with 32-bit busses at the expense of more memory.
2894 @strong{Warning:} if you use the @samp{-malign-int} switch, GNU CC will
2895 align structures containing the above types differently than
2896 most published application binary interface specifications for the m68k.
2901 @subsection VAX Options
2904 These @samp{-m} options are defined for the Vax:
2908 Do not output certain jump instructions (@code{aobleq} and so on)
2909 that the Unix assembler for the Vax cannot handle across long
2913 Do output those jump instructions, on the assumption that you
2914 will assemble with the GNU assembler.
2917 Output code for g-format floating point numbers instead of d-format.
2921 @subsection SPARC Options
2922 @cindex SPARC options
2924 These @samp{-m} switches are supported on the SPARC:
2929 Specify @samp{-mapp-regs} to generate output using the global registers
2930 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
2933 To be fully SVR4 ABI compliant at the cost of some performance loss,
2934 specify @samp{-mno-app-regs}. You should compile libraries and system
2935 software with this option.
2939 Generate output containing floating point instructions. This is the
2944 Generate output containing library calls for floating point.
2945 @strong{Warning:} the requisite libraries are not available for all SPARC
2946 targets. Normally the facilities of the machine's usual C compiler are
2947 used, but this cannot be done directly in cross-compilation. You must make
2948 your own arrangements to provide suitable library functions for
2949 cross-compilation. The embedded targets @samp{sparc-*-aout} and
2950 @samp{sparclite-*-*} do provide software floating point support.
2952 @samp{-msoft-float} changes the calling convention in the output file;
2953 therefore, it is only useful if you compile @emph{all} of a program with
2954 this option. In particular, you need to compile @file{libgcc.a}, the
2955 library that comes with GNU CC, with @samp{-msoft-float} in order for
2958 @item -mhard-quad-float
2959 Generate output containing quad-word (long double) floating point
2962 @item -msoft-quad-float
2963 Generate output containing library calls for quad-word (long double)
2964 floating point instructions. The functions called are those specified
2965 in the SPARC ABI. This is the default.
2967 As of this writing, there are no sparc implementations that have hardware
2968 support for the quad-word floating point instructions. They all invoke
2969 a trap handler for one of these instructions, and then the trap handler
2970 emulates the effect of the instruction. Because of the trap handler overhead,
2971 this is much slower than calling the ABI library routines. Thus the
2972 @samp{-msoft-quad-float} option is the default.
2976 With @samp{-mepilogue} (the default), the compiler always emits code for
2977 function exit at the end of each function. Any function exit in
2978 the middle of the function (such as a return statement in C) will
2979 generate a jump to the exit code at the end of the function.
2981 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
2982 at every function exit.
2986 With @samp{-mflat}, the compiler does not generate save/restore instructions
2987 and will use a "flat" or single register window calling convention.
2988 This model uses %i7 as the frame pointer and is compatible with the normal
2989 register window model. Code from either may be intermixed.
2990 The local registers and the input registers (0-5) are still treated as
2991 "call saved" registers and will be saved on the stack as necessary.
2993 With @samp{-mno-flat} (the default), the compiler emits save/restore
2994 instructions (except for leaf functions) and is the normal mode of operation.
2996 @item -mno-unaligned-doubles
2997 @itemx -munaligned-doubles
2998 Assume that doubles have 8 byte alignment. This is the default.
3000 With @samp{-munaligned-doubles}, GNU CC assumes that doubles have 8 byte
3001 alignment only if they are contained in another type, or if they have an
3002 absolute address. Otherwise, it assumes they have 4 byte alignment.
3003 Specifying this option avoids some rare compatibility problems with code
3004 generated by other compilers. It is not the default because it results
3005 in a performance loss, especially for floating point code.
3009 These two options select variations on the SPARC architecture.
3011 By default (unless specifically configured for the Fujitsu SPARClite),
3012 GCC generates code for the v7 variant of the SPARC architecture.
3014 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
3015 code is that the compiler emits the integer multiply and integer
3016 divide instructions which exist in SPARC v8 but not in SPARC v7.
3018 @samp{-msparclite} will give you SPARClite code. This adds the integer
3019 multiply, integer divide step and scan (@code{ffs}) instructions which
3020 exist in SPARClite but not in SPARC v7.
3022 These options are deprecated and will be deleted in GNU CC 2.9.
3023 They have been replaced with @samp{-mcpu=xxx}.
3027 These two options select the processor for which the code is optimised.
3029 With @samp{-mcypress} (the default), the compiler optimizes code for the
3030 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
3031 This is also appropriate for the older SparcStation 1, 2, IPX etc.
3033 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
3034 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
3035 of the full SPARC v8 instruction set.
3037 These options are deprecated and will be deleted in GNU CC 2.9.
3038 They have been replaced with @samp{-mcpu=xxx}.
3040 @item -mcpu=@var{cpu_type}
3041 Set architecture type and instruction scheduling parameters for machine
3042 type @var{cpu_type}. Supported values for @var{cpu_type} are
3043 @samp{common}, @samp{cypress}, @samp{v8}, @samp{supersparc},
3044 @samp{sparclite}, @samp{f930}, @samp{f934},
3045 @samp{sparclet}, @samp{90c701}, @samp{v8plus}, @samp{v9},
3046 and @samp{ultrasparc}. Specifying @samp{v9} is only supported on true
3049 @item -mtune=@var{cpu_type}
3050 Set the instruction scheduling parameters for machine type
3051 @var{cpu_type}, but do not set the architecture type as the option
3052 @samp{-mcpu=}@var{cpu_type} would. The same values for
3053 @samp{-mcpu=}@var{cpu_type} are used for @samp{-tune=}@var{cpu_type}.
3057 These @samp{-m} switches are supported in addition to the above
3058 on the SPARCLET processor.
3061 @item -mlittle-endian
3062 Generate code for a processor running in little-endian mode.
3065 Treat register @code{%g0} as a normal register.
3066 GCC will continue to clobber it as necessary but will not assume
3067 it always reads as 0.
3069 @item -mbroken-saverestore
3070 Generate code that does not use non-trivial forms of the @code{save} and
3071 @code{restore} instructions. Early versions of the SPARCLET processor do
3072 not correctly handle @code{save} and @code{restore} instructions used with
3073 arguments. They correctly handle them used without arguments. A @code{save}
3074 instruction used without arguments increments the current window pointer
3075 but does not allocate a new stack frame. It is assumed that the window
3076 overflow trap handler will properly handle this case as will interrupt
3080 These @samp{-m} switches are supported in addition to the above
3081 on SPARC V9 processors in 64 bit environments.
3084 @item -mlittle-endian
3085 Generate code for a processor running in little-endian mode.
3088 Generate code for the Medium/Low code model: assume a 32 bit address space.
3089 Programs are statically linked, PIC is not supported. Pointers are still
3092 It is very likely that a future version of GCC will rename this option.
3095 Generate code for the Medium/Anywhere code model: assume a 32 bit text
3096 and a 32 bit data segment, both starting anywhere (determined at link time).
3097 Programs are statically linked, PIC is not supported. Pointers are still
3100 It is very likely that a future version of GCC will rename this option.
3103 Generate code for the Full/Anywhere code model: assume a full 64 bit
3104 address space. PIC is not supported.
3106 It is very likely that a future version of GCC will rename this option.
3109 Types long and int are 64 bits.
3112 Types long and int are 32 bits.
3116 Type long is 64 bits, and type int is 32 bits.
3119 @itemx -mno-stack-bias
3120 With @samp{-mstack-bias}, GNU CC assumes that the stack pointer, and
3121 frame pointer if present, are offset by -2047 which must be added back
3122 when making stack frame references.
3123 Otherwise, assume no such offset is present.
3126 @node Convex Options
3127 @subsection Convex Options
3128 @cindex Convex options
3130 These @samp{-m} options are defined for Convex:
3134 Generate output for C1. The code will run on any Convex machine.
3135 The preprocessor symbol @code{__convex__c1__} is defined.
3138 Generate output for C2. Uses instructions not available on C1.
3139 Scheduling and other optimizations are chosen for max performance on C2.
3140 The preprocessor symbol @code{__convex_c2__} is defined.
3143 Generate output for C32xx. Uses instructions not available on C1.
3144 Scheduling and other optimizations are chosen for max performance on C32.
3145 The preprocessor symbol @code{__convex_c32__} is defined.
3148 Generate output for C34xx. Uses instructions not available on C1.
3149 Scheduling and other optimizations are chosen for max performance on C34.
3150 The preprocessor symbol @code{__convex_c34__} is defined.
3153 Generate output for C38xx. Uses instructions not available on C1.
3154 Scheduling and other optimizations are chosen for max performance on C38.
3155 The preprocessor symbol @code{__convex_c38__} is defined.
3158 Generate code which puts an argument count in the word preceding each
3159 argument list. This is compatible with regular CC, and a few programs
3160 may need the argument count word. GDB and other source-level debuggers
3161 do not need it; this info is in the symbol table.
3164 Omit the argument count word. This is the default.
3166 @item -mvolatile-cache
3167 Allow volatile references to be cached. This is the default.
3169 @item -mvolatile-nocache
3170 Volatile references bypass the data cache, going all the way to memory.
3171 This is only needed for multi-processor code that does not use standard
3172 synchronization instructions. Making non-volatile references to volatile
3173 locations will not necessarily work.
3176 Type long is 32 bits, the same as type int. This is the default.
3179 Type long is 64 bits, the same as type long long. This option is useless,
3180 because no library support exists for it.
3183 @node AMD29K Options
3184 @subsection AMD29K Options
3185 @cindex AMD29K options
3187 These @samp{-m} options are defined for the AMD Am29000:
3192 @cindex DW bit (29k)
3193 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
3194 halfword operations are directly supported by the hardware. This is the
3199 Generate code that assumes the @code{DW} bit is not set.
3203 @cindex byte writes (29k)
3204 Generate code that assumes the system supports byte and halfword write
3205 operations. This is the default.
3209 Generate code that assumes the systems does not support byte and
3210 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
3214 @cindex memory model (29k)
3215 Use a small memory model that assumes that all function addresses are
3216 either within a single 256 KB segment or at an absolute address of less
3217 than 256k. This allows the @code{call} instruction to be used instead
3218 of a @code{const}, @code{consth}, @code{calli} sequence.
3222 Use the normal memory model: Generate @code{call} instructions only when
3223 calling functions in the same file and @code{calli} instructions
3224 otherwise. This works if each file occupies less than 256 KB but allows
3225 the entire executable to be larger than 256 KB. This is the default.
3228 Always use @code{calli} instructions. Specify this option if you expect
3229 a single file to compile into more than 256 KB of code.
3233 @cindex processor selection (29k)
3234 Generate code for the Am29050.
3238 Generate code for the Am29000. This is the default.
3240 @item -mkernel-registers
3241 @kindex -mkernel-registers
3242 @cindex kernel and user registers (29k)
3243 Generate references to registers @code{gr64-gr95} instead of to
3244 registers @code{gr96-gr127}. This option can be used when compiling
3245 kernel code that wants a set of global registers disjoint from that used
3248 Note that when this option is used, register names in @samp{-f} flags
3249 must use the normal, user-mode, names.
3251 @item -muser-registers
3252 @kindex -muser-registers
3253 Use the normal set of global registers, @code{gr96-gr127}. This is the
3257 @itemx -mno-stack-check
3258 @kindex -mstack-check
3259 @cindex stack checks (29k)
3260 Insert (or do not insert) a call to @code{__msp_check} after each stack
3261 adjustment. This is often used for kernel code.
3264 @itemx -mno-storem-bug
3265 @kindex -mstorem-bug
3266 @cindex storem bug (29k)
3267 @samp{-mstorem-bug} handles 29k processors which cannot handle the
3268 separation of a mtsrim insn and a storem instruction (most 29000 chips
3269 to date, but not the 29050).
3271 @item -mno-reuse-arg-regs
3272 @itemx -mreuse-arg-regs
3273 @kindex -mreuse-arg-regs
3274 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
3275 registers for copying out arguments. This helps detect calling a function
3276 with fewer arguments than it was declared with.
3278 @item -mno-impure-text
3279 @itemx -mimpure-text
3280 @kindex -mimpure-text
3281 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
3282 not pass @samp{-assert pure-text} to the linker when linking a shared object.
3285 @kindex -msoft-float
3286 Generate output containing library calls for floating point.
3287 @strong{Warning:} the requisite libraries are not part of GNU CC.
3288 Normally the facilities of the machine's usual C compiler are used, but
3289 this can't be done directly in cross-compilation. You must make your
3290 own arrangements to provide suitable library functions for
3295 @subsection ARM Options
3298 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
3303 @kindex -mapcs-frame
3304 Generate a stack frame that is compliant with the ARM Procedure Call
3305 Standard for all functions, even if this is not strictly necessary for
3306 correct execution of the code.
3310 Generate code for a processor running with a 26-bit program counter,
3311 and conforming to the function calling standards for the APCS 26-bit
3312 option. This option replaces the @samp{-m2} and @samp{-m3} options
3313 of previous releases of the compiler.
3317 Generate code for a processor running with a 32-bit program counter,
3318 and conforming to the function calling standards for the APCS 32-bit
3319 option. This option replaces the @samp{-m6} option of previous releases
3323 Generate output containing floating point instructions. This is the
3327 Generate output containing library calls for floating point.
3328 @strong{Warning:} the requisite libraries are not available for all ARM
3329 targets. Normally the facilities of the machine's usual C compiler are
3330 used, but this cannot be done directly in cross-compilation. You must make
3331 your own arrangements to provide suitable library functions for
3334 @samp{-msoft-float} changes the calling convention in the output file;
3335 therefore, it is only useful if you compile @emph{all} of a program with
3336 this option. In particular, you need to compile @file{libgcc.a}, the
3337 library that comes with GNU CC, with @samp{-msoft-float} in order for
3340 @item -mlittle-endian
3341 Generate code for a processor running in little-endian mode. This is
3342 the default for all standard configurations.
3345 Generate code for a processor running in big-endian mode; the default is
3346 to compile code for a little-endian processor.
3348 @item -mwords-little-endian
3349 This option only applies when generating code for big-endian processors.
3350 Generate code for a little-endian word order but a big-endian byte
3351 order. That is, a byte order of the form @samp{32107654}. Note: this
3352 option should only be used if you require compatibility with code for
3353 big-endian ARM processors generated by versions of the compiler prior to
3356 @item -mshort-load-bytes
3357 @kindex -mshort-load-bytes
3358 Do not try to load half-words (eg @samp{short}s) by loading a word from
3359 an unaligned address. For some targets the MMU is configured to trap
3360 unaligned loads; use this option to generate code that is safe in these
3363 @item -mno-short-load-bytes
3364 @kindex -mno-short-load-bytes
3365 Use unaligned word loads to load half-words (eg @samp{short}s). This
3366 option produces more efficient code, but the MMU is sometimes configured
3367 to trap these instructions.
3371 This option only applies to RISC iX. Emulate the native BSD-mode
3372 compiler. This is the default if @samp{-ansi} is not specified.
3376 This option only applies to RISC iX. Emulate the native X/Open-mode
3379 @item -mno-symrename
3380 @kindex -mno-symrename
3381 This option only applies to RISC iX. Do not run the assembler
3382 post-processor, @samp{symrename}, after code has been assembled.
3383 Normally it is necessary to modify some of the standard symbols in
3384 preparation for linking with the RISC iX C library; this option
3385 suppresses this pass. The post-processor is never run when the
3386 compiler is built for cross-compilation.
3389 @node M32R/D Options
3390 @subsection M32R/D Options
3391 @cindex M32R/D options
3393 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
3396 @item -mcode-model=small
3397 Assume all objects live in the lower 16MB of memory (so that their addresses
3398 can be loaded with the @code{ld24} instruction), and assume all subroutines
3399 are reachable with the @code{bl} instruction.
3400 This is the default.
3402 The addressability of a particular object can be set with the
3403 @code{model} attribute.
3405 @item -mcode-model=medium
3406 Assume objects may be anywhere in the 32 bit address space (the compiler
3407 will generate @code{seth/add3} instructions to load their addresses), and
3408 assume all subroutines are reachable with the @code{bl} instruction.
3410 @item -mcode-model=large
3411 Assume objects may be anywhere in the 32 bit address space (the compiler
3412 will generate @code{seth/add3} instructions to load their addresses), and
3413 assume subroutines may not be reachable with the @code{bl} instruction
3414 (the compiler will generate the much slower @code{seth/add3/jl}
3415 instruction sequence).
3418 Disable use of the small data area. Variables will be put into
3419 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
3420 @code{section} attribute has been specified).
3421 This is the default.
3423 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
3424 Objects may be explicitly put in the small data area with the
3425 @code{section} attribute using one of these sections.
3428 Put small global and static data in the small data area, but do not
3429 generate special code to reference them.
3432 Put small global and static data in the small data area, and generate
3433 special instructions to reference them.
3436 @cindex smaller data references
3437 Put global and static objects less than or equal to @var{num} bytes
3438 into the small data or bss sections instead of the normal data or bss
3439 sections. The default value of @var{num} is 8.
3440 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
3441 for this option to have any effect.
3443 All modules should be compiled with the same @samp{-G @var{num}} value.
3444 Compiling with different values of @var{num} may or may not work; if it
3445 doesn't the linker will give an error message - incorrect code will not be
3451 @subsection M88K Options
3452 @cindex M88k options
3454 These @samp{-m} options are defined for Motorola 88k architectures:
3459 Generate code that works well on both the m88100 and the
3464 Generate code that works best for the m88100, but that also
3469 Generate code that works best for the m88110, and may not run
3474 Obsolete option to be removed from the next revision.
3477 @item -midentify-revision
3478 @kindex -midentify-revision
3480 @cindex identifying source, compiler (88k)
3481 Include an @code{ident} directive in the assembler output recording the
3482 source file name, compiler name and version, timestamp, and compilation
3485 @item -mno-underscores
3486 @kindex -mno-underscores
3487 @cindex underscores, avoiding (88k)
3488 In assembler output, emit symbol names without adding an underscore
3489 character at the beginning of each name. The default is to use an
3490 underscore as prefix on each name.
3492 @item -mocs-debug-info
3493 @itemx -mno-ocs-debug-info
3494 @kindex -mocs-debug-info
3495 @kindex -mno-ocs-debug-info
3497 @cindex debugging, 88k OCS
3498 Include (or omit) additional debugging information (about registers used
3499 in each stack frame) as specified in the 88open Object Compatibility
3500 Standard, ``OCS''. This extra information allows debugging of code that
3501 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
3502 Delta 88 SVr3.2 is to include this information; other 88k configurations
3503 omit this information by default.
3505 @item -mocs-frame-position
3506 @kindex -mocs-frame-position
3507 @cindex register positions in frame (88k)
3508 When emitting COFF debugging information for automatic variables and
3509 parameters stored on the stack, use the offset from the canonical frame
3510 address, which is the stack pointer (register 31) on entry to the
3511 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
3512 @samp{-mocs-frame-position}; other 88k configurations have the default
3513 @samp{-mno-ocs-frame-position}.
3515 @item -mno-ocs-frame-position
3516 @kindex -mno-ocs-frame-position
3517 @cindex register positions in frame (88k)
3518 When emitting COFF debugging information for automatic variables and
3519 parameters stored on the stack, use the offset from the frame pointer
3520 register (register 30). When this option is in effect, the frame
3521 pointer is not eliminated when debugging information is selected by the
3524 @item -moptimize-arg-area
3525 @itemx -mno-optimize-arg-area
3526 @kindex -moptimize-arg-area
3527 @kindex -mno-optimize-arg-area
3528 @cindex arguments in frame (88k)
3529 Control how function arguments are stored in stack frames.
3530 @samp{-moptimize-arg-area} saves space by optimizing them, but this
3531 conflicts with the 88open specifications. The opposite alternative,
3532 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
3533 GNU CC does not optimize the argument area.
3535 @item -mshort-data-@var{num}
3536 @kindex -mshort-data-@var{num}
3537 @cindex smaller data references (88k)
3538 @cindex r0-relative references (88k)
3539 Generate smaller data references by making them relative to @code{r0},
3540 which allows loading a value using a single instruction (rather than the
3541 usual two). You control which data references are affected by
3542 specifying @var{num} with this option. For example, if you specify
3543 @samp{-mshort-data-512}, then the data references affected are those
3544 involving displacements of less than 512 bytes.
3545 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
3548 @item -mserialize-volatile
3549 @kindex -mserialize-volatile
3550 @itemx -mno-serialize-volatile
3551 @kindex -mno-serialize-volatile
3552 @cindex sequential consistency on 88k
3553 Do, or don't, generate code to guarantee sequential consistency
3554 of volatile memory references. By default, consistency is
3557 The order of memory references made by the MC88110 processor does
3558 not always match the order of the instructions requesting those
3559 references. In particular, a load instruction may execute before
3560 a preceding store instruction. Such reordering violates
3561 sequential consistency of volatile memory references, when there
3562 are multiple processors. When consistency must be guaranteed,
3563 GNU C generates special instructions, as needed, to force
3564 execution in the proper order.
3566 The MC88100 processor does not reorder memory references and so
3567 always provides sequential consistency. However, by default, GNU
3568 C generates the special instructions to guarantee consistency
3569 even when you use @samp{-m88100}, so that the code may be run on an
3570 MC88110 processor. If you intend to run your code only on the
3571 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
3573 The extra code generated to guarantee consistency may affect the
3574 performance of your application. If you know that you can safely
3575 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
3581 @cindex assembler syntax, 88k
3583 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
3584 related to System V release 4 (SVr4). This controls the following:
3588 Which variant of the assembler syntax to emit.
3590 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
3591 that is used on System V release 4.
3593 @samp{-msvr4} makes GNU CC issue additional declaration directives used in
3597 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
3598 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
3599 other m88k configurations.
3601 @item -mversion-03.00
3602 @kindex -mversion-03.00
3603 This option is obsolete, and is ignored.
3604 @c ??? which asm syntax better for GAS? option there too?
3606 @item -mno-check-zero-division
3607 @itemx -mcheck-zero-division
3608 @kindex -mno-check-zero-division
3609 @kindex -mcheck-zero-division
3610 @cindex zero division on 88k
3611 Do, or don't, generate code to guarantee that integer division by
3612 zero will be detected. By default, detection is guaranteed.
3614 Some models of the MC88100 processor fail to trap upon integer
3615 division by zero under certain conditions. By default, when
3616 compiling code that might be run on such a processor, GNU C
3617 generates code that explicitly checks for zero-valued divisors
3618 and traps with exception number 503 when one is detected. Use of
3619 mno-check-zero-division suppresses such checking for code
3620 generated to run on an MC88100 processor.
3622 GNU C assumes that the MC88110 processor correctly detects all
3623 instances of integer division by zero. When @samp{-m88110} is
3624 specified, both @samp{-mcheck-zero-division} and
3625 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
3626 zero-valued divisors are generated.
3628 @item -muse-div-instruction
3629 @kindex -muse-div-instruction
3630 @cindex divide instruction, 88k
3631 Use the div instruction for signed integer division on the
3632 MC88100 processor. By default, the div instruction is not used.
3634 On the MC88100 processor the signed integer division instruction
3635 div) traps to the operating system on a negative operand. The
3636 operating system transparently completes the operation, but at a
3637 large cost in execution time. By default, when compiling code
3638 that might be run on an MC88100 processor, GNU C emulates signed
3639 integer division using the unsigned integer division instruction
3640 divu), thereby avoiding the large penalty of a trap to the
3641 operating system. Such emulation has its own, smaller, execution
3642 cost in both time and space. To the extent that your code's
3643 important signed integer division operations are performed on two
3644 nonnegative operands, it may be desirable to use the div
3645 instruction directly.
3647 On the MC88110 processor the div instruction (also known as the
3648 divs instruction) processes negative operands without trapping to
3649 the operating system. When @samp{-m88110} is specified,
3650 @samp{-muse-div-instruction} is ignored, and the div instruction is used
3651 for signed integer division.
3653 Note that the result of dividing INT_MIN by -1 is undefined. In
3654 particular, the behavior of such a division with and without
3655 @samp{-muse-div-instruction} may differ.
3657 @item -mtrap-large-shift
3658 @itemx -mhandle-large-shift
3659 @kindex -mtrap-large-shift
3660 @kindex -mhandle-large-shift
3661 @cindex bit shift overflow (88k)
3662 @cindex large bit shifts (88k)
3663 Include code to detect bit-shifts of more than 31 bits; respectively,
3664 trap such shifts or emit code to handle them properly. By default GNU CC
3665 makes no special provision for large bit shifts.
3667 @item -mwarn-passed-structs
3668 @kindex -mwarn-passed-structs
3669 @cindex structure passing (88k)
3670 Warn when a function passes a struct as an argument or result.
3671 Structure-passing conventions have changed during the evolution of the C
3672 language, and are often the source of portability problems. By default,
3673 GNU CC issues no such warning.
3676 @node RS/6000 and PowerPC Options
3677 @subsection IBM RS/6000 and PowerPC Options
3678 @cindex RS/6000 and PowerPC Options
3679 @cindex IBM RS/6000 and PowerPC Options
3681 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
3689 @itemx -mpowerpc-gpopt
3690 @itemx -mno-powerpc-gpopt
3691 @itemx -mpowerpc-gfxopt
3692 @itemx -mno-powerpc-gfxopt
3696 @kindex -mpowerpc-gpopt
3697 @kindex -mpowerpc-gfxopt
3698 GNU CC supports two related instruction set architectures for the
3699 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
3700 instructions supported by the @samp{rios} chip set used in the original
3701 RS/6000 systems and the @dfn{PowerPC} instruction set is the
3702 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
3703 the IBM 4xx microprocessors.
3705 Neither architecture is a subset of the other. However there is a
3706 large common subset of instructions supported by both. An MQ
3707 register is included in processors supporting the POWER architecture.
3709 You use these options to specify which instructions are available on the
3710 processor you are using. The default value of these options is
3711 determined when configuring GNU CC. Specifying the
3712 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
3713 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
3714 rather than the options listed above.
3716 The @samp{-mpower} option allows GNU CC to generate instructions that
3717 are found only in the POWER architecture and to use the MQ register.
3718 Specifying @samp{-mpower2} implies @samp{-power} and also allows GNU CC
3719 to generate instructions that are present in the POWER2 architecture but
3720 not the original POWER architecture.
3722 The @samp{-mpowerpc} option allows GNU CC to generate instructions that
3723 are found only in the 32-bit subset of the PowerPC architecture.
3724 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
3725 GNU CC to use the optional PowerPC architecture instructions in the
3726 General Purpose group, including floating-point square root. Specifying
3727 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GNU CC to
3728 use the optional PowerPC architecture instructions in the Graphics
3729 group, including floating-point select.
3731 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GNU CC
3732 will use only the instructions in the common subset of both
3733 architectures plus some special AIX common-mode calls, and will not use
3734 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
3735 permits GNU CC to use any instruction from either architecture and to
3736 allow use of the MQ register; specify this for the Motorola MPC601.
3738 @item -mnew-mnemonics
3739 @itemx -mold-mnemonics
3740 @kindex -mnew-mnemonics
3741 @kindex -mold-mnemonics
3742 Select which mnemonics to use in the generated assembler code.
3743 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
3744 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
3745 requests the assembler mnemonics defined for the POWER architecture.
3746 Instructions defined in only one architecture have only one mnemonic;
3747 GNU CC uses that mnemonic irrespective of which of these options is
3750 PowerPC assemblers support both the old and new mnemonics, as will later
3751 POWER assemblers. Current POWER assemblers only support the old
3752 mnemonics. Specify @samp{-mnew-mnemonics} if you have an assembler that
3753 supports them, otherwise specify @samp{-mold-mnemonics}.
3755 The default value of these options depends on how GNU CC was configured.
3756 Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the value of
3757 these option. Unless you are building a cross-compiler, you should
3758 normally not specify either @samp{-mnew-mnemonics} or
3759 @samp{-mold-mnemonics}, but should instead accept the default.
3761 @item -mcpu=@var{cpu_type}
3762 Set architecture type, register usage, choice of mnemonics, and
3763 instruction scheduling parameters for machine type @var{cpu_type}.
3764 Supported values for @var{cpu_type} are @samp{rs6000}, @samp{rios1},
3765 @samp{rios2}, @samp{rsc}, @samp{601}, @samp{602}, @samp{603},
3766 @samp{603e}, @samp{604}, @samp{604e}, @samp{620}, @samp{power},
3767 @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505}, @samp{801},
3768 @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
3769 @samp{-mcpu=power}, @samp{-mcpu=power2}, and @samp{-mcpu=powerpc}
3770 specify generic POWER, POWER2 and pure PowerPC (i.e., not MPC601)
3771 architecture machine types, with an appropriate, generic processor model
3772 assumed for scheduling purposes.@refill
3774 @c overfull hbox here --bob 22 jul96
3775 @c original text between ignore ... end ignore
3777 Specifying any of the @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
3778 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} options
3779 enables the @samp{-mpower} option and disables the @samp{-mpowerpc}
3780 option; @samp{-mcpu=601} enables both the @samp{-mpower} and
3781 @samp{-mpowerpc} options; all of @samp{-mcpu=602}, @samp{-mcpu=603},
3782 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=604e},
3783 @samp{-mcpu=620}, @samp{-mcpu=403}, @samp{-mcpu=505}, @samp{-mcpu=801},
3784 @samp{-mcpu=821}, @samp{-mcpu=823}, @samp{-mcpu=860} and
3785 @samp{-mcpu=powerpc} enable the @samp{-mpowerpc} option and disable the
3786 @samp{-mpower} option; @samp{-mcpu=common} disables both the
3787 @samp{-mpower} and @samp{-mpowerpc} options.@refill
3789 @c changed paragraph
3790 Specifying any of the following options:
3791 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
3792 @samp{-mcpu=power}, or @samp{-mcpu=power2}
3793 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
3794 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
3795 All of @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e},
3796 @samp{-mcpu=604}, @samp{-mcpu=620},
3797 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
3798 Exactly similarly, all of @samp{-mcpu=403},
3799 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
3800 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
3801 @samp{-mcpu=common} disables both the
3802 @samp{-mpower} and @samp{-mpowerpc} options.@refill
3803 @c end changes to prevent overfull hboxes
3805 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
3806 that code will operate on all members of the RS/6000 and PowerPC
3807 families. In that case, GNU CC will use only the instructions in the
3808 common subset of both architectures plus some special AIX common-mode
3809 calls, and will not use the MQ register. GNU CC assumes a generic
3810 processor model for scheduling purposes.
3812 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
3813 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
3814 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
3815 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
3816 @samp{620}, @samp{403}, or @samp{-mcpu=powerpc} also enables the
3817 @samp{new-mnemonics} option.@refill
3819 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
3820 enables the @samp{-msoft-float} option.
3822 @item -mtune=@var{cpu_type}
3823 Set the instruction scheduling parameters for machine type
3824 @var{cpu_type}, but do not set the architecture type, register usage,
3825 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
3826 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
3827 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
3828 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
3829 instruction scheduling parameters.
3832 @itemx -mno-fp-in-toc
3833 @itemx -mno-sum-in-toc
3834 @itemx -mminimal-toc
3835 Modify generation of the TOC (Table Of Contents), which is created for
3836 every executable file. The @samp{-mfull-toc} option is selected by
3837 default. In that case, GNU CC will allocate at least one TOC entry for
3838 each unique non-automatic variable reference in your program. GNU CC
3839 will also place floating-point constants in the TOC. However, only
3840 16,384 entries are available in the TOC.
3842 If you receive a linker error message that saying you have overflowed
3843 the available TOC space, you can reduce the amount of TOC space used
3844 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
3845 @samp{-mno-fp-in-toc} prevents GNU CC from putting floating-point
3846 constants in the TOC and @samp{-mno-sum-in-toc} forces GNU CC to
3847 generate code to calculate the sum of an address and a constant at
3848 run-time instead of putting that sum into the TOC. You may specify one
3849 or both of these options. Each causes GNU CC to produce very slightly
3850 slower and larger code at the expense of conserving TOC space.
3852 If you still run out of space in the TOC even when you specify both of
3853 these options, specify @samp{-mminimal-toc} instead. This option causes
3854 GNU CC to make only one TOC entry for every file. When you specify this
3855 option, GNU CC will produce code that is slower and larger but which
3856 uses extremely little TOC space. You may wish to use this option
3857 only on files that contain less frequently executed code. @refill
3861 On AIX, pass floating-point arguments to prototyped functions beyond the
3862 register save area (RSA) on the stack in addition to argument FPRs. The
3863 AIX calling convention was extended but not initially documented to
3864 handle an obscure K&R C case of calling a function that takes the
3865 address of its arguments with fewer arguments than declared. AIX XL
3866 compilers assume that floating point arguments which do not fit in the
3867 RSA are on the stack when they compile a subroutine without
3868 optimization. Because always storing floating-point arguments on the
3869 stack is inefficient and rarely needed, this option is not enabled by
3870 default and only is necessary when calling subroutines compiled by AIX
3871 XL compilers without optimization.
3874 Support @dfn{AIX Threads}. Link an application written to use
3875 @dfn{pthreads} with special libraries and startup code to enable the
3879 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
3880 application written to use message passing with special startup code to
3881 enable the application to run. The system must have PE installed in the
3882 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
3883 must be overridden with the @samp{-specs=} option to specify the
3884 appropriate directory location. The Parallel Environment does not
3885 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
3886 option are incompatible.
3890 Generate code that does not use (uses) the floating-point register set.
3891 Software floating point emulation is provided if you use the
3892 @samp{-msoft-float} option, and pass the option to GNU CC when linking.
3895 @itemx -mno-multiple
3896 Generate code that uses (does not use) the load multiple word
3897 instructions and the store multiple word instructions. These
3898 instructions are generated by default on POWER systems, and not
3899 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
3900 endian PowerPC systems, since those instructions do not work when the
3901 processor is in little endian mode.
3905 Generate code that uses (does not use) the load string instructions and the
3906 store string word instructions to save multiple registers and do small block
3907 moves. These instructions are generated by default on POWER systems, and not
3908 generated on PowerPC systems. Do not use @samp{-mstring} on little endian
3909 PowerPC systems, since those instructions do not work when the processor is in
3914 Generate code that uses (does not use) the load or store instructions
3915 that update the base register to the address of the calculated memory
3916 location. These instructions are generated by default. If you use
3917 @samp{-mno-update}, there is a small window between the time that the
3918 stack pointer is updated and the address of the previous frame is
3919 stored, which means code that walks the stack frame across interrupts or
3920 signals may get corrupted data.
3923 @itemx -mno-fused-madd
3924 Generate code that uses (does not use) the floating point multiply and
3925 accumulate instructions. These instructions are generated by default if
3926 hardware floating is used.
3928 @item -mno-bit-align
3930 On System V.4 and embedded PowerPC systems do not (do) force structures
3931 and unions that contain bit fields to be aligned to the base type of the
3934 For example, by default a structure containing nothing but 8
3935 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
3936 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
3937 the structure would be aligned to a 1 byte boundary and be one byte in
3940 @item -mno-strict-align
3941 @itemx -mstrict-align
3942 On System V.4 and embedded PowerPC systems do not (do) assume that
3943 unaligned memory references will be handled by the system.
3946 @itemx -mno-relocatable
3947 On embedded PowerPC systems generate code that allows (does not allow)
3948 the program to be relocated to a different address at runtime. If you
3949 use @samp{-mrelocatable} on any module, all objects linked together must
3950 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
3952 @item -mrelocatable-lib
3953 @itemx -mno-relocatable-lib
3954 On embedded PowerPC systems generate code that allows (does not allow)
3955 the program to be relocated to a different address at runtime. Modules
3956 compiled with @samp{-mreloctable-lib} can be linked with either modules
3957 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
3958 with modules compiled with the @samp{-mrelocatable} options.
3962 On System V.4 and embedded PowerPC systems do not (do) assume that
3963 register 2 contains a pointer to a global area pointing to the addresses
3964 used in the program.
3966 @item -mno-traceback
3968 On embedded PowerPC systems do not (do) generate a traceback tag before
3969 the start of the function. This tag can be used by the debugger to
3970 identify where the start of a function is.
3973 @itemx -mlittle-endian
3974 On System V.4 and embedded PowerPC systems compile code for the
3975 processor in little endian mode. The @samp{-mlittle-endian} option is
3976 the same as @samp{-mlittle}.
3980 On System V.4 and embedded PowerPC systems compile code for the
3981 processor in big endian mode. The @samp{-mbig-endian} option is
3982 the same as @samp{-mbig}.
3985 On System V.4 and embedded PowerPC systems compile code using calling
3986 conventions that adheres to the March 1995 draft of the System V
3987 Application Binary Interface, PowerPC processor supplement. This is the
3988 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
3990 @item -mcall-sysv-eabi
3991 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
3993 @item -mcall-sysv-noeabi
3994 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
3997 On System V.4 and embedded PowerPC systems compile code using calling
3998 conventions that are similar to those used on AIX. This is the
3999 default if you configured GCC using @samp{powerpc-*-eabiaix}.
4001 @item -mcall-solaris
4002 On System V.4 and embedded PowerPC systems compile code for the Solaris
4006 On System V.4 and embedded PowerPC systems compile code for the
4007 Linux-based GNU system.
4010 @item -mno-prototype
4011 On System V.4 and embedded PowerPC systems assume that all calls to
4012 variable argument functions are properly prototyped. Otherwise, the
4013 compiler must insert an instruction before every non prototyped call to
4014 set or clear bit 6 of the condition code register (@var{CR}) to
4015 indicate whether floating point values were passed in the floating point
4016 registers in case the function takes a variable arguments. With
4017 @samp{-mprototype}, only calls to prototyped variable argument functions
4018 will set or clear the bit.
4021 On embedded PowerPC systems, assume that the startup module is called
4022 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
4023 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
4027 On embedded PowerPC systems, assume that the startup module is called
4028 @file{mvme-crt0.o} and the standard C libraries are @file{libmvme.a} and
4032 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
4033 header to indicate that @samp{eabi} extended relocations are used.
4037 On System V.4 and embedded PowerPC systems do (do not) adhere to the
4038 Embedded Applications Binary Interface (eabi) which is a set of
4039 modifications to the System V.4 specifications. Selecting @code{-meabi}
4040 means that the stack is aligned to an 8 byte boundary, a function
4041 @code{__eabi} is called to from @code{main} to set up the eabi
4042 environment, and the @samp{-msdata} option can use both @code{r2} and
4043 @code{r13} to point to two separate small data areas. Selecting
4044 @code{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
4045 do not call an initialization function from @code{main}, and the
4046 @samp{-msdata} option will only use @code{r13} to point to a single
4047 small data area. The @samp{-meabi} option is on by default if you
4048 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
4051 On System V.4 and embedded PowerPC systems, put small initialized
4052 @code{const} global and static data in the @samp{.sdata2} section, which
4053 is pointed to by register @code{r2}. Put small initialized
4054 non-@code{const} global and static data in the @samp{.sdata} section,
4055 which is pointed to by register @code{r13}. Put small uninitialized
4056 global and static data in the @samp{.sbss} section, which is adjacent to
4057 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
4058 incompatible with the @samp{-mrelocatable} option. The
4059 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
4062 On System V.4 and embedded PowerPC systems, put small global and static
4063 data in the @samp{.sdata} section, which is pointed to by register
4064 @code{r13}. Put small uninitialized global and static data in the
4065 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
4066 The @samp{-msdata=sysv} option is incompatible with the
4067 @samp{-mrelocatable} option.
4069 @item -msdata=default
4071 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
4072 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
4073 same as @samp{-msdata=sysv}.
4076 On System V.4 and embedded PowerPC systems, put small global and static
4077 data in the @samp{.sdata} section. Put small uninitialized global and
4078 static data in the @samp{.sbss} section. Do not use register @code{r13}
4079 to address small data however. This is the default behavior unless
4080 other @samp{-msdata} options are used.
4084 On embedded PowerPC systems, put all initialized global and static data
4085 in the @samp{.data} section, and all uninitialized data in the
4086 @samp{.bss} section.
4089 @cindex smaller data references (PowerPC)
4090 @cindex .sdata/.sdata2 references (PowerPC)
4091 On embbeded PowerPC systems, put global and static items less than or
4092 equal to @var{num} bytes into the small data or bss sections instead of
4093 the normal data or bss section. By default, @var{num} is 8. The
4094 @samp{-G @var{num}} switch is also passed to the linker.
4095 All modules should be compiled with the same @samp{-G @var{num}} value.
4098 @itemx -mno-regnames
4099 On System V.4 and embedded PowerPC systems do (do not) emit register
4100 names in the assembly language output using symbolic forms.
4103 @subsection IBM RT Options
4105 @cindex IBM RT options
4107 These @samp{-m} options are defined for the IBM RT PC:
4111 Use an in-line code sequence for integer multiplies. This is the
4114 @item -mcall-lib-mul
4115 Call @code{lmul$$} for integer multiples.
4117 @item -mfull-fp-blocks
4118 Generate full-size floating point data blocks, including the minimum
4119 amount of scratch space recommended by IBM. This is the default.
4121 @item -mminimum-fp-blocks
4122 Do not include extra scratch space in floating point data blocks. This
4123 results in smaller code, but slower execution, since scratch space must
4124 be allocated dynamically.
4126 @cindex @file{varargs.h} and RT PC
4127 @cindex @file{stdarg.h} and RT PC
4128 @item -mfp-arg-in-fpregs
4129 Use a calling sequence incompatible with the IBM calling convention in
4130 which floating point arguments are passed in floating point registers.
4131 Note that @code{varargs.h} and @code{stdargs.h} will not work with
4132 floating point operands if this option is specified.
4134 @item -mfp-arg-in-gregs
4135 Use the normal calling convention for floating point arguments. This is
4138 @item -mhc-struct-return
4139 Return structures of more than one word in memory, rather than in a
4140 register. This provides compatibility with the MetaWare HighC (hc)
4141 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
4142 with the Portable C Compiler (pcc).
4144 @item -mnohc-struct-return
4145 Return some structures of more than one word in registers, when
4146 convenient. This is the default. For compatibility with the
4147 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
4148 option @samp{-mhc-struct-return}.
4152 @subsection MIPS Options
4153 @cindex MIPS options
4155 These @samp{-m} options are defined for the MIPS family of computers:
4158 @item -mcpu=@var{cpu type}
4159 Assume the defaults for the machine type @var{cpu type} when scheduling
4160 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
4161 @samp{r4000}, @samp{r4400}, @samp{r4600}, and @samp{r6000}. While picking a
4162 specific @var{cpu type} will schedule things appropriately for that
4163 particular chip, the compiler will not generate any code that does not
4164 meet level 1 of the MIPS ISA (instruction set architecture) without
4165 the @samp{-mips2} or @samp{-mips3} switches being used.
4168 Issue instructions from level 1 of the MIPS ISA. This is the default.
4169 @samp{r3000} is the default @var{cpu type} at this ISA level.
4172 Issue instructions from level 2 of the MIPS ISA (branch likely, square
4173 root instructions). @samp{r6000} is the default @var{cpu type} at this
4177 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
4178 @samp{r4000} is the default @var{cpu type} at this ISA level.
4179 This option does not change the sizes of any of the C data types.
4182 Assume that 32 32-bit floating point registers are available. This is
4186 Assume that 32 64-bit floating point registers are available. This is
4187 the default when the @samp{-mips3} option is used.
4190 Assume that 32 32-bit general purpose registers are available. This is
4194 Assume that 32 64-bit general purpose registers are available. This is
4195 the default when the @samp{-mips3} option is used.
4198 Types long, int, and pointer are 64 bits. This works only if @samp{-mips3}
4202 Types long and pointer are 64 bits, and type int is 32 bits.
4203 This works only if @samp{-mips3} is also specified.
4206 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
4207 add normal debug information. This is the default for all
4208 platforms except for the OSF/1 reference platform, using the OSF/rose
4209 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
4210 switches are used, the @file{mips-tfile} program will encapsulate the
4211 stabs within MIPS ECOFF.
4214 Generate code for the GNU assembler. This is the default on the OSF/1
4215 reference platform, using the OSF/rose object format. Also, this is
4216 the default if the configure option @samp{--with-gnu-as} is used.
4218 @item -msplit-addresses
4219 @itemx -mno-split-addresses
4220 Generate code to load the high and low parts of address constants separately.
4221 This allows @code{gcc} to optimize away redundant loads of the high order
4222 bits of addresses. This optimization requires GNU as and GNU ld.
4223 This optimization is enabled by default for some embedded targets where
4224 GNU as and GNU ld are standard.
4228 The @samp{-mrnames} switch says to output code using the MIPS software
4229 names for the registers, instead of the hardware names (ie, @var{a0}
4230 instead of @var{$4}). The only known assembler that supports this option
4231 is the Algorithmics assembler.
4235 The @samp{-mgpopt} switch says to write all of the data declarations
4236 before the instructions in the text section, this allows the MIPS
4237 assembler to generate one word memory references instead of using two
4238 words for short global or static data items. This is on by default if
4239 optimization is selected.
4243 For each non-inline function processed, the @samp{-mstats} switch
4244 causes the compiler to emit one line to the standard error file to
4245 print statistics about the program (number of registers saved, stack
4250 The @samp{-mmemcpy} switch makes all block moves call the appropriate
4251 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
4252 generating inline code.
4255 @itemx -mno-mips-tfile
4256 The @samp{-mno-mips-tfile} switch causes the compiler not
4257 postprocess the object file with the @file{mips-tfile} program,
4258 after the MIPS assembler has generated it to add debug support. If
4259 @file{mips-tfile} is not run, then no local variables will be
4260 available to the debugger. In addition, @file{stage2} and
4261 @file{stage3} objects will have the temporary file names passed to the
4262 assembler embedded in the object file, which means the objects will
4263 not compare the same. The @samp{-mno-mips-tfile} switch should only
4264 be used when there are bugs in the @file{mips-tfile} program that
4265 prevents compilation.
4268 Generate output containing library calls for floating point.
4269 @strong{Warning:} the requisite libraries are not part of GNU CC.
4270 Normally the facilities of the machine's usual C compiler are used, but
4271 this can't be done directly in cross-compilation. You must make your
4272 own arrangements to provide suitable library functions for
4276 Generate output containing floating point instructions. This is the
4277 default if you use the unmodified sources.
4280 @itemx -mno-abicalls
4281 Emit (or do not emit) the pseudo operations @samp{.abicalls},
4282 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
4283 position independent code.
4286 @itemx -mno-long-calls
4287 Do all calls with the @samp{JALR} instruction, which requires
4288 loading up a function's address into a register before the call.
4289 You need to use this switch, if you call outside of the current
4290 512 megabyte segment to functions that are not through pointers.
4293 @itemx -mno-half-pic
4294 Put pointers to extern references into the data section and load them
4295 up, rather than put the references in the text section.
4297 @item -membedded-pic
4298 @itemx -mno-embedded-pic
4299 Generate PIC code suitable for some embedded systems. All calls are made
4300 using PC relative address, and all data is addressed using the $gp register.
4301 This requires GNU as and GNU ld which do most of the work.
4303 @item -membedded-data
4304 @itemx -mno-embedded-data
4305 Allocate variables to the read-only data section first if possible, then
4306 next in the small data section if possible, otherwise in data. This gives
4307 slightly slower code than the default, but reduces the amount of RAM required
4308 when executing, and thus may be preferred for some embedded systems.
4310 @item -msingle-float
4311 @itemx -mdouble-float
4312 The @samp{-msingle-float} switch tells gcc to assume that the floating
4313 point coprocessor only supports single precision operations, as on the
4314 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
4315 double precision operations. This is the default.
4319 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
4320 as on the @samp{r4650} chip.
4323 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
4327 Compile code for the processor in little endian mode.
4328 The requisite libraries are assumed to exist.
4331 Compile code for the processor in big endian mode.
4332 The requisite libraries are assumed to exist.
4335 @cindex smaller data references (MIPS)
4336 @cindex gp-relative references (MIPS)
4337 Put global and static items less than or equal to @var{num} bytes into
4338 the small data or bss sections instead of the normal data or bss
4339 section. This allows the assembler to emit one word memory reference
4340 instructions based on the global pointer (@var{gp} or @var{$28}),
4341 instead of the normal two words used. By default, @var{num} is 8 when
4342 the MIPS assembler is used, and 0 when the GNU assembler is used. The
4343 @samp{-G @var{num}} switch is also passed to the assembler and linker.
4344 All modules should be compiled with the same @samp{-G @var{num}}
4348 Tell the MIPS assembler to not run it's preprocessor over user
4349 assembler files (with a @samp{.s} suffix) when assembling them.
4353 These options are defined by the macro
4354 @code{TARGET_SWITCHES} in the machine description. The default for the
4355 options is also defined by that macro, which enables you to change the
4360 @subsection Intel 386 Options
4361 @cindex i386 Options
4362 @cindex Intel 386 Options
4364 These @samp{-m} options are defined for the i386 family of computers:
4369 Control whether or not code is optimized for a 486 instead of an
4370 386. Code generated for an 486 will run on a 386 and vice versa.
4374 Control whether or not the compiler uses IEEE floating point
4375 comparisons. These handle correctly the case where the result of a
4376 comparison is unordered.
4379 Generate output containing library calls for floating point.
4380 @strong{Warning:} the requisite libraries are not part of GNU CC.
4381 Normally the facilities of the machine's usual C compiler are used, but
4382 this can't be done directly in cross-compilation. You must make your
4383 own arrangements to provide suitable library functions for
4386 On machines where a function returns floating point results in the 80387
4387 register stack, some floating point opcodes may be emitted even if
4388 @samp{-msoft-float} is used.
4390 @item -mno-fp-ret-in-387
4391 Do not use the FPU registers for return values of functions.
4393 The usual calling convention has functions return values of types
4394 @code{float} and @code{double} in an FPU register, even if there
4395 is no FPU. The idea is that the operating system should emulate
4398 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
4399 in ordinary CPU registers instead.
4401 @item -mno-fancy-math-387
4402 Some 387 emulators do not support the @code{sin}, @code{cos} and
4403 @code{sqrt} instructions for the 387. Specify this option to avoid
4404 generating those instructions. This option is the default on FreeBSD.
4405 As of revision 2.6.1, these instructions are not generated unless you
4406 also use the @samp{-ffast-math} switch.
4408 @item -malign-double
4409 @itemx -mno-align-double
4410 Control whether GNU CC aligns @code{double}, @code{long double}, and
4411 @code{long long} variables on a two word boundary or a one word
4412 boundary. Aligning @code{double} variables on a two word boundary will
4413 produce code that runs somewhat faster on a @samp{Pentium} at the
4414 expense of more memory.
4416 @strong{Warning:} if you use the @samp{-malign-double} switch,
4417 structures containing the above types will be aligned differently than
4418 the published application binary interface specifications for the 386.
4421 @itemx -mno-svr3-shlib
4422 Control whether GNU CC places uninitialized locals into @code{bss} or
4423 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
4424 These options are meaningful only on System V Release 3.
4426 @item -mno-wide-multiply
4427 @itemx -mwide-multiply
4428 Control whether GNU CC uses the @code{mul} and @code{imul} that produce
4429 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
4430 long} multiplies and 32-bit division by constants.
4433 Use a different function-calling convention, in which functions that
4434 take a fixed number of arguments return with the @code{ret} @var{num}
4435 instruction, which pops their arguments while returning. This saves one
4436 instruction in the caller since there is no need to pop the arguments
4439 You can specify that an individual function is called with this calling
4440 sequence with the function attribute @samp{stdcall}. You can also
4441 override the @samp{-mrtd} option by using the function attribute
4442 @samp{cdecl}. @xref{Function Attributes}
4444 @strong{Warning:} this calling convention is incompatible with the one
4445 normally used on Unix, so you cannot use it if you need to call
4446 libraries compiled with the Unix compiler.
4448 Also, you must provide function prototypes for all functions that
4449 take variable numbers of arguments (including @code{printf});
4450 otherwise incorrect code will be generated for calls to those
4453 In addition, seriously incorrect code will result if you call a
4454 function with too many arguments. (Normally, extra arguments are
4455 harmlessly ignored.)
4457 @item -mreg-alloc=@var{regs}
4458 Control the default allocation order of integer registers. The
4459 string @var{regs} is a series of letters specifying a register. The
4460 supported letters are: @code{a} allocate EAX; @code{b} allocate EBX;
4461 @code{c} allocate ECX; @code{d} allocate EDX; @code{S} allocate ESI;
4462 @code{D} allocate EDI; @code{B} allocate EBP.
4464 @item -mregparm=@var{num}
4465 Control how many registers are used to pass integer arguments. By
4466 default, no registers are used to pass arguments, and at most 3
4467 registers can be used. You can control this behavior for a specific
4468 function by using the function attribute @samp{regparm}. @xref{Function Attributes}
4470 @strong{Warning:} if you use this switch, and
4471 @var{num} is nonzero, then you must build all modules with the same
4472 value, including any libraries. This includes the system libraries and
4475 @item -malign-loops=@var{num}
4476 Align loops to a 2 raised to a @var{num} byte boundary. If
4477 @samp{-malign-loops} is not specified, the default is 2.
4479 @item -malign-jumps=@var{num}
4480 Align instructions that are only jumped to to a 2 raised to a @var{num}
4481 byte boundary. If @samp{-malign-jumps} is not specified, the default is
4482 2 if optimizing for a 386, and 4 if optimizing for a 486.
4484 @item -malign-functions=@var{num}
4485 Align the start of functions to a 2 raised to @var{num} byte boundary.
4486 If @samp{-malign-jumps} is not specified, the default is 2 if optimizing
4487 for a 386, and 4 if optimizing for a 486.
4491 @subsection HPPA Options
4492 @cindex HPPA Options
4494 These @samp{-m} options are defined for the HPPA family of computers:
4498 Generate code for a PA 1.0 processor.
4501 Generate code for a PA 1.1 processor.
4504 Generate code suitable for big switch tables. Use this option only if
4505 the assembler/linker complain about out of range branches within a switch
4508 @item -mjump-in-delay
4509 Fill delay slots of function calls with unconditional jump instructions
4510 by modifying the return pointer for the function call to be the target
4511 of the conditional jump.
4513 @item -mdisable-fpregs
4514 Prevent floating point registers from being used in any manner. This is
4515 necessary for compiling kernels which perform lazy context switching of
4516 floating point registers. If you use this option and attempt to perform
4517 floating point operations, the compiler will abort.
4519 @item -mdisable-indexing
4520 Prevent the compiler from using indexing address modes. This avoids some
4521 rather obscure problems when compiling MIG generated code under MACH.
4523 @item -mno-space-regs
4524 Generate code that assumes the target has no space registers. This allows
4525 GCC to generate faster indirect calls and use unscaled index address modes.
4527 Such code is suitable for level 0 PA systems and kernels.
4529 @item -mfast-indirect-calls
4530 Generate code that assumes calls never cross space boundaries. This
4531 allows GCC to emit code which performs faster indirect calls.
4533 This option will not work in the presense of shared libraries or nested
4537 Optimize for space rather than execution time. Currently this only
4538 enables out of line function prologues and epilogues. This option is
4539 incompatable with PIC code generation and profiling.
4541 @item -mlong-load-store
4542 Generate 3-instruction load and store sequences as sometimes required by
4543 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
4546 @item -mportable-runtime
4547 Use the portable calling conventions proposed by HP for ELF systems.
4550 Enable the use of assembler directives only GAS understands.
4552 @item -mschedule=@var{cpu type}
4553 Schedule code according to the constraints for the machine type
4554 @var{cpu type}. The choices for @var{cpu type} are @samp{700} for
4555 7@var{n}0 machines, @samp{7100} for 7@var{n}5 machines, and @samp{7100}
4556 for 7@var{n}2 machines. @samp{7100} is the default for @var{cpu type}.
4558 Note the @samp{7100LC} scheduling information is incomplete and using
4559 @samp{7100LC} often leads to bad schedules. For now it's probably best
4560 to use @samp{7100} instead of @samp{7100LC} for the 7@var{n}2 machines.
4563 Enable the optimization pass in the HPUX linker. Note this makes symbolic
4564 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
4565 in which they give bogus error messages when linking some programs.
4568 Generate output containing library calls for floating point.
4569 @strong{Warning:} the requisite libraries are not available for all HPPA
4570 targets. Normally the facilities of the machine's usual C compiler are
4571 used, but this cannot be done directly in cross-compilation. You must make
4572 your own arrangements to provide suitable library functions for
4573 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
4574 does provide software floating point support.
4576 @samp{-msoft-float} changes the calling convention in the output file;
4577 therefore, it is only useful if you compile @emph{all} of a program with
4578 this option. In particular, you need to compile @file{libgcc.a}, the
4579 library that comes with GNU CC, with @samp{-msoft-float} in order for
4583 @node Intel 960 Options
4584 @subsection Intel 960 Options
4586 These @samp{-m} options are defined for the Intel 960 implementations:
4589 @item -m@var{cpu type}
4590 Assume the defaults for the machine type @var{cpu type} for some of
4591 the other options, including instruction scheduling, floating point
4592 support, and addressing modes. The choices for @var{cpu type} are
4593 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
4594 @samp{sa}, and @samp{sb}.
4600 The @samp{-mnumerics} option indicates that the processor does support
4601 floating-point instructions. The @samp{-msoft-float} option indicates
4602 that floating-point support should not be assumed.
4604 @item -mleaf-procedures
4605 @itemx -mno-leaf-procedures
4606 Do (or do not) attempt to alter leaf procedures to be callable with the
4607 @code{bal} instruction as well as @code{call}. This will result in more
4608 efficient code for explicit calls when the @code{bal} instruction can be
4609 substituted by the assembler or linker, but less efficient code in other
4610 cases, such as calls via function pointers, or using a linker that doesn't
4611 support this optimization.
4614 @itemx -mno-tail-call
4615 Do (or do not) make additional attempts (beyond those of the
4616 machine-independent portions of the compiler) to optimize tail-recursive
4617 calls into branches. You may not want to do this because the detection of
4618 cases where this is not valid is not totally complete. The default is
4619 @samp{-mno-tail-call}.
4621 @item -mcomplex-addr
4622 @itemx -mno-complex-addr
4623 Assume (or do not assume) that the use of a complex addressing mode is a
4624 win on this implementation of the i960. Complex addressing modes may not
4625 be worthwhile on the K-series, but they definitely are on the C-series.
4626 The default is currently @samp{-mcomplex-addr} for all processors except
4630 @itemx -mno-code-align
4631 Align code to 8-byte boundaries for faster fetching (or don't bother).
4632 Currently turned on by default for C-series implementations only.
4635 @item -mclean-linkage
4636 @itemx -mno-clean-linkage
4637 These options are not fully implemented.
4641 @itemx -mic2.0-compat
4642 @itemx -mic3.0-compat
4643 Enable compatibility with iC960 v2.0 or v3.0.
4647 Enable compatibility with the iC960 assembler.
4649 @item -mstrict-align
4650 @itemx -mno-strict-align
4651 Do not permit (do permit) unaligned accesses.
4654 Enable structure-alignment compatibility with Intel's gcc release version
4655 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
4658 @node DEC Alpha Options
4659 @subsection DEC Alpha Options
4661 These @samp{-m} options are defined for the DEC Alpha implementations:
4664 @item -mno-soft-float
4666 Use (do not use) the hardware floating-point instructions for
4667 floating-point operations. When @code{-msoft-float} is specified,
4668 functions in @file{libgcc1.c} will be used to perform floating-point
4669 operations. Unless they are replaced by routines that emulate the
4670 floating-point operations, or compiled in such a way as to call such
4671 emulations routines, these routines will issue floating-point
4672 operations. If you are compiling for an Alpha without floating-point
4673 operations, you must ensure that the library is built so as not to call
4676 Note that Alpha implementations without floating-point operations are
4677 required to have floating-point registers.
4681 Generate code that uses (does not use) the floating-point register set.
4682 @code{-mno-fp-regs} implies @code{-msoft-float}. If the floating-point
4683 register set is not used, floating point operands are passed in integer
4684 registers as if they were integers and floating-point results are passed
4685 in $0 instead of $f0. This is a non-standard calling sequence, so any
4686 function with a floating-point argument or return value called by code
4687 compiled with @code{-mno-fp-regs} must also be compiled with that
4690 A typical use of this option is building a kernel that does not use,
4691 and hence need not save and restore, any floating-point registers.
4694 The Alpha architecture implements floating-point hardware optimized for
4695 maximum performance. It is mostly compliant with the IEEE floating
4696 point standard. However, for full compliance, software assistance is
4697 required. This option generates code fully IEEE compliant code
4698 @emph{except} that the @var{inexact flag} is not maintained (see below).
4699 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
4700 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
4701 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
4702 code is less efficient but is able to correctly support denormalized
4703 numbers and exceptional IEEE values such as not-a-number and plus/minus
4704 infinity. Other Alpha compilers call this option
4705 @code{-ieee_with_no_inexact}.
4707 @item -mieee-with-inexact
4708 @c overfull hbox here --bob 22 jul96
4709 @c original text between ignore ... end ignore
4711 This is like @samp{-mieee} except the generated code also maintains the
4712 IEEE @var{inexact flag}. Turning on this option causes the generated
4713 code to implement fully-compliant IEEE math. The option is a shorthand
4714 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
4715 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
4716 implementations the resulting code may execute significantly slower than
4717 the code generated by default. Since there is very little code that
4718 depends on the @var{inexact flag}, you should normally not specify this
4719 option. Other Alpha compilers call this option
4720 @samp{-ieee_with_inexact}.
4722 @c changed paragraph
4723 This is like @samp{-mieee} except the generated code also maintains the
4724 IEEE @var{inexact flag}. Turning on this option causes the generated
4725 code to implement fully-compliant IEEE math. The option is a shorthand
4726 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
4727 @samp{-mieee-conformant},
4728 @samp{-mfp-trap-mode=sui},
4729 and @samp{-mtrap-precision=i}.
4730 On some Alpha implementations the resulting code may execute
4731 significantly slower than the code generated by default. Since there
4732 is very little code that depends on the @var{inexact flag}, you should
4733 normally not specify this option. Other Alpha compilers call this
4734 option @samp{-ieee_with_inexact}.
4735 @c end changes to prevent overfull hboxes
4737 @item -mfp-trap-mode=@var{trap mode}
4738 This option controls what floating-point related traps are enabled.
4739 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
4740 The trap mode can be set to one of four values:
4744 This is the default (normal) setting. The only traps that are enabled
4745 are the ones that cannot be disabled in software (e.g., division by zero
4749 In addition to the traps enabled by @samp{n}, underflow traps are enabled
4753 Like @samp{su}, but the instructions are marked to be safe for software
4754 completion (see Alpha architecture manual for details).
4757 Like @samp{su}, but inexact traps are enabled as well.
4760 @item -mfp-rounding-mode=@var{rounding mode}
4761 Selects the IEEE rounding mode. Other Alpha compilers call this option
4762 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
4767 Normal IEEE rounding mode. Floating point numbers are rounded towards
4768 the nearest machine number or towards the even machine number in case
4772 Round towards minus infinity.
4775 Chopped rounding mode. Floating point numbers are rounded towards zero.
4778 Dynamic rounding mode. A field in the floating point control register
4779 (@var{fpcr}, see Alpha architecture reference manual) controls the
4780 rounding mode in effect. The C library initializes this register for
4781 rounding towards plus infinity. Thus, unless your program modifies the
4782 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.@end table
4784 @item -mtrap-precision=@var{trap precision}
4785 In the Alpha architecture, floating point traps are imprecise. This
4786 means without software assistance it is impossible to recover from a
4787 floating trap and program execution normally needs to be terminated.
4788 GNU CC can generate code that can assist operating system trap handlers
4789 in determining the exact location that caused a floating point trap.
4790 Depending on the requirements of an application, different levels of
4791 precisions can be selected:
4795 Program precision. This option is the default and means a trap handler
4796 can only identify which program caused a floating point exception.
4799 Function precision. The trap handler can determine the function that
4800 caused a floating point exception.
4803 Instruction precision. The trap handler can determine the exact
4804 instruction that caused a floating point exception.
4807 Other Alpha compilers provide the equivalent options called
4808 @samp{-scope_safe} and @samp{-resumption_safe}.
4810 @item -mieee-conformant
4811 This option marks the generated code as IEEE conformant. You must not
4812 use this option unless you also specify @samp{-mtrap-precision=i} and either
4813 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
4814 is to emit the line @samp{.eflag 48} in the function prologue of the
4815 generated assembly file. Under DEC Unix, this has the effect that
4816 IEEE-conformant math library routines will be linked in.
4818 @item -mbuild-constants
4819 Normally GNU CC examines a 32- or 64-bit integer constant to
4820 see if it can construct it from smaller constants in two or three
4821 instructions. If it cannot, it will output the constant as a literal and
4822 generate code to load it from the data segement at runtime.
4824 Use this option to require GNU CC to construct @emph{all} integer constants
4825 using code, even if it takes more instructions (the maximum is six).
4827 You would typically use this option to build a shared library dynamic
4828 loader. Itself a shared library, it must relocate itself in memory
4829 before it can find the variables and constants in its own data segment.
4832 @node Clipper Options
4833 @subsection Clipper Options
4835 These @samp{-m} options are defined for the Clipper implementations:
4839 Produce code for a C300 Clipper processor. This is the default.
4842 Produce code for a C400 Clipper processor i.e. use floating point
4846 @node H8/300 Options
4847 @subsection H8/300 Options
4849 These @samp{-m} options are defined for the H8/300 implementations:
4853 Shorten some address references at link time, when possible; uses the
4854 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
4855 ld.info, Using ld}, for a fuller description.
4858 Generate code for the H8/300H.
4861 Make @code{int} data 32 bits by default.
4864 On the h8/300h, use the same alignment rules as for the h8/300.
4865 The default for the h8/300h is to align longs and floats on 4 byte boundaries.
4866 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
4867 This option has no effect on the h8/300.
4871 @subsection SH Options
4873 These @samp{-m} options are defined for the SH implementations:
4877 Generate code for the SH1.
4880 Generate code for the SH2.
4883 Generate code for the SH3.
4886 Generate code for the SH3e.
4889 Compile code for the processor in big endian mode.
4892 Compile code for the processor in little endian mode.
4895 Shorten some address references at link time, when possible; uses the
4896 linker option @samp{-relax}.
4899 @node System V Options
4900 @subsection Options for System V
4902 These additional options are available on System V Release 4 for
4903 compatibility with other compilers on those systems:
4907 Create a shared object.
4908 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
4911 Identify the versions of each tool used by the compiler, in a
4912 @code{.ident} assembler directive in the output.
4915 Refrain from adding @code{.ident} directives to the output file (this is
4918 @item -YP,@var{dirs}
4919 Search the directories @var{dirs}, and no others, for libraries
4920 specified with @samp{-l}.
4923 Look in the directory @var{dir} to find the M4 preprocessor.
4924 The assembler uses this option.
4925 @c This is supposed to go with a -Yd for predefined M4 macro files, but
4926 @c the generic assembler that comes with Solaris takes just -Ym.
4930 @subsection V850 Options
4931 @cindex V850 Options
4933 These @samp{-m} options are defined for V850 implementations:
4937 @itemx -mno-long-calls
4938 Treat all calls as being far away (near). If calls are assumed to be
4939 far away, the compiler will always load the functions address up into a
4940 register, and call indirect through the pointer.
4944 Do not optimize (do optimize) basic blocks that use the same index
4945 pointer 4 or more times to copy pointer into the @code{ep} register, and
4946 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
4947 option is on by default if you optimize.
4949 @item -mno-prolog-function
4950 @itemx -mprolog-function
4951 Do not use (do use) external functions to save and restore registers at
4952 the prolog and epilog of a function. The external functions are slower,
4953 but use less code space if more than one function saves the same number
4954 of registers. The @samp{-mprolog-function} option is on by default if
4958 Try to make the code as small as possible. At present, this just turns
4959 on the @samp{-mep} and @samp{-mprolog-function} options.
4962 Put static or global variables whose size is @var{n} bytes or less into
4963 the tiny data area that register @code{ep} points to. The tiny data
4964 area can hold up to 256 bytes in total (128 bytes for byte references).
4967 Put static or global variables whose size is @var{n} bytes or less into
4968 the small data area that register @code{gp} points to. The small data
4969 area can hold up to 64 kilobytes.
4972 Put static or global variables whose size is @var{n} bytes or less into
4973 the first 32 kilobytes of memory.
4976 @node Code Gen Options
4977 @section Options for Code Generation Conventions
4978 @cindex code generation conventions
4979 @cindex options, code generation
4980 @cindex run-time options
4982 These machine-independent options control the interface conventions
4983 used in code generation.
4985 Most of them have both positive and negative forms; the negative form
4986 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
4987 one of the forms is listed---the one which is not the default. You
4988 can figure out the other form by either removing @samp{no-} or adding
4992 @item -fpcc-struct-return
4993 Return ``short'' @code{struct} and @code{union} values in memory like
4994 longer ones, rather than in registers. This convention is less
4995 efficient, but it has the advantage of allowing intercallability between
4996 GNU CC-compiled files and files compiled with other compilers.
4998 The precise convention for returning structures in memory depends
4999 on the target configuration macros.
5001 Short structures and unions are those whose size and alignment match
5002 that of some integer type.
5004 @item -freg-struct-return
5005 Use the convention that @code{struct} and @code{union} values are
5006 returned in registers when possible. This is more efficient for small
5007 structures than @samp{-fpcc-struct-return}.
5009 If you specify neither @samp{-fpcc-struct-return} nor its contrary
5010 @samp{-freg-struct-return}, GNU CC defaults to whichever convention is
5011 standard for the target. If there is no standard convention, GNU CC
5012 defaults to @samp{-fpcc-struct-return}, except on targets where GNU CC
5013 is the principal compiler. In those cases, we can choose the standard,
5014 and we chose the more efficient register return alternative.
5017 Allocate to an @code{enum} type only as many bytes as it needs for the
5018 declared range of possible values. Specifically, the @code{enum} type
5019 will be equivalent to the smallest integer type which has enough room.
5021 @item -fshort-double
5022 Use the same size for @code{double} as for @code{float}.
5025 Requests that the data and non-@code{const} variables of this
5026 compilation be shared data rather than private data. The distinction
5027 makes sense only on certain operating systems, where shared data is
5028 shared between processes running the same program, while private data
5029 exists in one copy per process.
5032 Allocate even uninitialized global variables in the bss section of the
5033 object file, rather than generating them as common blocks. This has the
5034 effect that if the same variable is declared (without @code{extern}) in
5035 two different compilations, you will get an error when you link them.
5036 The only reason this might be useful is if you wish to verify that the
5037 program will work on other systems which always work this way.
5040 Ignore the @samp{#ident} directive.
5042 @item -fno-gnu-linker
5043 Do not output global initializations (such as C++ constructors and
5044 destructors) in the form used by the GNU linker (on systems where the GNU
5045 linker is the standard method of handling them). Use this option when
5046 you want to use a non-GNU linker, which also requires using the
5047 @code{collect2} program to make sure the system linker includes
5048 constructors and destructors. (@code{collect2} is included in the GNU CC
5049 distribution.) For systems which @emph{must} use @code{collect2}, the
5050 compiler driver @code{gcc} is configured to do this automatically.
5052 @item -finhibit-size-directive
5053 Don't output a @code{.size} assembler directive, or anything else that
5054 would cause trouble if the function is split in the middle, and the
5055 two halves are placed at locations far apart in memory. This option is
5056 used when compiling @file{crtstuff.c}; you should not need to use it
5060 Put extra commentary information in the generated assembly code to
5061 make it more readable. This option is generally only of use to those
5062 who actually need to read the generated assembly code (perhaps while
5063 debugging the compiler itself).
5065 @samp{-fverbose-asm} is the default. @samp{-fno-verbose-asm} causes the
5066 extra information to be omitted and is useful when comparing two assembler
5070 Consider all memory references through pointers to be volatile.
5072 @item -fvolatile-global
5073 Consider all memory references to extern and global data items to
5077 @cindex global offset table
5079 Generate position-independent code (PIC) suitable for use in a shared
5080 library, if supported for the target machine. Such code accesses all
5081 constant addresses through a global offset table (GOT). The dynamic
5082 loader resolves the GOT entries when the program starts (the dynamic
5083 loader is not part of GNU CC; it is part of the operating system). If
5084 the GOT size for the linked executable exceeds a machine-specific
5085 maximum size, you get an error message from the linker indicating that
5086 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
5087 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
5088 on the m68k and RS/6000. The 386 has no such limit.)
5090 Position-independent code requires special support, and therefore works
5091 only on certain machines. For the 386, GNU CC supports PIC for System V
5092 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
5093 position-independent.
5096 If supported for the target machine, emit position-independent code,
5097 suitable for dynamic linking and avoiding any limit on the size of the
5098 global offset table. This option makes a difference on the m68k, m88k,
5101 Position-independent code requires special support, and therefore works
5102 only on certain machines.
5104 @item -ffixed-@var{reg}
5105 Treat the register named @var{reg} as a fixed register; generated code
5106 should never refer to it (except perhaps as a stack pointer, frame
5107 pointer or in some other fixed role).
5109 @var{reg} must be the name of a register. The register names accepted
5110 are machine-specific and are defined in the @code{REGISTER_NAMES}
5111 macro in the machine description macro file.
5113 This flag does not have a negative form, because it specifies a
5116 @item -fcall-used-@var{reg}
5117 Treat the register named @var{reg} as an allocatable register that is
5118 clobbered by function calls. It may be allocated for temporaries or
5119 variables that do not live across a call. Functions compiled this way
5120 will not save and restore the register @var{reg}.
5122 Use of this flag for a register that has a fixed pervasive role in the
5123 machine's execution model, such as the stack pointer or frame pointer,
5124 will produce disastrous results.
5126 This flag does not have a negative form, because it specifies a
5129 @item -fcall-saved-@var{reg}
5130 Treat the register named @var{reg} as an allocatable register saved by
5131 functions. It may be allocated even for temporaries or variables that
5132 live across a call. Functions compiled this way will save and restore
5133 the register @var{reg} if they use it.
5135 Use of this flag for a register that has a fixed pervasive role in the
5136 machine's execution model, such as the stack pointer or frame pointer,
5137 will produce disastrous results.
5139 A different sort of disaster will result from the use of this flag for
5140 a register in which function values may be returned.
5142 This flag does not have a negative form, because it specifies a
5146 Pack all structure members together without holes. Usually you would
5147 not want to use this option, since it makes the code suboptimal, and
5148 the offsets of structure members won't agree with system libraries.
5150 @item -fcheck-memory-usage
5151 Generate extra code to check each memory access. GNU CC will generate
5152 code that is suitable for a detector of bad memory accesses such as
5153 @file{Checker}. If you specify this option, you can not use the
5154 @code{asm} or @code{__asm__} keywords.
5156 You must also specify this option when you compile functions you call that
5157 have side effects. If you do not, you may get erronous messages from
5158 the detector. Normally, you should compile all your code with this option.
5159 If you use functions from a library that have side-effects (such as
5160 @code{read}), you may not be able to recompile the library and
5161 specify this option. In that case, you can enable the
5162 @samp{-fprefix-function-name} option, which requests GNU CC to encapsulate
5163 your code and make other functions look as if they were compiled with
5164 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
5165 which are provided by the detector. If you cannot find or build
5166 stubs for every function you call, you may have to specify
5167 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
5169 @item -fprefix-function-name
5170 Request GNU CC to add a prefix to the symbols generated for function names.
5171 GNU CC adds a prefix to the names of functions defined as well as
5172 functions called. Code compiled with this option and code compiled
5173 without the option can't be linked together, unless or stubs are used.
5175 If you compile the following code with @samp{-fprefix-function-name}
5177 extern void bar (int);
5187 GNU CC will compile the code as if it was written:
5189 extern void prefix_bar (int);
5193 return prefix_bar (a + 5);
5196 This option is designed to be used with @samp{-fcheck-memory-usage}.
5199 Generate code to verify that you do not go beyond the boundary of the
5200 stack. You should specify this flag if you are running in an
5201 environment with multiple threads, but only rarely need to specify it in
5202 a single-threaded environment since stack overflow is automatically
5203 detected on nearly all systems if there is only one stack.
5207 Control whether virtual function definitions in classes are used to
5208 generate code, or only to define interfaces for their callers. (C++
5211 These options are provided for compatibility with @code{cfront} 1.x
5212 usage; the recommended alternative GNU C++ usage is in flux. @xref{C++
5213 Interface,,Declarations and Definitions in One Header}.
5215 With @samp{+e0}, virtual function definitions in classes are declared
5216 @code{extern}; the declaration is used only as an interface
5217 specification, not to generate code for the virtual functions (in this
5220 With @samp{+e1}, G++ actually generates the code implementing virtual
5221 functions defined in the code, and makes them publicly visible.
5223 @cindex aliasing of parameters
5224 @cindex parameters, aliased
5225 @item -fargument-alias
5226 @item -fargument-noalias
5227 @item -fargument-noalias-global
5228 Specify the possible relationships among parameters and between
5229 parameters and global data.
5231 @samp{-fargument-alias} specifies that arguments (parameters) may
5232 alias each other and may alias global storage.
5233 @samp{-fargument-noalias} specifies that arguments do not alias
5234 each other, but may alias global storage.
5235 @samp{-fargument-noalias-global} specifies that arguments do not
5236 alias each other and do not alias global storage.
5238 Each language will automatically use whatever option is required by
5239 the language standard. You should not need to use these options yourself.
5242 @node Environment Variables
5243 @section Environment Variables Affecting GNU CC
5244 @cindex environment variables
5246 This section describes several environment variables that affect how GNU
5247 CC operates. They work by specifying directories or prefixes to use
5248 when searching for various kinds of files.
5251 Note that you can also specify places to search using options such as
5252 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
5253 take precedence over places specified using environment variables, which
5254 in turn take precedence over those specified by the configuration of GNU
5258 Note that you can also specify places to search using options such as
5259 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
5260 take precedence over places specified using environment variables, which
5261 in turn take precedence over those specified by the configuration of GNU
5268 If @code{TMPDIR} is set, it specifies the directory to use for temporary
5269 files. GNU CC uses temporary files to hold the output of one stage of
5270 compilation which is to be used as input to the next stage: for example,
5271 the output of the preprocessor, which is the input to the compiler
5274 @item GCC_EXEC_PREFIX
5275 @findex GCC_EXEC_PREFIX
5276 If @code{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
5277 names of the subprograms executed by the compiler. No slash is added
5278 when this prefix is combined with the name of a subprogram, but you can
5279 specify a prefix that ends with a slash if you wish.
5281 If GNU CC cannot find the subprogram using the specified prefix, it
5282 tries looking in the usual places for the subprogram.
5284 The default value of @code{GCC_EXEC_PREFIX} is
5285 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
5286 of @code{prefix} when you ran the @file{configure} script.
5288 Other prefixes specified with @samp{-B} take precedence over this prefix.
5290 This prefix is also used for finding files such as @file{crt0.o} that are
5293 In addition, the prefix is used in an unusual way in finding the
5294 directories to search for header files. For each of the standard
5295 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
5296 (more precisely, with the value of @code{GCC_INCLUDE_DIR}), GNU CC tries
5297 replacing that beginning with the specified prefix to produce an
5298 alternate directory name. Thus, with @samp{-Bfoo/}, GNU CC will search
5299 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
5300 These alternate directories are searched first; the standard directories
5304 @findex COMPILER_PATH
5305 The value of @code{COMPILER_PATH} is a colon-separated list of
5306 directories, much like @code{PATH}. GNU CC tries the directories thus
5307 specified when searching for subprograms, if it can't find the
5308 subprograms using @code{GCC_EXEC_PREFIX}.
5311 @findex LIBRARY_PATH
5312 The value of @code{LIBRARY_PATH} is a colon-separated list of
5313 directories, much like @code{PATH}. When configured as a native compiler,
5314 GNU CC tries the directories thus specified when searching for special
5315 linker files, if it can't find them using @code{GCC_EXEC_PREFIX}. Linking
5316 using GNU CC also uses these directories when searching for ordinary
5317 libraries for the @samp{-l} option (but directories specified with
5318 @samp{-L} come first).
5320 @item C_INCLUDE_PATH
5321 @itemx CPLUS_INCLUDE_PATH
5322 @itemx OBJC_INCLUDE_PATH
5323 @findex C_INCLUDE_PATH
5324 @findex CPLUS_INCLUDE_PATH
5325 @findex OBJC_INCLUDE_PATH
5326 @c @itemx OBJCPLUS_INCLUDE_PATH
5327 These environment variables pertain to particular languages. Each
5328 variable's value is a colon-separated list of directories, much like
5329 @code{PATH}. When GNU CC searches for header files, it tries the
5330 directories listed in the variable for the language you are using, after
5331 the directories specified with @samp{-I} but before the standard header
5334 @item DEPENDENCIES_OUTPUT
5335 @findex DEPENDENCIES_OUTPUT
5336 @cindex dependencies for make as output
5337 If this variable is set, its value specifies how to output dependencies
5338 for Make based on the header files processed by the compiler. This
5339 output looks much like the output from the @samp{-M} option
5340 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
5341 in addition to the usual results of compilation.
5343 The value of @code{DEPENDENCIES_OUTPUT} can be just a file name, in
5344 which case the Make rules are written to that file, guessing the target
5345 name from the source file name. Or the value can have the form
5346 @samp{@var{file} @var{target}}, in which case the rules are written to
5347 file @var{file} using @var{target} as the target name.
5350 @node Running Protoize
5351 @section Running Protoize
5353 The program @code{protoize} is an optional part of GNU C. You can use
5354 it to add prototypes to a program, thus converting the program to ANSI
5355 C in one respect. The companion program @code{unprotoize} does the
5356 reverse: it removes argument types from any prototypes that are found.
5358 When you run these programs, you must specify a set of source files as
5359 command line arguments. The conversion programs start out by compiling
5360 these files to see what functions they define. The information gathered
5361 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
5363 After scanning comes actual conversion. The specified files are all
5364 eligible to be converted; any files they include (whether sources or
5365 just headers) are eligible as well.
5367 But not all the eligible files are converted. By default,
5368 @code{protoize} and @code{unprotoize} convert only source and header
5369 files in the current directory. You can specify additional directories
5370 whose files should be converted with the @samp{-d @var{directory}}
5371 option. You can also specify particular files to exclude with the
5372 @samp{-x @var{file}} option. A file is converted if it is eligible, its
5373 directory name matches one of the specified directory names, and its
5374 name within the directory has not been excluded.
5376 Basic conversion with @code{protoize} consists of rewriting most
5377 function definitions and function declarations to specify the types of
5378 the arguments. The only ones not rewritten are those for varargs
5381 @code{protoize} optionally inserts prototype declarations at the
5382 beginning of the source file, to make them available for any calls that
5383 precede the function's definition. Or it can insert prototype
5384 declarations with block scope in the blocks where undeclared functions
5387 Basic conversion with @code{unprotoize} consists of rewriting most
5388 function declarations to remove any argument types, and rewriting
5389 function definitions to the old-style pre-ANSI form.
5391 Both conversion programs print a warning for any function declaration or
5392 definition that they can't convert. You can suppress these warnings
5395 The output from @code{protoize} or @code{unprotoize} replaces the
5396 original source file. The original file is renamed to a name ending
5397 with @samp{.save}. If the @samp{.save} file already exists, then
5398 the source file is simply discarded.
5400 @code{protoize} and @code{unprotoize} both depend on GNU CC itself to
5401 scan the program and collect information about the functions it uses.
5402 So neither of these programs will work until GNU CC is installed.
5404 Here is a table of the options you can use with @code{protoize} and
5405 @code{unprotoize}. Each option works with both programs unless
5409 @item -B @var{directory}
5410 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
5411 usual directory (normally @file{/usr/local/lib}). This file contains
5412 prototype information about standard system functions. This option
5413 applies only to @code{protoize}.
5415 @item -c @var{compilation-options}
5416 Use @var{compilation-options} as the options when running @code{gcc} to
5417 produce the @samp{.X} files. The special option @samp{-aux-info} is
5418 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
5420 Note that the compilation options must be given as a single argument to
5421 @code{protoize} or @code{unprotoize}. If you want to specify several
5422 @code{gcc} options, you must quote the entire set of compilation options
5423 to make them a single word in the shell.
5425 There are certain @code{gcc} arguments that you cannot use, because they
5426 would produce the wrong kind of output. These include @samp{-g},
5427 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
5428 the @var{compilation-options}, they are ignored.
5431 Rename files to end in @samp{.C} instead of @samp{.c}.
5432 This is convenient if you are converting a C program to C++.
5433 This option applies only to @code{protoize}.
5436 Add explicit global declarations. This means inserting explicit
5437 declarations at the beginning of each source file for each function
5438 that is called in the file and was not declared. These declarations
5439 precede the first function definition that contains a call to an
5440 undeclared function. This option applies only to @code{protoize}.
5442 @item -i @var{string}
5443 Indent old-style parameter declarations with the string @var{string}.
5444 This option applies only to @code{protoize}.
5446 @code{unprotoize} converts prototyped function definitions to old-style
5447 function definitions, where the arguments are declared between the
5448 argument list and the initial @samp{@{}. By default, @code{unprotoize}
5449 uses five spaces as the indentation. If you want to indent with just
5450 one space instead, use @samp{-i " "}.
5453 Keep the @samp{.X} files. Normally, they are deleted after conversion
5457 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
5458 a prototype declaration for each function in each block which calls the
5459 function without any declaration. This option applies only to
5463 Make no real changes. This mode just prints information about the conversions
5464 that would have been done without @samp{-n}.
5467 Make no @samp{.save} files. The original files are simply deleted.
5468 Use this option with caution.
5470 @item -p @var{program}
5471 Use the program @var{program} as the compiler. Normally, the name
5475 Work quietly. Most warnings are suppressed.
5478 Print the version number, just like @samp{-v} for @code{gcc}.
5481 If you need special compiler options to compile one of your program's
5482 source files, then you should generate that file's @samp{.X} file
5483 specially, by running @code{gcc} on that source file with the
5484 appropriate options and the option @samp{-aux-info}. Then run
5485 @code{protoize} on the entire set of files. @code{protoize} will use
5486 the existing @samp{.X} file because it is newer than the source file.
5490 gcc -Dfoo=bar file1.c -aux-info
5495 You need to include the special files along with the rest in the
5496 @code{protoize} command, even though their @samp{.X} files already
5497 exist, because otherwise they won't get converted.
5499 @xref{Protoize Caveats}, for more information on how to use
5500 @code{protoize} successfully.