1 @c Copyright (C) 1988, 89, 92-98, 1999 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 GCC Command Options
7 @cindex GCC command options
8 @cindex command options
9 @cindex options, GCC command
11 When you invoke GCC, 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 GCC 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 * Spec Files:: How to pass switches to sub-processes.
72 * Target Options:: Running a cross-compiler, or an old version of GCC.
73 * Submodel Options:: Specifying minor hardware or convention variations,
74 such as 68010 vs 68020.
75 * Code Gen Options:: Specifying conventions for function calls, data layout
77 * Environment Variables:: Env vars that affect GCC.
78 * Running Protoize:: Automatically adding or removing function prototypes.
82 @section Option Summary
84 Here is a summary of all the options, grouped by type. Explanations are
85 in the following sections.
89 @xref{Overall Options,,Options Controlling the Kind of Output}.
91 -c -S -E -o @var{file} -pipe -v --help -x @var{language}
94 @item C Language Options
95 @xref{C Dialect Options,,Options Controlling C Dialect}.
97 -ansi -flang-isoc9x -fallow-single-precision -fcond-mismatch -fno-asm
98 -fno-builtin -ffreestanding -fhosted -fsigned-bitfields -fsigned-char
99 -funsigned-bitfields -funsigned-char -fwritable-strings
100 -traditional -traditional-cpp -trigraphs
103 @item C++ Language Options
104 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
106 -fno-access-control -fcheck-new -fconserve-space -fdollars-in-identifiers
107 -fno-elide-constructors -fexternal-templates -ffor-scope
108 -fno-for-scope -fno-gnu-keywords -fguiding-decls
109 -fhonor-std -fhuge-objects -fno-implicit-templates -finit-priority
110 -fno-implement-inlines -fname-mangling-version-@var{n} -fno-default-inline
111 -foperator-names -fno-optional-diags -fpermissive -frepo -fstrict-prototype
112 -fsquangle -ftemplate-depth-@var{n} -fthis-is-variable -fvtable-thunks
113 -nostdinc++ -Wctor-dtor-privacy -Wno-deprecated -Weffc++
114 -Wno-non-template-friend
115 -Wnon-virtual-dtor -Wold-style-cast -Woverloaded-virtual
116 -Wno-pmf-conversions -Wreorder -Wsign-promo -Wsynth
119 @item Warning Options
120 @xref{Warning Options,,Options to Request or Suppress Warnings}.
122 -fsyntax-only -pedantic -pedantic-errors
123 -w -W -Wall -Waggregate-return -Wbad-function-cast
124 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment
125 -Wconversion -Werror -Wformat
126 -Wid-clash-@var{len} -Wimplicit -Wimplicit-int
127 -Wimplicit-function-declaration -Wimport
128 -Werror-implicit-function-declaration -Winline
129 -Wlarger-than-@var{len} -Wlong-long
130 -Wmain -Wmissing-declarations -Wmissing-noreturn
131 -Wmissing-prototypes -Wmultichar -Wnested-externs -Wno-import
132 -Wparentheses -Wpointer-arith -Wredundant-decls
133 -Wreturn-type -Wshadow -Wsign-compare -Wstrict-prototypes
134 -Wswitch -Wtraditional
135 -Wtrigraphs -Wundef -Wuninitialized -Wunknown-pragmas -Wunreachable-code
136 -Wunused -Wwrite-strings
139 @item Debugging Options
140 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
142 -a -ax -d@var{letters} -fdump-unnumbered -fdump-translation-unit-@var{file}
143 -fpretend-float -fprofile-arcs -ftest-coverage
144 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2
145 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+
146 -p -pg -print-file-name=@var{library} -print-libgcc-file-name
147 -print-prog-name=@var{program} -print-search-dirs -save-temps -time
150 @item Optimization Options
151 @xref{Optimize Options,,Options that Control Optimization}.
153 -falign-functions=@var{n} -falign-labels=@var{n} -falign-loops=@var{n}
154 -falign-jumps=@var{n} -fbranch-probabilities
155 -fcaller-saves -fcse-follow-jumps -fcse-skip-blocks
156 -fdelayed-branch -fexpensive-optimizations
157 -ffast-math -ffloat-store -fforce-addr -fforce-mem
158 -fdata-sections -ffunction-sections -fgcse
159 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions
160 -fmove-all-movables -fno-default-inline -fno-defer-pop
161 -fno-function-cse -fno-inline -fno-peephole
162 -fomit-frame-pointer -foptimize-register-moves -fregmove
163 -frerun-cse-after-loop -frerun-loop-opt -freduce-all-givs
164 -fschedule-insns -fschedule-insns2 -fstrength-reduce
165 -fstrict-aliasing -fthread-jumps -funroll-all-loops
167 -O -O0 -O1 -O2 -O3 -Os
170 @item Preprocessor Options
171 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
173 -A@var{question}(@var{answer}) -C -dD -dM -dN
174 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H
176 -include @var{file} -imacros @var{file}
177 -iprefix @var{file} -iwithprefix @var{dir}
178 -iwithprefixbefore @var{dir} -isystem @var{dir} -isystem-c++ @var{dir}
179 -M -MD -MM -MMD -MG -nostdinc -P -trigraphs
180 -undef -U@var{macro} -Wp,@var{option}
183 @item Assembler Option
184 @xref{Assembler Options,,Passing Options to the Assembler}.
190 @xref{Link Options,,Options for Linking}.
192 @var{object-file-name} -l@var{library}
193 -nostartfiles -nodefaultlibs -nostdlib
194 -s -static -shared -symbolic
195 -Wl,@var{option} -Xlinker @var{option}
199 @item Directory Options
200 @xref{Directory Options,,Options for Directory Search}.
202 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}
206 @c I wrote this xref this way to avoid overfull hbox. -- rms
207 @xref{Target Options}.
209 -b @var{machine} -V @var{version}
212 @item Machine Dependent Options
213 @xref{Submodel Options,,Hardware Models and Configurations}.
215 @emph{M680x0 Options}
216 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040
217 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020
218 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel
226 -mtune=@var{cpu type}
227 -mcmodel=@var{code model}
229 -mapp-regs -mbroken-saverestore -mcypress -mepilogue
230 -mflat -mfpu -mhard-float -mhard-quad-float
231 -mimpure-text -mlive-g0 -mno-app-regs -mno-epilogue
232 -mno-flat -mno-fpu -mno-impure-text
233 -mno-stack-bias -mno-unaligned-doubles
234 -msoft-float -msoft-quad-float -msparclite -mstack-bias
235 -msupersparc -munaligned-doubles -mv8
237 @emph{Convex Options}
238 -mc1 -mc2 -mc32 -mc34 -mc38
239 -margcount -mnoargcount
241 -mvolatile-cache -mvolatile-nocache
243 @emph{AMD29K Options}
244 -m29000 -m29050 -mbw -mnbw -mdw -mndw
245 -mlarge -mnormal -msmall
246 -mkernel-registers -mno-reuse-arg-regs
247 -mno-stack-check -mno-storem-bug
248 -mreuse-arg-regs -msoft-float -mstack-check
249 -mstorem-bug -muser-registers
252 -mapcs-frame -mno-apcs-frame
254 -mapcs-stack-check -mno-apcs-stack-check
255 -mapcs-float -mno-apcs-float
256 -mapcs-reentrant -mno-apcs-reentrant
257 -msched-prolog -mno-sched-prolog
258 -mlittle-endian -mbig-endian -mwords-little-endian
259 -mshort-load-bytes -mno-short-load-bytes -mshort-load-words -mno-short-load-words
260 -msoft-float -mhard-float -mfpe
261 -mthumb-interwork -mno-thumb-interwork
262 -mcpu= -march= -mfpe=
263 -mstructure-size-boundary=
264 -mbsd -mxopen -mno-symrename
266 -mnop-fun-dllimport -mno-nop-fun-dllimport
267 -msingle-pic-base -mno-single-pic-base
271 -mtpcs-frame -mno-tpcs-frame
272 -mtpcs-leaf-frame -mno-tpcs-leaf-frame
273 -mlittle-endian -mbig-endian
274 -mthumb-interwork -mno-thumb-interwork
275 -mstructure-size-boundary=
276 -mnop-fun-dllimport -mno-nop-fun-dllimport
277 -mcallee-super-interworking -mno-callee-super-interworking
278 -mcaller-super-interworking -mno-caller-super-interworking
279 -msingle-pic-base -mno-single-pic-base
282 @emph{MN10200 Options}
285 @emph{MN10300 Options}
290 @emph{M32R/D Options}
291 -mcode-model=@var{model type} -msdata=@var{sdata type}
295 -m88000 -m88100 -m88110 -mbig-pic
296 -mcheck-zero-division -mhandle-large-shift
297 -midentify-revision -mno-check-zero-division
298 -mno-ocs-debug-info -mno-ocs-frame-position
299 -mno-optimize-arg-area -mno-serialize-volatile
300 -mno-underscores -mocs-debug-info
301 -mocs-frame-position -moptimize-arg-area
302 -mserialize-volatile -mshort-data-@var{num} -msvr3
303 -msvr4 -mtrap-large-shift -muse-div-instruction
304 -mversion-03.00 -mwarn-passed-structs
306 @emph{RS/6000 and PowerPC Options}
308 -mtune=@var{cpu type}
309 -mpower -mno-power -mpower2 -mno-power2
310 -mpowerpc -mpowerpc64 -mno-powerpc
311 -mpowerpc-gpopt -mno-powerpc-gpopt
312 -mpowerpc-gfxopt -mno-powerpc-gfxopt
313 -mnew-mnemonics -mno-new-mnemonics
314 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc
315 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe
316 -msoft-float -mhard-float -mmultiple -mno-multiple
317 -mstring -mno-string -mupdate -mno-update
318 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align
319 -mstrict-align -mno-strict-align -mrelocatable
320 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib
321 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian
322 -mcall-aix -mcall-sysv -mprototype -mno-prototype
323 -msim -mmvme -mads -myellowknife -memb -msdata
324 -msdata=@var{opt} -G @var{num}
327 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs
328 -mfull-fp-blocks -mhc-struct-return -min-line-mul
329 -mminimum-fp-blocks -mnohc-struct-return
332 -mabicalls -mcpu=@var{cpu type} -membedded-data
333 -membedded-pic -mfp32 -mfp64 -mgas -mgp32 -mgp64
334 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1
335 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy
336 -mmips-as -mmips-tfile -mno-abicalls
337 -mno-embedded-data -mno-embedded-pic
338 -mno-gpopt -mno-long-calls
339 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats
340 -mrnames -msoft-float
341 -m4650 -msingle-float -mmad
342 -mstats -EL -EB -G @var{num} -nocpp
343 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi
347 -march=@var{cpu type}
348 -mieee-fp -mno-fancy-math-387
349 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib
350 -mno-wide-multiply -mrtd -malign-double
351 -mreg-alloc=@var{list} -mregparm=@var{num}
352 -malign-jumps=@var{num} -malign-loops=@var{num}
353 -malign-functions=@var{num} -mpreferred-stack-boundary=@var{num}
356 -march=@var{architecture type}
357 -mbig-switch -mdisable-fpregs -mdisable-indexing
358 -mfast-indirect-calls -mgas -mjump-in-delay
359 -mlong-load-store -mno-big-switch -mno-disable-fpregs
360 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas
361 -mno-jump-in-delay -mno-long-load-store
362 -mno-portable-runtime -mno-soft-float
363 -mno-space-regs -msoft-float -mpa-risc-1-0
364 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime
365 -mschedule=@var{cpu type} -mspace-regs
367 @emph{Intel 960 Options}
368 -m@var{cpu type} -masm-compat -mclean-linkage
369 -mcode-align -mcomplex-addr -mleaf-procedures
370 -mic-compat -mic2.0-compat -mic3.0-compat
371 -mintel-asm -mno-clean-linkage -mno-code-align
372 -mno-complex-addr -mno-leaf-procedures
373 -mno-old-align -mno-strict-align -mno-tail-call
374 -mnumerics -mold-align -msoft-float -mstrict-align
377 @emph{DEC Alpha Options}
378 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float
380 -mieee -mieee-with-inexact -mieee-conformant
381 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode}
382 -mtrap-precision=@var{mode} -mbuild-constants
384 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max
385 -mmemory-latency=@var{time}
387 @emph{Clipper Options}
390 @emph{H8/300 Options}
391 -mrelax -mh -ms -mint32 -malign-300
394 -m1 -m2 -m3 -m3e -mb -ml -mdalign -mrelax
396 @emph{System V Options}
397 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}
401 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text section}
402 -mdata=@var{data section} -mrodata=@var{readonly data section}
404 @emph{TMS320C3x/C4x Options}
405 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm
406 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload
407 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned
408 -mparallel-insns -mparallel-mpy -mpreserve-float
411 -mlong-calls -mno-long-calls -mep -mno-ep
412 -mprolog-function -mno-prolog-function -mspace
413 -mtda=@var{n} -msda=@var{n} -mzda=@var{n}
417 -m32032 -m32332 -m32532 -m32081 -m32381 -mmult-add -mnomult-add
418 -msoft-float -mrtd -mnortd -mregparam -mnoregparam -msb -mnosb
419 -mbitfield -mnobitfield -mhimem -mnohimem
422 @item Code Generation Options
423 @xref{Code Gen Options,,Options for Code Generation Conventions}.
425 -fcall-saved-@var{reg} -fcall-used-@var{reg}
426 -fexceptions -ffixed-@var{reg} -finhibit-size-directive
427 -fcheck-memory-usage -fprefix-function-name
428 -fno-common -fno-ident -fno-gnu-linker
429 -fpcc-struct-return -fpic -fPIC
430 -freg-struct-return -fshared-data -fshort-enums
431 -fshort-double -fvolatile -fvolatile-global -fvolatile-static
432 -fverbose-asm -fpack-struct -fstack-check
433 -fargument-alias -fargument-noalias
434 -fargument-noalias-global
440 * Overall Options:: Controlling the kind of output:
441 an executable, object files, assembler files,
442 or preprocessed source.
443 * C Dialect Options:: Controlling the variant of C language compiled.
444 * C++ Dialect Options:: Variations on C++.
445 * Warning Options:: How picky should the compiler be?
446 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
447 * Optimize Options:: How much optimization?
448 * Preprocessor Options:: Controlling header files and macro definitions.
449 Also, getting dependency information for Make.
450 * Assembler Options:: Passing options to the assembler.
451 * Link Options:: Specifying libraries and so on.
452 * Directory Options:: Where to find header files and libraries.
453 Where to find the compiler executable files.
454 * Spec Files:: How to pass switches to sub-processes.
455 * Target Options:: Running a cross-compiler, or an old version of GCC.
458 @node Overall Options
459 @section Options Controlling the Kind of Output
461 Compilation can involve up to four stages: preprocessing, compilation
462 proper, assembly and linking, always in that order. The first three
463 stages apply to an individual source file, and end by producing an
464 object file; linking combines all the object files (those newly
465 compiled, and those specified as input) into an executable file.
467 @cindex file name suffix
468 For any given input file, the file name suffix determines what kind of
473 C source code which must be preprocessed.
476 C source code which should not be preprocessed.
479 C++ source code which should not be preprocessed.
482 Objective-C source code. Note that you must link with the library
483 @file{libobjc.a} to make an Objective-C program work.
486 C header file (not to be compiled or linked).
489 @itemx @var{file}.cxx
490 @itemx @var{file}.cpp
492 C++ source code which must be preprocessed. Note that in @samp{.cxx},
493 the last two letters must both be literally @samp{x}. Likewise,
494 @samp{.C} refers to a literal capital C.
500 Assembler code which must be preprocessed.
503 An object file to be fed straight into linking.
504 Any file name with no recognized suffix is treated this way.
507 You can specify the input language explicitly with the @samp{-x} option:
510 @item -x @var{language}
511 Specify explicitly the @var{language} for the following input files
512 (rather than letting the compiler choose a default based on the file
513 name suffix). This option applies to all following input files until
514 the next @samp{-x} option. Possible values for @var{language} are:
517 c-header cpp-output c++-cpp-output
518 assembler assembler-with-cpp
522 Turn off any specification of a language, so that subsequent files are
523 handled according to their file name suffixes (as they are if @samp{-x}
524 has not been used at all).
527 If you only want some of the stages of compilation, you can use
528 @samp{-x} (or filename suffixes) to tell @code{gcc} where to start, and
529 one of the options @samp{-c}, @samp{-S}, or @samp{-E} to say where
530 @code{gcc} is to stop. Note that some combinations (for example,
531 @samp{-x cpp-output -E} instruct @code{gcc} to do nothing at all.
535 Compile or assemble the source files, but do not link. The linking
536 stage simply is not done. The ultimate output is in the form of an
537 object file for each source file.
539 By default, the object file name for a source file is made by replacing
540 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
542 Unrecognized input files, not requiring compilation or assembly, are
546 Stop after the stage of compilation proper; do not assemble. The output
547 is in the form of an assembler code file for each non-assembler input
550 By default, the assembler file name for a source file is made by
551 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
553 Input files that don't require compilation are ignored.
556 Stop after the preprocessing stage; do not run the compiler proper. The
557 output is in the form of preprocessed source code, which is sent to the
560 Input files which don't require preprocessing are ignored.
562 @cindex output file option
564 Place output in file @var{file}. This applies regardless to whatever
565 sort of output is being produced, whether it be an executable file,
566 an object file, an assembler file or preprocessed C code.
568 Since only one output file can be specified, it does not make sense to
569 use @samp{-o} when compiling more than one input file, unless you are
570 producing an executable file as output.
572 If @samp{-o} is not specified, the default is to put an executable file
573 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
574 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
575 all preprocessed C source on standard output.@refill
578 Print (on standard error output) the commands executed to run the stages
579 of compilation. Also print the version number of the compiler driver
580 program and of the preprocessor and the compiler proper.
583 Use pipes rather than temporary files for communication between the
584 various stages of compilation. This fails to work on some systems where
585 the assembler is unable to read from a pipe; but the GNU assembler has
589 Print (on the standard output) a description of the command line options
590 understood by @code{gcc}. If the @code{-v} option is also specified
591 then @code{--help} will also be passed on to the various processes
592 invoked by @code{gcc}, so that they can display the command line options
593 they accept. If the @code{-W} option is also specified then command
594 line options which have no documentation associated with them will also
599 @section Compiling C++ Programs
601 @cindex suffixes for C++ source
602 @cindex C++ source file suffixes
603 C++ source files conventionally use one of the suffixes @samp{.C},
604 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
605 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
606 files with these names and compiles them as C++ programs even if you
607 call the compiler the same way as for compiling C programs (usually with
608 the name @code{gcc}).
612 However, C++ programs often require class libraries as well as a
613 compiler that understands the C++ language---and under some
614 circumstances, you might want to compile programs from standard input,
615 or otherwise without a suffix that flags them as C++ programs.
616 @code{g++} is a program that calls GCC with the default language
617 set to C++, and automatically specifies linking against the C++
618 library. On many systems, the script @code{g++} is also
619 installed with the name @code{c++}.
621 @cindex invoking @code{g++}
622 When you compile C++ programs, you may specify many of the same
623 command-line options that you use for compiling programs in any
624 language; or command-line options meaningful for C and related
625 languages; or options that are meaningful only for C++ programs.
626 @xref{C Dialect Options,,Options Controlling C Dialect}, for
627 explanations of options for languages related to C.
628 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
629 explanations of options that are meaningful only for C++ programs.
631 @node C Dialect Options
632 @section Options Controlling C Dialect
633 @cindex dialect options
634 @cindex language dialect options
635 @cindex options, dialect
637 The following options control the dialect of C (or languages derived
638 from C, such as C++ and Objective C) that the compiler accepts:
643 In C mode, support all ANSI standard C programs. In C++ mode,
644 remove GNU extensions that conflict with ANSI C++.
645 @c shouldn't we be saying "ISO"?
647 This turns off certain features of GCC that are incompatible with ANSI
648 C (when compiling C code), or of ANSI standard C++ (when compiling C++ code),
649 such as the @code{asm} and @code{typeof} keywords, and
650 predefined macros such as @code{unix} and @code{vax} that identify the
651 type of system you are using. It also enables the undesirable and
652 rarely used ANSI trigraph feature. For the C compiler,
653 it disables recognition of C++ style @samp{//} comments as well as
654 the @code{inline} keyword. For the C++ compiler,
655 @samp{-foperator-names} is enabled as well.
658 The alternate keywords @code{__asm__}, @code{__extension__},
659 @code{__inline__} and @code{__typeof__} continue to work despite
660 @samp{-ansi}. You would not want to use them in an ANSI C program, of
661 course, but it is useful to put them in header files that might be included
662 in compilations done with @samp{-ansi}. Alternate predefined macros
663 such as @code{__unix__} and @code{__vax__} are also available, with or
664 without @samp{-ansi}.
666 The @samp{-ansi} option does not cause non-ANSI programs to be
667 rejected gratuitously. For that, @samp{-pedantic} is required in
668 addition to @samp{-ansi}. @xref{Warning Options}.
670 The macro @code{__STRICT_ANSI__} is predefined when the @samp{-ansi}
671 option is used. Some header files may notice this macro and refrain
672 from declaring certain functions or defining certain macros that the
673 ANSI standard doesn't call for; this is to avoid interfering with any
674 programs that might use these names for other things.
676 The functions @code{alloca}, @code{abort}, @code{exit}, and
677 @code{_exit} are not builtin functions when @samp{-ansi} is used.
680 Enable support for features found in the C9X standard. In particular,
681 enable support for the C9X @code{restrict} keyword.
683 Even when this option is not specified, you can still use some C9X
684 features in so far as they do not conflict with previous C standards.
685 For example, you may use @code{__restrict__} even when -flang-isoc9x
689 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
690 keyword, so that code can use these words as identifiers. You can use
691 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
692 instead. @samp{-ansi} implies @samp{-fno-asm}.
694 In C++, this switch only affects the @code{typeof} keyword, since
695 @code{asm} and @code{inline} are standard keywords. You may want to
696 use the @samp{-fno-gnu-keywords} flag instead, as it also disables the
697 other, C++-specific, extension keywords such as @code{headof}.
700 @cindex builtin functions
716 Don't recognize builtin functions that do not begin with @samp{__builtin_}
717 as prefix. Currently, the functions affected include @code{abort},
718 @code{abs}, @code{alloca}, @code{cos}, @code{exit}, @code{fabs},
719 @code{ffs}, @code{labs}, @code{memcmp}, @code{memcpy}, @code{sin},
720 @code{sqrt}, @code{strcmp}, @code{strcpy}, and @code{strlen}.
722 GCC normally generates special code to handle certain builtin functions
723 more efficiently; for instance, calls to @code{alloca} may become single
724 instructions that adjust the stack directly, and calls to @code{memcpy}
725 may become inline copy loops. The resulting code is often both smaller
726 and faster, but since the function calls no longer appear as such, you
727 cannot set a breakpoint on those calls, nor can you change the behavior
728 of the functions by linking with a different library.
730 The @samp{-ansi} option prevents @code{alloca} and @code{ffs} from being
731 builtin functions, since these functions do not have an ANSI standard
735 @cindex hosted environment
737 Assert that compilation takes place in a hosted environment. This implies
738 @samp{-fbuiltin}. A hosted environment is one in which the
739 entire standard library is available, and in which @code{main} has a return
740 type of @code{int}. Examples are nearly everything except a kernel.
741 This is equivalent to @samp{-fno-freestanding}.
744 @cindex hosted environment
746 Assert that compilation takes place in a freestanding environment. This
747 implies @samp{-fno-builtin}. A freestanding environment
748 is one in which the standard library may not exist, and program startup may
749 not necessarily be at @code{main}. The most obvious example is an OS kernel.
750 This is equivalent to @samp{-fno-hosted}.
753 Support ANSI C trigraphs. You don't want to know about this
754 brain-damage. The @samp{-ansi} option implies @samp{-trigraphs}.
756 @cindex traditional C language
757 @cindex C language, traditional
759 Attempt to support some aspects of traditional C compilers.
764 All @code{extern} declarations take effect globally even if they
765 are written inside of a function definition. This includes implicit
766 declarations of functions.
769 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
770 and @code{volatile} are not recognized. (You can still use the
771 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
775 Comparisons between pointers and integers are always allowed.
778 Integer types @code{unsigned short} and @code{unsigned char} promote
779 to @code{unsigned int}.
782 Out-of-range floating point literals are not an error.
785 Certain constructs which ANSI regards as a single invalid preprocessing
786 number, such as @samp{0xe-0xd}, are treated as expressions instead.
789 String ``constants'' are not necessarily constant; they are stored in
790 writable space, and identical looking constants are allocated
791 separately. (This is the same as the effect of
792 @samp{-fwritable-strings}.)
794 @cindex @code{longjmp} and automatic variables
796 All automatic variables not declared @code{register} are preserved by
797 @code{longjmp}. Ordinarily, GNU C follows ANSI C: automatic variables
798 not declared @code{volatile} may be clobbered.
803 @cindex escape sequences, traditional
804 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
805 literal characters @samp{x} and @samp{a} respectively. Without
806 @w{@samp{-traditional}}, @samp{\x} is a prefix for the hexadecimal
807 representation of a character, and @samp{\a} produces a bell.
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 The @samp{-traditional} option also enables @samp{-traditional-cpp},
820 which is described next.
822 @item -traditional-cpp
823 Attempt to support some aspects of traditional C preprocessors.
828 Comments convert to nothing at all, rather than to a space. This allows
829 traditional token concatenation.
832 In a preprocessing directive, the @samp{#} symbol must appear as the first
836 Macro arguments are recognized within string constants in a macro
837 definition (and their values are stringified, though without additional
838 quote marks, when they appear in such a context). The preprocessor
839 always considers a string constant to end at a newline.
842 @cindex detecting @w{@samp{-traditional}}
843 The predefined macro @code{__STDC__} is not defined when you use
844 @samp{-traditional}, but @code{__GNUC__} is (since the GNU extensions
845 which @code{__GNUC__} indicates are not affected by
846 @samp{-traditional}). If you need to write header files that work
847 differently depending on whether @samp{-traditional} is in use, by
848 testing both of these predefined macros you can distinguish four
849 situations: GNU C, traditional GNU C, other ANSI C compilers, and other
850 old C compilers. The predefined macro @code{__STDC_VERSION__} is also
851 not defined when you use @samp{-traditional}. @xref{Standard
852 Predefined,,Standard Predefined Macros,cpp.info,The C Preprocessor},
853 for more discussion of these and other predefined macros.
856 @cindex string constants vs newline
857 @cindex newline vs string constants
858 The preprocessor considers a string constant to end at a newline (unless
859 the newline is escaped with @samp{\}). (Without @w{@samp{-traditional}},
860 string constants can contain the newline character as typed.)
863 @item -fcond-mismatch
864 Allow conditional expressions with mismatched types in the second and
865 third arguments. The value of such an expression is void.
867 @item -funsigned-char
868 Let the type @code{char} be unsigned, like @code{unsigned char}.
870 Each kind of machine has a default for what @code{char} should
871 be. It is either like @code{unsigned char} by default or like
872 @code{signed char} by default.
874 Ideally, a portable program should always use @code{signed char} or
875 @code{unsigned char} when it depends on the signedness of an object.
876 But many programs have been written to use plain @code{char} and
877 expect it to be signed, or expect it to be unsigned, depending on the
878 machines they were written for. This option, and its inverse, let you
879 make such a program work with the opposite default.
881 The type @code{char} is always a distinct type from each of
882 @code{signed char} or @code{unsigned char}, even though its behavior
883 is always just like one of those two.
886 Let the type @code{char} be signed, like @code{signed char}.
888 Note that this is equivalent to @samp{-fno-unsigned-char}, which is
889 the negative form of @samp{-funsigned-char}. Likewise, the option
890 @samp{-fno-signed-char} is equivalent to @samp{-funsigned-char}.
892 You may wish to use @samp{-fno-builtin} as well as @samp{-traditional}
893 if your program uses names that are normally GNU C builtin functions for
894 other purposes of its own.
896 You cannot use @samp{-traditional} if you include any header files that
897 rely on ANSI C features. Some vendors are starting to ship systems with
898 ANSI C header files and you cannot use @samp{-traditional} on such
899 systems to compile files that include any system headers.
901 @item -fsigned-bitfields
902 @itemx -funsigned-bitfields
903 @itemx -fno-signed-bitfields
904 @itemx -fno-unsigned-bitfields
905 These options control whether a bitfield is signed or unsigned, when the
906 declaration does not use either @code{signed} or @code{unsigned}. By
907 default, such a bitfield is signed, because this is consistent: the
908 basic integer types such as @code{int} are signed types.
910 However, when @samp{-traditional} is used, bitfields are all unsigned
913 @item -fwritable-strings
914 Store string constants in the writable data segment and don't uniquize
915 them. This is for compatibility with old programs which assume they can
916 write into string constants. The option @samp{-traditional} also has
919 Writing into string constants is a very bad idea; ``constants'' should
922 @item -fallow-single-precision
923 Do not promote single precision math operations to double precision,
924 even when compiling with @samp{-traditional}.
926 Traditional K&R C promotes all floating point operations to double
927 precision, regardless of the sizes of the operands. On the
928 architecture for which you are compiling, single precision may be faster
929 than double precision. If you must use @samp{-traditional}, but want
930 to use single precision operations when the operands are single
931 precision, use this option. This option has no effect when compiling
932 with ANSI or GNU C conventions (the default).
936 @node C++ Dialect Options
937 @section Options Controlling C++ Dialect
939 @cindex compiler options, C++
940 @cindex C++ options, command line
942 This section describes the command-line options that are only meaningful
943 for C++ programs; but you can also use most of the GNU compiler options
944 regardless of what language your program is in. For example, you
945 might compile a file @code{firstClass.C} like this:
948 g++ -g -frepo -O -c firstClass.C
952 In this example, only @samp{-frepo} is an option meant
953 only for C++ programs; you can use the other options with any
954 language supported by GCC.
956 Here is a list of options that are @emph{only} for compiling C++ programs:
959 @item -fno-access-control
960 Turn off all access checking. This switch is mainly useful for working
961 around bugs in the access control code.
964 Check that the pointer returned by @code{operator new} is non-null
965 before attempting to modify the storage allocated. The current Working
966 Paper requires that @code{operator new} never return a null pointer, so
967 this check is normally unnecessary.
969 An alternative to using this option is to specify that your
970 @code{operator new} does not throw any exceptions; if you declare it
971 @samp{throw()}, g++ will check the return value. See also @samp{new
974 @item -fconserve-space
975 Put uninitialized or runtime-initialized global variables into the
976 common segment, as C does. This saves space in the executable at the
977 cost of not diagnosing duplicate definitions. If you compile with this
978 flag and your program mysteriously crashes after @code{main()} has
979 completed, you may have an object that is being destroyed twice because
980 two definitions were merged.
982 This option is no longer useful on most targets, now that support has
983 been added for putting variables into BSS without making them common.
985 @item -fdollars-in-identifiers
986 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
987 @samp{$} with the option @samp{-fno-dollars-in-identifiers}. (GNU C allows
988 @samp{$} by default on most target systems, but there are a few exceptions.)
989 Traditional C allowed the character @samp{$} to form part of
990 identifiers. However, ANSI C and C++ forbid @samp{$} in identifiers.
992 @item -fno-elide-constructors
993 The C++ standard allows an implementation to omit creating a temporary
994 which is only used to initialize another object of the same type.
995 Specifying this option disables that optimization, and forces g++ to
996 call the copy constructor in all cases.
998 @item -fexternal-templates
999 Cause template instantiations to obey @samp{#pragma interface} and
1000 @samp{implementation}; template instances are emitted or not according
1001 to the location of the template definition. @xref{Template
1002 Instantiation}, for more information.
1004 This option is deprecated.
1006 @item -falt-external-templates
1007 Similar to -fexternal-templates, but template instances are emitted or
1008 not according to the place where they are first instantiated.
1009 @xref{Template Instantiation}, for more information.
1011 This option is deprecated.
1014 @itemx -fno-for-scope
1015 If -ffor-scope is specified, the scope of variables declared in
1016 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1017 as specified by the draft C++ standard.
1018 If -fno-for-scope is specified, the scope of variables declared in
1019 a @i{for-init-statement} extends to the end of the enclosing scope,
1020 as was the case in old versions of gcc, and other (traditional)
1021 implementations of C++.
1023 The default if neither flag is given to follow the standard,
1024 but to allow and give a warning for old-style code that would
1025 otherwise be invalid, or have different behavior.
1027 @item -fno-gnu-keywords
1028 Do not recognize @code{classof}, @code{headof}, or @code{typeof} as a
1029 keyword, so that code can use these words as identifiers. You can use
1030 the keywords @code{__classof__}, @code{__headof__}, and
1031 @code{__typeof__} instead. @samp{-ansi} implies
1032 @samp{-fno-gnu-keywords}.
1034 @item -fguiding-decls
1035 Treat a function declaration with the same type as a potential function
1036 template instantiation as though it declares that instantiation, not a
1037 normal function. If a definition is given for the function later in the
1038 translation unit (or another translation unit if the target supports
1039 weak symbols), that definition will be used; otherwise the template will
1040 be instantiated. This behavior reflects the C++ language prior to
1041 September 1996, when guiding declarations were removed.
1043 This option implies @samp{-fname-mangling-version-0}, and will not work
1044 with other name mangling versions. Like all options that change the
1045 ABI, all C++ code, @emph{including libgcc.a} must be built with the same
1046 setting of this option.
1049 Treat the @code{namespace std} as a namespace, instead of ignoring
1050 it. For compatibility with earlier versions of g++, the compiler will,
1051 by default, ignore @code{namespace-declarations},
1052 @code{using-declarations}, @code{using-directives}, and
1053 @code{namespace-names}, if they involve @code{std}.
1055 @item -fhuge-objects
1056 Support virtual function calls for objects that exceed the size
1057 representable by a @samp{short int}. Users should not use this flag by
1058 default; if you need to use it, the compiler will tell you so.
1060 This flag is not useful when compiling with -fvtable-thunks.
1062 Like all options that change the ABI, all C++ code, @emph{including
1063 libgcc} must be built with the same setting of this option.
1065 @item -fno-implicit-templates
1066 Never emit code for non-inline templates which are instantiated
1067 implicitly (i.e. by use); only emit code for explicit instantiations.
1068 @xref{Template Instantiation}, for more information.
1070 @item -fno-implicit-inline-templates
1071 Don't emit code for implicit instantiations of inline templates, either.
1072 The default is to handle inlines differently so that compiles with and
1073 without optimization will need the same set of explicit instantiations.
1075 @item -finit-priority
1076 Support @samp{__attribute__ ((init_priority (n)))} for controlling the
1077 order of initialization of file-scope objects. On ELF targets, this
1078 requires GNU ld 2.10 or later.
1080 @item -fno-implement-inlines
1081 To save space, do not emit out-of-line copies of inline functions
1082 controlled by @samp{#pragma implementation}. This will cause linker
1083 errors if these functions are not inlined everywhere they are called.
1085 @item -fms-extensions
1086 Disable pedwarns about constructs used in MFC, such as implicit int and
1087 getting a pointer to member function via non-standard syntax.
1089 @item -fname-mangling-version-@var{n}
1090 Control the way in which names are mangled. Version 0 is compatible
1091 with versions of g++ before 2.8. Version 1 is the default. Version 1
1092 will allow correct mangling of function templates. For example,
1093 version 0 mangling does not mangle foo<int, double> and foo<int, char>
1094 given this declaration:
1097 template <class T, class U> void foo(T t);
1100 Like all options that change the ABI, all C++ code, @emph{including
1101 libgcc} must be built with the same setting of this option.
1103 @item -foperator-names
1104 Recognize the operator name keywords @code{and}, @code{bitand},
1105 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1106 synonyms for the symbols they refer to. @samp{-ansi} implies
1107 @samp{-foperator-names}.
1109 @item -fno-optional-diags
1110 Disable diagnostics that the standard says a compiler does not need to
1111 issue. Currently, the only such diagnostic issued by g++ is the one for
1112 a name having multiple meanings within a class.
1115 Downgrade messages about nonconformant code from errors to warnings. By
1116 default, g++ effectively sets @samp{-pedantic-errors} without
1117 @samp{-pedantic}; this option reverses that. This behavior and this
1118 option are superseded by @samp{-pedantic}, which works as it does for GNU C.
1121 Enable automatic template instantiation. This option also implies
1122 @samp{-fno-implicit-templates}. @xref{Template Instantiation}, for more
1126 Disable generation of the information used by C++ runtime type
1127 identification features (@samp{dynamic_cast} and @samp{typeid}). If you
1128 don't use those parts of the language (or exception handling, which uses
1129 @samp{dynamic_cast} internally), you can save some space by using this
1132 @item -fstrict-prototype
1133 Within an @samp{extern "C"} linkage specification, treat a function
1134 declaration with no arguments, such as @samp{int foo ();}, as declaring
1135 the function to take no arguments. Normally, such a declaration means
1136 that the function @code{foo} can take any combination of arguments, as
1137 in C. @samp{-pedantic} implies @samp{-fstrict-prototype} unless
1138 overridden with @samp{-fno-strict-prototype}.
1140 Specifying this option will also suppress implicit declarations of
1143 This flag no longer affects declarations with C++ linkage.
1146 @itemx -fno-squangle
1147 @samp{-fsquangle} will enable a compressed form of name mangling for
1148 identifiers. In particular, it helps to shorten very long names by recognizing
1149 types and class names which occur more than once, replacing them with special
1150 short ID codes. This option also requires any C++ libraries being used to
1151 be compiled with this option as well. The compiler has this disabled (the
1152 equivalent of @samp{-fno-squangle}) by default.
1154 Like all options that change the ABI, all C++ code, @emph{including
1155 libgcc.a} must be built with the same setting of this option.
1157 @item -ftemplate-depth-@var{n}
1158 Set the maximum instantiation depth for template classes to @var{n}.
1159 A limit on the template instantiation depth is needed to detect
1160 endless recursions during template class instantiation. ANSI/ISO C++
1161 conforming programs must not rely on a maximum depth greater than 17.
1163 @item -fthis-is-variable
1164 Permit assignment to @code{this}. The incorporation of user-defined
1165 free store management into C++ has made assignment to @samp{this} an
1166 anachronism. Therefore, by default it is invalid to assign to
1167 @code{this} within a class member function; that is, GNU C++ treats
1168 @samp{this} in a member function of class @code{X} as a non-lvalue of
1169 type @samp{X *}. However, for backwards compatibility, you can make it
1170 valid with @samp{-fthis-is-variable}.
1172 @item -fvtable-thunks
1173 Use @samp{thunks} to implement the virtual function dispatch table
1174 (@samp{vtable}). The traditional (cfront-style) approach to
1175 implementing vtables was to store a pointer to the function and two
1176 offsets for adjusting the @samp{this} pointer at the call site. Newer
1177 implementations store a single pointer to a @samp{thunk} function which
1178 does any necessary adjustment and then calls the target function.
1180 This option also enables a heuristic for controlling emission of
1181 vtables; if a class has any non-inline virtual functions, the vtable
1182 will be emitted in the translation unit containing the first one of
1185 Like all options that change the ABI, all C++ code, @emph{including
1186 libgcc.a} must be built with the same setting of this option.
1189 Do not search for header files in the standard directories specific to
1190 C++, but do still search the other standard directories. (This option
1191 is used when building the C++ library.)
1194 In addition, these optimization, warning, and code generation options
1195 have meanings only for C++ programs:
1198 @item -fno-default-inline
1199 Do not assume @samp{inline} for functions defined inside a class scope.
1200 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1201 functions will have linkage like inline functions; they just won't be
1204 @item -Wctor-dtor-privacy (C++ only)
1205 Warn when a class seems unusable, because all the constructors or
1206 destructors in a class are private and the class has no friends or
1207 public static member functions.
1209 @item -Wnon-virtual-dtor (C++ only)
1210 Warn when a class declares a non-virtual destructor that should probably
1211 be virtual, because it looks like the class will be used polymorphically.
1213 @item -Wreorder (C++ only)
1214 @cindex reordering, warning
1215 @cindex warning for reordering of member initializers
1216 Warn when the order of member initializers given in the code does not
1217 match the order in which they must be executed. For instance:
1223 A(): j (0), i (1) @{ @}
1227 Here the compiler will warn that the member initializers for @samp{i}
1228 and @samp{j} will be rearranged to match the declaration order of the
1232 The following @samp{-W@dots{}} options are not affected by @samp{-Wall}.
1235 @item -Weffc++ (C++ only)
1236 Warn about violations of various style guidelines from Scott Meyers'
1237 @cite{Effective C++} books. If you use this option, you should be aware
1238 that the standard library headers do not obey all of these guidelines;
1239 you can use @samp{grep -v} to filter out those warnings.
1241 @item -Wno-deprecated (C++ only)
1242 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1244 @item -Wno-non-template-friend (C++ only)
1245 Disable warnings when non-templatized friend functions are declared
1246 within a template. With the advent of explicit template specification
1247 support in g++, if the name of the friend is an unqualified-id (ie,
1248 @samp{friend foo(int)}), the C++ language specification demands that the
1249 friend declare or define an ordinary, nontemplate function. (Section
1250 14.5.3). Before g++ implemented explicit specification, unqualified-ids
1251 could be interpreted as a particular specialization of a templatized
1252 function. Because this non-conforming behavior is no longer the default
1253 behavior for g++, @samp{-Wnon-template-friend} allows the compiler to
1254 check existing code for potential trouble spots, and is on by default.
1255 This new compiler behavior can also be turned off with the flag
1256 @samp{-fguiding-decls}, which activates the older, non-specification
1257 compiler code, or with @samp{-Wno-non-template-friend} which keeps the
1258 conformant compiler code but disables the helpful warning.
1260 @item -Wold-style-cast (C++ only)
1261 Warn if an old-style (C-style) cast is used within a C++ program. The
1262 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1263 @samp{const_cast}) are less vulnerable to unintended effects.
1265 @item -Woverloaded-virtual (C++ only)
1266 @cindex overloaded virtual fn, warning
1267 @cindex warning for overloaded virtual fn
1268 Warn when a derived class function declaration may be an error in
1269 defining a virtual function. In a derived class, the
1270 definitions of virtual functions must match the type signature of a
1271 virtual function declared in the base class. With this option, the
1272 compiler warns when you define a function with the same name as a
1273 virtual function, but with a type signature that does not match any
1274 declarations from the base class.
1276 @item -Wno-pmf-conversions (C++ only)
1277 Disable the diagnostic for converting a bound pointer to member function
1280 @item -Wsign-promo (C++ only)
1281 Warn when overload resolution chooses a promotion from unsigned or
1282 enumeral type to a signed type over a conversion to an unsigned type of
1283 the same size. Previous versions of g++ would try to preserve
1284 unsignedness, but the standard mandates the current behavior.
1286 @item -Wsynth (C++ only)
1287 @cindex warning for synthesized methods
1288 @cindex synthesized methods, warning
1289 Warn when g++'s synthesis behavior does not match that of cfront. For
1295 A& operator = (int);
1305 In this example, g++ will synthesize a default @samp{A& operator =
1306 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1309 @node Warning Options
1310 @section Options to Request or Suppress Warnings
1311 @cindex options to control warnings
1312 @cindex warning messages
1313 @cindex messages, warning
1314 @cindex suppressing warnings
1316 Warnings are diagnostic messages that report constructions which
1317 are not inherently erroneous but which are risky or suggest there
1318 may have been an error.
1320 You can request many specific warnings with options beginning @samp{-W},
1321 for example @samp{-Wimplicit} to request warnings on implicit
1322 declarations. Each of these specific warning options also has a
1323 negative form beginning @samp{-Wno-} to turn off warnings;
1324 for example, @samp{-Wno-implicit}. This manual lists only one of the
1325 two forms, whichever is not the default.
1327 These options control the amount and kinds of warnings produced by GCC:
1330 @cindex syntax checking
1332 Check the code for syntax errors, but don't do anything beyond that.
1335 Issue all the warnings demanded by strict ANSI C and ISO C++;
1336 reject all programs that use forbidden extensions.
1338 Valid ANSI C and ISO C++ programs should compile properly with or without
1339 this option (though a rare few will require @samp{-ansi}). However,
1340 without this option, certain GNU extensions and traditional C and C++
1341 features are supported as well. With this option, they are rejected.
1343 @samp{-pedantic} does not cause warning messages for use of the
1344 alternate keywords whose names begin and end with @samp{__}. Pedantic
1345 warnings are also disabled in the expression that follows
1346 @code{__extension__}. However, only system header files should use
1347 these escape routes; application programs should avoid them.
1348 @xref{Alternate Keywords}.
1350 This option is not intended to be @i{useful}; it exists only to satisfy
1351 pedants who would otherwise claim that GCC fails to support the ANSI
1354 Some users try to use @samp{-pedantic} to check programs for strict ANSI
1355 C conformance. They soon find that it does not do quite what they want:
1356 it finds some non-ANSI practices, but not all---only those for which
1357 ANSI C @emph{requires} a diagnostic.
1359 A feature to report any failure to conform to ANSI C might be useful in
1360 some instances, but would require considerable additional work and would
1361 be quite different from @samp{-pedantic}. We don't have plans to
1362 support such a feature in the near future.
1364 @item -pedantic-errors
1365 Like @samp{-pedantic}, except that errors are produced rather than
1369 Inhibit all warning messages.
1372 Inhibit warning messages about the use of @samp{#import}.
1374 @item -Wchar-subscripts
1375 Warn if an array subscript has type @code{char}. This is a common cause
1376 of error, as programmers often forget that this type is signed on some
1380 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1381 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1384 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1385 the arguments supplied have types appropriate to the format string
1388 @item -Wimplicit-int
1389 Warn when a declaration does not specify a type.
1391 @item -Wimplicit-function-declaration
1392 @itemx -Werror-implicit-function-declaration
1393 Give a warning (or error) whenever a function is used before being
1397 Same as @samp{-Wimplicit-int} and @samp{-Wimplicit-function-}@*
1401 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1402 function with external linkage, returning int, taking either zero
1403 arguments, two, or three arguments of appropriate types.
1406 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1407 indicate a typo in the user's code, as they have implementation-defined
1408 values, and should not be used in portable code.
1411 Warn if parentheses are omitted in certain contexts, such
1412 as when there is an assignment in a context where a truth value
1413 is expected, or when operators are nested whose precedence people
1414 often get confused about.
1416 Also warn about constructions where there may be confusion to which
1417 @code{if} statement an @code{else} branch belongs. Here is an example of
1430 In C, every @code{else} branch belongs to the innermost possible @code{if}
1431 statement, which in this example is @code{if (b)}. This is often not
1432 what the programmer expected, as illustrated in the above example by
1433 indentation the programmer chose. When there is the potential for this
1434 confusion, GNU C will issue a warning when this flag is specified.
1435 To eliminate the warning, add explicit braces around the innermost
1436 @code{if} statement so there is no way the @code{else} could belong to
1437 the enclosing @code{if}. The resulting code would look like this:
1452 Warn whenever a function is defined with a return-type that defaults
1453 to @code{int}. Also warn about any @code{return} statement with no
1454 return-value in a function whose return-type is not @code{void}.
1457 Warn whenever a @code{switch} statement has an index of enumeral type
1458 and lacks a @code{case} for one or more of the named codes of that
1459 enumeration. (The presence of a @code{default} label prevents this
1460 warning.) @code{case} labels outside the enumeration range also
1461 provoke warnings when this option is used.
1464 Warn if any trigraphs are encountered (assuming they are enabled).
1467 Warn whenever a variable is unused aside from its declaration,
1468 whenever a function is declared static but never defined, whenever a
1469 label is declared but not used, and whenever a statement computes a
1470 result that is explicitly not used.
1472 In order to get a warning about an unused function parameter, you must
1473 specify both @samp{-W} and @samp{-Wunused}.
1475 To suppress this warning for an expression, simply cast it to void. For
1476 unused variables, parameters and labels, use the @samp{unused} attribute
1477 (@pxref{Variable Attributes}).
1479 @item -Wuninitialized
1480 Warn if an automatic variable is used without first being initialized or
1481 if a variable may be clobbered by a @code{setjmp} call.
1483 These warnings are possible only in optimizing compilation,
1484 because they require data flow information that is computed only
1485 when optimizing. If you don't specify @samp{-O}, you simply won't
1488 These warnings occur only for variables that are candidates for
1489 register allocation. Therefore, they do not occur for a variable that
1490 is declared @code{volatile}, or whose address is taken, or whose size
1491 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
1492 structures, unions or arrays, even when they are in registers.
1494 Note that there may be no warning about a variable that is used only
1495 to compute a value that itself is never used, because such
1496 computations may be deleted by data flow analysis before the warnings
1499 These warnings are made optional because GCC is not smart
1500 enough to see all the reasons why the code might be correct
1501 despite appearing to have an error. Here is one example of how
1520 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
1521 always initialized, but GCC doesn't know this. Here is
1522 another common case:
1527 if (change_y) save_y = y, y = new_y;
1529 if (change_y) y = save_y;
1534 This has no bug because @code{save_y} is used only if it is set.
1536 @cindex @code{longjmp} warnings
1537 This option also warns when a nonvolatile automatic variable might be
1538 changed by a call to @code{longjmp}. These warnings as well are possible
1539 only in optimizing compilation.
1541 The compiler sees only the calls to @code{setjmp}. It cannot know
1542 where @code{longjmp} will be called; in fact, a signal handler could
1543 call it at any point in the code. As a result, you may get a warning
1544 even when there is in fact no problem because @code{longjmp} cannot
1545 in fact be called at the place which would cause a problem.
1547 Some spurious warnings can be avoided if you declare all the functions
1548 you use that never return as @code{noreturn}. @xref{Function
1551 @item -Wreorder (C++ only)
1552 @cindex reordering, warning
1553 @cindex warning for reordering of member initializers
1554 Warn when the order of member initializers given in the code does not
1555 match the order in which they must be executed. For instance:
1557 @item -Wunknown-pragmas
1558 @cindex warning for unknown pragmas
1559 @cindex unknown pragmas, warning
1560 @cindex pragmas, warning of unknown
1561 Warn when a #pragma directive is encountered which is not understood by
1562 GCC. If this command line option is used, warnings will even be issued
1563 for unknown pragmas in system header files. This is not the case if
1564 the warnings were only enabled by the @samp{-Wall} command line option.
1567 All of the above @samp{-W} options combined. This enables all the
1568 warnings about constructions that some users consider questionable, and
1569 that are easy to avoid (or modify to prevent the warning), even in
1570 conjunction with macros.
1573 The following @samp{-W@dots{}} options are not implied by @samp{-Wall}.
1574 Some of them warn about constructions that users generally do not
1575 consider questionable, but which occasionally you might wish to check
1576 for; others warn about constructions that are necessary or hard to avoid
1577 in some cases, and there is no simple way to modify the code to suppress
1582 Print extra warning messages for these events:
1586 A function can return either with or without a value. (Falling
1587 off the end of the function body is considered returning without
1588 a value.) For example, this function would evoke such a
1602 An expression-statement or the left-hand side of a comma expression
1603 contains no side effects.
1604 To suppress the warning, cast the unused expression to void.
1605 For example, an expression such as @samp{x[i,j]} will cause a warning,
1606 but @samp{x[(void)i,j]} will not.
1609 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
1612 A comparison like @samp{x<=y<=z} appears; this is equivalent to
1613 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
1614 that of ordinary mathematical notation.
1617 Storage-class specifiers like @code{static} are not the first things in
1618 a declaration. According to the C Standard, this usage is obsolescent.
1621 If @samp{-Wall} or @samp{-Wunused} is also specified, warn about unused
1625 A comparison between signed and unsigned values could produce an
1626 incorrect result when the signed value is converted to unsigned.
1627 (But don't warn if @samp{-Wno-sign-compare} is also specified.)
1630 An aggregate has a partly bracketed initializer.
1631 For example, the following code would evoke such a warning,
1632 because braces are missing around the initializer for @code{x.h}:
1635 struct s @{ int f, g; @};
1636 struct t @{ struct s h; int i; @};
1637 struct t x = @{ 1, 2, 3 @};
1641 An aggregate has an initializer which does not initialize all members.
1642 For example, the following code would cause such a warning, because
1643 @code{x.h} would be implicitly initialized to zero:
1646 struct s @{ int f, g, h; @};
1647 struct s x = @{ 3, 4 @};
1652 Warn about certain constructs that behave differently in traditional and
1657 Macro arguments occurring within string constants in the macro body.
1658 These would substitute the argument in traditional C, but are part of
1659 the constant in ANSI C.
1662 A function declared external in one block and then used after the end of
1666 A @code{switch} statement has an operand of type @code{long}.
1669 A non-@code{static} function declaration follows a @code{static} one.
1670 This construct is not accepted by some traditional C compilers.
1674 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
1677 Warn whenever a local variable shadows another local variable.
1679 @item -Wid-clash-@var{len}
1680 Warn whenever two distinct identifiers match in the first @var{len}
1681 characters. This may help you prepare a program that will compile
1682 with certain obsolete, brain-damaged compilers.
1684 @item -Wlarger-than-@var{len}
1685 Warn whenever an object of larger than @var{len} bytes is defined.
1687 @item -Wpointer-arith
1688 Warn about anything that depends on the ``size of'' a function type or
1689 of @code{void}. GNU C assigns these types a size of 1, for
1690 convenience in calculations with @code{void *} pointers and pointers
1693 @item -Wbad-function-cast
1694 Warn whenever a function call is cast to a non-matching type.
1695 For example, warn if @code{int malloc()} is cast to @code{anything *}.
1698 Warn whenever a pointer is cast so as to remove a type qualifier from
1699 the target type. For example, warn if a @code{const char *} is cast
1700 to an ordinary @code{char *}.
1703 Warn whenever a pointer is cast such that the required alignment of the
1704 target is increased. For example, warn if a @code{char *} is cast to
1705 an @code{int *} on machines where integers can only be accessed at
1706 two- or four-byte boundaries.
1708 @item -Wwrite-strings
1709 Give string constants the type @code{const char[@var{length}]} so that
1710 copying the address of one into a non-@code{const} @code{char *}
1711 pointer will get a warning. These warnings will help you find at
1712 compile time code that can try to write into a string constant, but
1713 only if you have been very careful about using @code{const} in
1714 declarations and prototypes. Otherwise, it will just be a nuisance;
1715 this is why we did not make @samp{-Wall} request these warnings.
1718 Warn if a prototype causes a type conversion that is different from what
1719 would happen to the same argument in the absence of a prototype. This
1720 includes conversions of fixed point to floating and vice versa, and
1721 conversions changing the width or signedness of a fixed point argument
1722 except when the same as the default promotion.
1724 Also, warn if a negative integer constant expression is implicitly
1725 converted to an unsigned type. For example, warn about the assignment
1726 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
1727 casts like @code{(unsigned) -1}.
1729 @item -Wsign-compare
1730 @cindex warning for comparison of signed and unsigned values
1731 @cindex comparison of signed and unsigned values, warning
1732 @cindex signed and unsigned values, comparison warning
1733 Warn when a comparison between signed and unsigned values could produce
1734 an incorrect result when the signed value is converted to unsigned.
1735 This warning is also enabled by @samp{-W}; to get the other warnings
1736 of @samp{-W} without this warning, use @samp{-W -Wno-sign-compare}.
1738 @item -Waggregate-return
1739 Warn if any functions that return structures or unions are defined or
1740 called. (In languages where you can return an array, this also elicits
1743 @item -Wstrict-prototypes
1744 Warn if a function is declared or defined without specifying the
1745 argument types. (An old-style function definition is permitted without
1746 a warning if preceded by a declaration which specifies the argument
1749 @item -Wmissing-prototypes
1750 Warn if a global function is defined without a previous prototype
1751 declaration. This warning is issued even if the definition itself
1752 provides a prototype. The aim is to detect global functions that fail
1753 to be declared in header files.
1755 @item -Wmissing-declarations
1756 Warn if a global function is defined without a previous declaration.
1757 Do so even if the definition itself provides a prototype.
1758 Use this option to detect global functions that are not declared in
1761 @item -Wmissing-noreturn
1762 Warn about functions which might be candidates for attribute @code{noreturn}.
1763 Note these are only possible candidates, not absolute ones. Care should
1764 be taken to manually verify functions actually do not ever return before
1765 adding the @code{noreturn} attribute, otherwise subtle code generation
1766 bugs could be introduced.
1768 @item -Wredundant-decls
1769 Warn if anything is declared more than once in the same scope, even in
1770 cases where multiple declaration is valid and changes nothing.
1772 @item -Wnested-externs
1773 Warn if an @code{extern} declaration is encountered within a function.
1775 @item -Wunreachable-code
1776 Warn if the compiler detects that code will never be executed.
1778 This option is intended to warn when the compiler detects that at
1779 least a whole line of source code will never be executed, because
1780 some condition is never satisfied or because it is after a
1781 procedure that never returns.
1783 It is possible for this option to produce a warning even though there
1784 are circumstances under which part of the affected line can be executed,
1785 so care should be taken when removing apparently-unreachable code.
1787 For instance, when a function is inlined, a warning may mean that the
1788 line is unreachable in only one inlined copy of the function.
1790 This option is not made part of @samp{-Wall} because in a debugging
1791 version of a program there is often substantial code which checks
1792 correct functioning of the program and is, hopefully, unreachable
1793 because the program does work. Another common use of unreachable
1794 code is to provide behaviour which is selectable at compile-time.
1797 Warn if a function can not be inlined and it was declared as inline.
1800 Warn if @samp{long long} type is used. This is default. To inhibit
1801 the warning messages, use @samp{-Wno-long-long}. Flags
1802 @samp{-Wlong-long} and @samp{-Wno-long-long} are taken into account
1803 only when @samp{-pedantic} flag is used.
1806 Make all warnings into errors.
1809 @node Debugging Options
1810 @section Options for Debugging Your Program or GCC
1811 @cindex options, debugging
1812 @cindex debugging information options
1814 GCC has various special options that are used for debugging
1815 either your program or GCC:
1819 Produce debugging information in the operating system's native format
1820 (stabs, COFF, XCOFF, or DWARF). GDB can work with this debugging
1823 On most systems that use stabs format, @samp{-g} enables use of extra
1824 debugging information that only GDB can use; this extra information
1825 makes debugging work better in GDB but will probably make other debuggers
1827 refuse to read the program. If you want to control for certain whether
1828 to generate the extra information, use @samp{-gstabs+}, @samp{-gstabs},
1829 @samp{-gxcoff+}, @samp{-gxcoff}, @samp{-gdwarf-1+}, or @samp{-gdwarf-1}
1832 Unlike most other C compilers, GCC allows you to use @samp{-g} with
1833 @samp{-O}. The shortcuts taken by optimized code may occasionally
1834 produce surprising results: some variables you declared may not exist
1835 at all; flow of control may briefly move where you did not expect it;
1836 some statements may not be executed because they compute constant
1837 results or their values were already at hand; some statements may
1838 execute in different places because they were moved out of loops.
1840 Nevertheless it proves possible to debug optimized output. This makes
1841 it reasonable to use the optimizer for programs that might have bugs.
1843 The following options are useful when GCC is generated with the
1844 capability for more than one debugging format.
1847 Produce debugging information for use by GDB. This means to use the
1848 most expressive format available (DWARF 2, stabs, or the native format
1849 if neither of those are supported), including GDB extensions if at all
1853 Produce debugging information in stabs format (if that is supported),
1854 without GDB extensions. This is the format used by DBX on most BSD
1855 systems. On MIPS, Alpha and System V Release 4 systems this option
1856 produces stabs debugging output which is not understood by DBX or SDB.
1857 On System V Release 4 systems this option requires the GNU assembler.
1860 Produce debugging information in stabs format (if that is supported),
1861 using GNU extensions understood only by the GNU debugger (GDB). The
1862 use of these extensions is likely to make other debuggers crash or
1863 refuse to read the program.
1866 Produce debugging information in COFF format (if that is supported).
1867 This is the format used by SDB on most System V systems prior to
1871 Produce debugging information in XCOFF format (if that is supported).
1872 This is the format used by the DBX debugger on IBM RS/6000 systems.
1875 Produce debugging information in XCOFF format (if that is supported),
1876 using GNU extensions understood only by the GNU debugger (GDB). The
1877 use of these extensions is likely to make other debuggers crash or
1878 refuse to read the program, and may cause assemblers other than the GNU
1879 assembler (GAS) to fail with an error.
1882 Produce debugging information in DWARF version 1 format (if that is
1883 supported). This is the format used by SDB on most System V Release 4
1887 Produce debugging information in DWARF version 1 format (if that is
1888 supported), using GNU extensions understood only by the GNU debugger
1889 (GDB). The use of these extensions is likely to make other debuggers
1890 crash or refuse to read the program.
1893 Produce debugging information in DWARF version 2 format (if that is
1894 supported). This is the format used by DBX on IRIX 6.
1897 @itemx -ggdb@var{level}
1898 @itemx -gstabs@var{level}
1899 @itemx -gcoff@var{level}
1900 @itemx -gxcoff@var{level}
1901 @itemx -gdwarf@var{level}
1902 @itemx -gdwarf-2@var{level}
1903 Request debugging information and also use @var{level} to specify how
1904 much information. The default level is 2.
1906 Level 1 produces minimal information, enough for making backtraces in
1907 parts of the program that you don't plan to debug. This includes
1908 descriptions of functions and external variables, but no information
1909 about local variables and no line numbers.
1911 Level 3 includes extra information, such as all the macro definitions
1912 present in the program. Some debuggers support macro expansion when
1917 Generate extra code to write profile information suitable for the
1918 analysis program @code{prof}. You must use this option when compiling
1919 the source files you want data about, and you must also use it when
1922 @cindex @code{gprof}
1924 Generate extra code to write profile information suitable for the
1925 analysis program @code{gprof}. You must use this option when compiling
1926 the source files you want data about, and you must also use it when
1931 Generate extra code to write profile information for basic blocks, which will
1932 record the number of times each basic block is executed, the basic block start
1933 address, and the function name containing the basic block. If @samp{-g} is
1934 used, the line number and filename of the start of the basic block will also be
1935 recorded. If not overridden by the machine description, the default action is
1936 to append to the text file @file{bb.out}.
1938 This data could be analyzed by a program like @code{tcov}. Note,
1939 however, that the format of the data is not what @code{tcov} expects.
1940 Eventually GNU @code{gprof} should be extended to process this data.
1943 Makes the compiler print out each function name as it is compiled, and
1944 print some statistics about each pass when it finishes.
1947 Generate extra code to profile basic blocks. Your executable will
1948 produce output that is a superset of that produced when @samp{-a} is
1949 used. Additional output is the source and target address of the basic
1950 blocks where a jump takes place, the number of times a jump is executed,
1951 and (optionally) the complete sequence of basic blocks being executed.
1952 The output is appended to file @file{bb.out}.
1954 You can examine different profiling aspects without recompilation. Your
1955 executable will read a list of function names from file @file{bb.in}.
1956 Profiling starts when a function on the list is entered and stops when
1957 that invocation is exited. To exclude a function from profiling, prefix
1958 its name with `-'. If a function name is not unique, you can
1959 disambiguate it by writing it in the form
1960 @samp{/path/filename.d:functionname}. Your executable will write the
1961 available paths and filenames in file @file{bb.out}.
1963 Several function names have a special meaning:
1966 Write source, target and frequency of jumps to file @file{bb.out}.
1967 @item __bb_hidecall__
1968 Exclude function calls from frequency count.
1969 @item __bb_showret__
1970 Include function returns in frequency count.
1972 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
1973 The file will be compressed using the program @samp{gzip}, which must
1974 exist in your @code{PATH}. On systems without the @samp{popen}
1975 function, the file will be named @file{bbtrace} and will not be
1976 compressed. @strong{Profiling for even a few seconds on these systems
1977 will produce a very large file.} Note: @code{__bb_hidecall__} and
1978 @code{__bb_showret__} will not affect the sequence written to
1982 Here's a short example using different profiling parameters
1983 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
1984 1 and 2 and is called twice from block 3 of function @code{main}. After
1985 the calls, block 3 transfers control to block 4 of @code{main}.
1987 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
1988 the following sequence of blocks is written to file @file{bbtrace.gz}:
1989 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
1990 the return is to a point inside the block and not to the top. The
1991 block address 0 always indicates, that control is transferred
1992 to the trace from somewhere outside the observed functions. With
1993 @samp{-foo} added to @file{bb.in}, the blocks of function
1994 @code{foo} are removed from the trace, so only 0 3 4 remains.
1996 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
1997 jump frequencies will be written to file @file{bb.out}. The
1998 frequencies are obtained by constructing a trace of blocks
1999 and incrementing a counter for every neighbouring pair of blocks
2000 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2004 Jump from block 0x0 to block 0x3 executed 1 time(s)
2005 Jump from block 0x3 to block 0x1 executed 1 time(s)
2006 Jump from block 0x1 to block 0x2 executed 2 time(s)
2007 Jump from block 0x2 to block 0x1 executed 1 time(s)
2008 Jump from block 0x2 to block 0x4 executed 1 time(s)
2011 With @code{__bb_hidecall__}, control transfer due to call instructions
2012 is removed from the trace, that is the trace is cut into three parts: 0
2013 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2014 to return instructions is added to the trace. The trace becomes: 0 3 1
2015 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2016 written to @file{bbtrace.gz}. It is solely used for counting jump
2019 @item -fprofile-arcs
2020 Instrument @dfn{arcs} during compilation. For each function of your
2021 program, GCC creates a program flow graph, then finds a spanning tree
2022 for the graph. Only arcs that are not on the spanning tree have to be
2023 instrumented: the compiler adds code to count the number of times that these
2024 arcs are executed. When an arc is the only exit or only entrance to a
2025 block, the instrumentation code can be added to the block; otherwise, a
2026 new basic block must be created to hold the instrumentation code.
2028 Since not every arc in the program must be instrumented, programs
2029 compiled with this option run faster than programs compiled with
2030 @samp{-a}, which adds instrumentation code to every basic block in the
2031 program. The tradeoff: since @code{gcov} does not have
2032 execution counts for all branches, it must start with the execution
2033 counts for the instrumented branches, and then iterate over the program
2034 flow graph until the entire graph has been solved. Hence, @code{gcov}
2035 runs a little more slowly than a program which uses information from
2038 @samp{-fprofile-arcs} also makes it possible to estimate branch
2039 probabilities, and to calculate basic block execution counts. In
2040 general, basic block execution counts do not give enough information to
2041 estimate all branch probabilities. When the compiled program exits, it
2042 saves the arc execution counts to a file called
2043 @file{@var{sourcename}.da}. Use the compiler option
2044 @samp{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2045 Control Optimization}) when recompiling, to optimize using estimated
2046 branch probabilities.
2049 @item -ftest-coverage
2050 Create data files for the @code{gcov} code-coverage utility
2051 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2052 The data file names begin with the name of your source file:
2055 @item @var{sourcename}.bb
2056 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2057 associate basic block execution counts with line numbers.
2059 @item @var{sourcename}.bbg
2060 A list of all arcs in the program flow graph. This allows @code{gcov}
2061 to reconstruct the program flow graph, so that it can compute all basic
2062 block and arc execution counts from the information in the
2063 @code{@var{sourcename}.da} file (this last file is the output from
2064 @samp{-fprofile-arcs}).
2068 Makes the compiler print out each function name as it is compiled, and
2069 print some statistics about each pass when it finishes.
2071 @item -d@var{letters}
2072 Says to make debugging dumps during compilation at times specified by
2073 @var{letters}. This is used for debugging the compiler. The file names
2074 for most of the dumps are made by appending a word to the source file
2075 name (e.g. @file{foo.c.rtl} or @file{foo.c.jump}). Here are the
2076 possible letters for use in @var{letters}, and their meanings:
2080 Dump after computing branch probabilities, to @file{@var{file}.bp}.
2082 Dump after instruction combination, to the file @file{@var{file}.combine}.
2084 Dump after delayed branch scheduling, to @file{@var{file}.dbr}.
2086 Dump all macro definitions, at the end of preprocessing, in addition to
2089 Dump after RTL generation, to @file{@var{file}.rtl}.
2091 Dump after first jump optimization, to @file{@var{file}.jump}.
2093 Dump after purging ADDRESSOF, to @file{@var{file}.addressof}.
2095 Dump after flow analysis, to @file{@var{file}.flow}.
2097 Dump after global register allocation, to @file{@var{file}.greg}.
2099 Dump after GCSE, to @file{@var{file}.gcse}.
2101 Dump after first jump optimization, to @file{@var{file}.jump}.
2103 Dump after last jump optimization, to @file{@var{file}.jump2}.
2105 Dump after conversion from registers to stack, to @file{@var{file}.stack}.
2107 Dump after local register allocation, to @file{@var{file}.lreg}.
2109 Dump after loop optimization, to @file{@var{file}.loop}.
2111 Dump after performing the machine dependent reorganisation pass, to
2112 @file{@var{file}.mach}.
2114 Dump after the register move pass, to @file{@var{file}.regmove}.
2116 Dump after RTL generation, to @file{@var{file}.rtl}.
2118 Dump after the second instruction scheduling pass, to @file{@var{file}.sched2}.
2120 Dump after CSE (including the jump optimization that sometimes follows
2121 CSE), to @file{@var{file}.cse}.
2123 Dump after the first instruction scheduling pass, to @file{@var{file}.sched}.
2125 Dump after the second CSE pass (including the jump optimization that
2126 sometimes follows CSE), to @file{@var{file}.cse2}.
2128 Produce all the dumps listed above.
2130 Print statistics on memory usage, at the end of the run, to
2133 Annotate the assembler output with a comment indicating which
2134 pattern and alternative was used. The length of each instruction is
2137 Just generate RTL for a function instead of compiling it. Usually used
2140 Dump debugging information during parsing, to standard error.
2142 Annotate the assembler output with miscellaneous debugging information.
2145 @item -fdump-unnumbered
2146 When doing debugging dumps (see -d option above), suppress instruction
2147 numbers and line number note output. This makes it more feasible to
2148 use diff on debugging dumps for compiler invokations with different
2149 options, in particular with and without -g.
2151 @item -fdump-translation-unit-@var{file} (C++ only)
2152 Dump a representation of the tree structure for the entire translation
2155 @item -fpretend-float
2156 When running a cross-compiler, pretend that the target machine uses the
2157 same floating point format as the host machine. This causes incorrect
2158 output of the actual floating constants, but the actual instruction
2159 sequence will probably be the same as GCC would make when running on
2163 Store the usual ``temporary'' intermediate files permanently; place them
2164 in the current directory and name them based on the source file. Thus,
2165 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
2166 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}.
2169 Report the CPU time taken by each subprocess in the compilation
2170 sequence. For C source files, this is the preprocessor, compiler
2171 proper, and assembler. The output looks like this:
2179 The first number on each line is the ``user time,'' that is time spent
2180 executing the program itself. The second number is ``system time,''
2181 time spent executing operating system routines on behalf of the program.
2182 Both numbers are in seconds.
2184 @item -print-file-name=@var{library}
2185 Print the full absolute name of the library file @var{library} that
2186 would be used when linking---and don't do anything else. With this
2187 option, GCC does not compile or link anything; it just prints the
2190 @item -print-prog-name=@var{program}
2191 Like @samp{-print-file-name}, but searches for a program such as @samp{cpp}.
2193 @item -print-libgcc-file-name
2194 Same as @samp{-print-file-name=libgcc.a}.
2196 This is useful when you use @samp{-nostdlib} or @samp{-nodefaultlibs}
2197 but you do want to link with @file{libgcc.a}. You can do
2200 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
2203 @item -print-search-dirs
2204 Print the name of the configured installation directory and a list of
2205 program and library directories gcc will search---and don't do anything else.
2207 This is useful when gcc prints the error message
2208 @samp{installation problem, cannot exec cpp: No such file or directory}.
2209 To resolve this you either need to put @file{cpp} and the other compiler
2210 components where gcc expects to find them, or you can set the environment
2211 variable @code{GCC_EXEC_PREFIX} to the directory where you installed them.
2212 Don't forget the trailing '/'.
2213 @xref{Environment Variables}.
2216 @node Optimize Options
2217 @section Options That Control Optimization
2218 @cindex optimize options
2219 @cindex options, optimization
2221 These options control various sorts of optimizations:
2226 Optimize. Optimizing compilation takes somewhat more time, and a lot
2227 more memory for a large function.
2229 Without @samp{-O}, the compiler's goal is to reduce the cost of
2230 compilation and to make debugging produce the expected results.
2231 Statements are independent: if you stop the program with a breakpoint
2232 between statements, you can then assign a new value to any variable or
2233 change the program counter to any other statement in the function and
2234 get exactly the results you would expect from the source code.
2236 Without @samp{-O}, the compiler only allocates variables declared
2237 @code{register} in registers. The resulting compiled code is a little
2238 worse than produced by PCC without @samp{-O}.
2240 With @samp{-O}, the compiler tries to reduce code size and execution
2243 When you specify @samp{-O}, the compiler turns on @samp{-fthread-jumps}
2244 and @samp{-fdefer-pop} on all machines. The compiler turns on
2245 @samp{-fdelayed-branch} on machines that have delay slots, and
2246 @samp{-fomit-frame-pointer} on machines that can support debugging even
2247 without a frame pointer. On some machines the compiler also turns
2248 on other flags.@refill
2251 Optimize even more. GCC performs nearly all supported optimizations
2252 that do not involve a space-speed tradeoff. The compiler does not
2253 perform loop unrolling or function inlining when you specify @samp{-O2}.
2254 As compared to @samp{-O}, this option increases both compilation time
2255 and the performance of the generated code.
2257 @samp{-O2} turns on all optional optimizations except for loop unrolling
2258 and function inlining. It also turns on the @samp{-fforce-mem} option
2259 on all machines and frame pointer elimination on machines where doing so
2260 does not interfere with debugging.
2263 Optimize yet more. @samp{-O3} turns on all optimizations specified by
2264 @samp{-O2} and also turns on the @samp{inline-functions} option.
2270 Optimize for size. @samp{-Os} enables all @samp{-O2} optimizations that
2271 do not typically increase code size. It also performs further
2272 optimizations designed to reduce code size.
2274 If you use multiple @samp{-O} options, with or without level numbers,
2275 the last such option is the one that is effective.
2278 Options of the form @samp{-f@var{flag}} specify machine-independent
2279 flags. Most flags have both positive and negative forms; the negative
2280 form of @samp{-ffoo} would be @samp{-fno-foo}. In the table below,
2281 only one of the forms is listed---the one which is not the default.
2282 You can figure out the other form by either removing @samp{no-} or
2287 Do not store floating point variables in registers, and inhibit other
2288 options that might change whether a floating point value is taken from a
2291 @cindex floating point precision
2292 This option prevents undesirable excess precision on machines such as
2293 the 68000 where the floating registers (of the 68881) keep more
2294 precision than a @code{double} is supposed to have. Similarly for the
2295 x86 architecture. For most programs, the excess precision does only
2296 good, but a few programs rely on the precise definition of IEEE floating
2297 point. Use @samp{-ffloat-store} for such programs, after modifying
2298 them to store all pertinent intermediate computations into variables.
2300 @item -fno-default-inline
2301 Do not make member functions inline by default merely because they are
2302 defined inside the class scope (C++ only). Otherwise, when you specify
2303 @w{@samp{-O}}, member functions defined inside class scope are compiled
2304 inline by default; i.e., you don't need to add @samp{inline} in front of
2305 the member function name.
2307 @item -fno-defer-pop
2308 Always pop the arguments to each function call as soon as that function
2309 returns. For machines which must pop arguments after a function call,
2310 the compiler normally lets arguments accumulate on the stack for several
2311 function calls and pops them all at once.
2314 Force memory operands to be copied into registers before doing
2315 arithmetic on them. This produces better code by making all memory
2316 references potential common subexpressions. When they are not common
2317 subexpressions, instruction combination should eliminate the separate
2318 register-load. The @samp{-O2} option turns on this option.
2321 Force memory address constants to be copied into registers before
2322 doing arithmetic on them. This may produce better code just as
2323 @samp{-fforce-mem} may.
2325 @item -fomit-frame-pointer
2326 Don't keep the frame pointer in a register for functions that
2327 don't need one. This avoids the instructions to save, set up and
2328 restore frame pointers; it also makes an extra register available
2329 in many functions. @strong{It also makes debugging impossible on
2333 On some machines, such as the Vax, this flag has no effect, because
2334 the standard calling sequence automatically handles the frame pointer
2335 and nothing is saved by pretending it doesn't exist. The
2336 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2337 whether a target machine supports this flag. @xref{Registers}.@refill
2340 On some machines, such as the Vax, this flag has no effect, because
2341 the standard calling sequence automatically handles the frame pointer
2342 and nothing is saved by pretending it doesn't exist. The
2343 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
2344 whether a target machine supports this flag. @xref{Registers,,Register
2345 Usage, gcc.info, Using and Porting GCC}.@refill
2349 Don't pay attention to the @code{inline} keyword. Normally this option
2350 is used to keep the compiler from expanding any functions inline.
2351 Note that if you are not optimizing, no functions can be expanded inline.
2353 @item -finline-functions
2354 Integrate all simple functions into their callers. The compiler
2355 heuristically decides which functions are simple enough to be worth
2356 integrating in this way.
2358 If all calls to a given function are integrated, and the function is
2359 declared @code{static}, then the function is normally not output as
2360 assembler code in its own right.
2362 @item -finline-limit=@var{n}
2363 By default, gcc limits the size of functions that can be inlined. This flag
2364 allows the control of this limit for functions that are explicitly marked as
2365 inline (ie marked with the inline keyword or defined within the class
2366 definition in c++). @var{n} is the size of functions that can be inlined in
2367 number of pseudo instructions (not counting parameter handling). The default
2368 value of n is 10000. Increasing this value can result in more inlined code at
2369 the cost of compilation time and memory consumption. Decreasing usually makes
2370 the compilation faster and less code will be inlined (which presumably
2371 means slower programs). This option is particularly useful for programs that
2372 use inlining heavily such as those based on recursive templates with c++.
2374 @emph{Note:} pseudo instruction represents, in this particular context, an
2375 abstract measurement of function's size. In no way, it represents a count
2376 of assembly instructions and as such its exact meaning might change from one
2377 release to an another.
2379 @item -fkeep-inline-functions
2380 Even if all calls to a given function are integrated, and the function
2381 is declared @code{static}, nevertheless output a separate run-time
2382 callable version of the function. This switch does not affect
2383 @code{extern inline} functions.
2385 @item -fkeep-static-consts
2386 Emit variables declared @code{static const} when optimization isn't turned
2387 on, even if the variables aren't referenced.
2389 GCC enables this option by default. If you want to force the compiler to
2390 check if the variable was referenced, regardless of whether or not
2391 optimization is turned on, use the @samp{-fno-keep-static-consts} option.
2393 @item -fno-function-cse
2394 Do not put function addresses in registers; make each instruction that
2395 calls a constant function contain the function's address explicitly.
2397 This option results in less efficient code, but some strange hacks
2398 that alter the assembler output may be confused by the optimizations
2399 performed when this option is not used.
2402 This option allows GCC to violate some ANSI or IEEE rules and/or
2403 specifications in the interest of optimizing code for speed. For
2404 example, it allows the compiler to assume arguments to the @code{sqrt}
2405 function are non-negative numbers and that no floating-point values
2408 This option should never be turned on by any @samp{-O} option since
2409 it can result in incorrect output for programs which depend on
2410 an exact implementation of IEEE or ANSI rules/specifications for
2414 @c following causes underfulls.. they don't look great, but we deal.
2416 The following options control specific optimizations. The @samp{-O2}
2417 option turns on all of these optimizations except @samp{-funroll-loops}
2418 and @samp{-funroll-all-loops}. On most machines, the @samp{-O} option
2419 turns on the @samp{-fthread-jumps} and @samp{-fdelayed-branch} options,
2420 but specific machines may handle it differently.
2422 You can use the following flags in the rare cases when ``fine-tuning''
2423 of optimizations to be performed is desired.
2426 @item -fstrength-reduce
2427 Perform the optimizations of loop strength reduction and
2428 elimination of iteration variables.
2430 @item -fthread-jumps
2431 Perform optimizations where we check to see if a jump branches to a
2432 location where another comparison subsumed by the first is found. If
2433 so, the first branch is redirected to either the destination of the
2434 second branch or a point immediately following it, depending on whether
2435 the condition is known to be true or false.
2437 @item -fcse-follow-jumps
2438 In common subexpression elimination, scan through jump instructions
2439 when the target of the jump is not reached by any other path. For
2440 example, when CSE encounters an @code{if} statement with an
2441 @code{else} clause, CSE will follow the jump when the condition
2444 @item -fcse-skip-blocks
2445 This is similar to @samp{-fcse-follow-jumps}, but causes CSE to
2446 follow jumps which conditionally skip over blocks. When CSE
2447 encounters a simple @code{if} statement with no else clause,
2448 @samp{-fcse-skip-blocks} causes CSE to follow the jump around the
2449 body of the @code{if}.
2451 @item -frerun-cse-after-loop
2452 Re-run common subexpression elimination after loop optimizations has been
2455 @item -frerun-loop-opt
2456 Run the loop optimizer twice.
2459 Perform a global common subexpression elimination pass.
2460 This pass also performs global constant and copy propagation.
2462 @item -fexpensive-optimizations
2463 Perform a number of minor optimizations that are relatively expensive.
2465 @item -foptimize-register-moves
2467 Attempt to reassign register numbers in move instructions and as
2468 operands of other simple instructions in order to maximize the amount of
2469 register tying. This is especially helpful on machines with two-operand
2470 instructions. GCC enables this optimization by default with @samp{-O2}
2473 Note @code{-fregmove} and @code{-foptimize-register-moves} are the same
2476 @item -fdelayed-branch
2477 If supported for the target machine, attempt to reorder instructions
2478 to exploit instruction slots available after delayed branch
2481 @item -fschedule-insns
2482 If supported for the target machine, attempt to reorder instructions to
2483 eliminate execution stalls due to required data being unavailable. This
2484 helps machines that have slow floating point or memory load instructions
2485 by allowing other instructions to be issued until the result of the load
2486 or floating point instruction is required.
2488 @item -fschedule-insns2
2489 Similar to @samp{-fschedule-insns}, but requests an additional pass of
2490 instruction scheduling after register allocation has been done. This is
2491 especially useful on machines with a relatively small number of
2492 registers and where memory load instructions take more than one cycle.
2494 @item -ffunction-sections
2495 @itemx -fdata-sections
2496 Place each function or data item into its own section in the output
2497 file if the target supports arbitrary sections. The name of the
2498 function or the name of the data item determines the section's name
2501 Use these options on systems where the linker can perform optimizations
2502 to improve locality of reference in the instruction space. HPPA
2503 processors running HP-UX and Sparc processors running Solaris 2 have
2504 linkers with such optimizations. Other systems using the ELF object format
2505 as well as AIX may have these optimizations in the future.
2507 Only use these options when there are significant benefits from doing
2508 so. When you specify these options, the assembler and linker will
2509 create larger object and executable files and will also be slower.
2510 You will not be able to use @code{gprof} on all systems if you
2511 specify this option and you may have problems with debugging if
2512 you specify both this option and @samp{-g}.
2514 @item -fcaller-saves
2515 Enable values to be allocated in registers that will be clobbered by
2516 function calls, by emitting extra instructions to save and restore the
2517 registers around such calls. Such allocation is done only when it
2518 seems to result in better code than would otherwise be produced.
2520 This option is always enabled by default on certain machines, usually
2521 those which have no call-preserved registers to use instead.
2523 For all machines, optimization level 2 and higher enables this flag by
2526 @item -funroll-loops
2527 Perform the optimization of loop unrolling. This is only done for loops
2528 whose number of iterations can be determined at compile time or run time.
2529 @samp{-funroll-loops} implies both @samp{-fstrength-reduce} and
2530 @samp{-frerun-cse-after-loop}.
2532 @item -funroll-all-loops
2533 Perform the optimization of loop unrolling. This is done for all loops
2534 and usually makes programs run more slowly. @samp{-funroll-all-loops}
2535 implies @samp{-fstrength-reduce} as well as @samp{-frerun-cse-after-loop}.
2537 @item -fmove-all-movables
2538 Forces all invariant computations in loops to be moved
2541 @item -freduce-all-givs
2542 Forces all general-induction variables in loops to be
2545 @emph{Note:} When compiling programs written in Fortran,
2546 @samp{-fmove-all-movables} and @samp{-freduce-all-givs} are enabled
2547 by default when you use the optimizer.
2549 These options may generate better or worse code; results are highly
2550 dependent on the structure of loops within the source code.
2552 These two options are intended to be removed someday, once
2553 they have helped determine the efficacy of various
2554 approaches to improving loop optimizations.
2556 Please let us (@code{gcc@@gcc.gnu.org} and @code{fortran@@gnu.org})
2557 know how use of these options affects
2558 the performance of your production code.
2559 We're very interested in code that runs @emph{slower}
2560 when these options are @emph{enabled}.
2563 Disable any machine-specific peephole optimizations.
2565 @item -fbranch-probabilities
2566 After running a program compiled with @samp{-fprofile-arcs}
2567 (@pxref{Debugging Options,, Options for Debugging Your Program or
2568 @code{gcc}}), you can compile it a second time using
2569 @samp{-fbranch-probabilities}, to improve optimizations based on
2570 guessing the path a branch might take.
2573 With @samp{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
2574 note on the first instruction of each basic block, and a
2575 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
2576 These can be used to improve optimization. Currently, they are only
2577 used in one place: in @file{reorg.c}, instead of guessing which path a
2578 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
2579 exactly determine which path is taken more often.
2582 @item -fstrict-aliasing
2583 Allows the compiler to assume the strictest aliasing rules applicable to
2584 the language being compiled. For C (and C++), this activates
2585 optimizations based on the type of expressions. In particular, an
2586 object of one type is assumed never to reside at the same address as an
2587 object of a different type, unless the types are almost the same. For
2588 example, an @code{unsigned int} can alias an @code{int}, but not a
2589 @code{void*} or a @code{double}. A character type may alias any other
2592 Pay special attention to code like this:
2605 The practice of reading from a different union member than the one most
2606 recently written to (called ``type-punning'') is common. Even with
2607 @samp{-fstrict-aliasing}, type-punning is allowed, provided the memory
2608 is accessed through the union type. So, the code above will work as
2609 expected. However, this code might not:
2621 Every language that wishes to perform language-specific alias analysis
2622 should define a function that computes, given an @code{tree}
2623 node, an alias set for the node. Nodes in different alias sets are not
2624 allowed to alias. For an example, see the C front-end function
2625 @code{c_get_alias_set}.
2628 @item -falign-functions
2629 @itemx -falign-functions=@var{n}
2630 Align the start of functions to the next power-of-two greater than
2631 @var{n}, skipping up to @var{n} bytes. For instance,
2632 @samp{-falign-functions=32} aligns functions to the next 32-byte
2633 boundary, but @samp{-falign-functions=24} would align to the next
2634 32-byte boundary only if this can be done by skipping 23 bytes or less.
2636 @samp{-fno-align-functions} and @samp{-falign-functions=1} are
2637 equivalent and mean that functions will not be aligned.
2639 Some assemblers only support this flag when @var{n} is a power of two;
2640 in that case, it is rounded up.
2642 If @var{n} is not specified, use a machine-dependent default.
2644 @item -falign-labels
2645 @itemx -falign-labels=@var{n}
2646 Align all branch targets to a power-of-two boundary, skipping up to
2647 @var{n} bytes like @samp{-falign-functions}. This option can easily
2648 make code slower, because it must insert dummy operations for when the
2649 branch target is reached in the usual flow of the code.
2651 If @samp{-falign-loops} or @samp{-falign-jumps} are applicable and
2652 are greater than this value, then their values are used instead.
2654 If @var{n} is not specified, use a machine-dependent default which is
2655 very likely to be @samp{1}, meaning no alignment.
2658 @itemx -falign-loops=@var{n}
2659 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
2660 like @samp{-falign-functions}. The hope is that the loop will be
2661 executed many times, which will make up for any execution of the dummy
2664 If @var{n} is not specified, use a machine-dependent default.
2667 @itemx -falign-jumps=@var{n}
2668 Align branch targets to a power-of-two boundary, for branch targets
2669 where the targets can only be reached by jumping, skipping up to @var{n}
2670 bytes like @samp{-falign-functions}. In this case, no dummy operations
2673 If @var{n} is not specified, use a machine-dependent default.
2677 @node Preprocessor Options
2678 @section Options Controlling the Preprocessor
2679 @cindex preprocessor options
2680 @cindex options, preprocessor
2682 These options control the C preprocessor, which is run on each C source
2683 file before actual compilation.
2685 If you use the @samp{-E} option, nothing is done except preprocessing.
2686 Some of these options make sense only together with @samp{-E} because
2687 they cause the preprocessor output to be unsuitable for actual
2691 @item -include @var{file}
2692 Process @var{file} as input before processing the regular input file.
2693 In effect, the contents of @var{file} are compiled first. Any @samp{-D}
2694 and @samp{-U} options on the command line are always processed before
2695 @samp{-include @var{file}}, regardless of the order in which they are
2696 written. All the @samp{-include} and @samp{-imacros} options are
2697 processed in the order in which they are written.
2699 @item -imacros @var{file}
2700 Process @var{file} as input, discarding the resulting output, before
2701 processing the regular input file. Because the output generated from
2702 @var{file} is discarded, the only effect of @samp{-imacros @var{file}}
2703 is to make the macros defined in @var{file} available for use in the
2706 Any @samp{-D} and @samp{-U} options on the command line are always
2707 processed before @samp{-imacros @var{file}}, regardless of the order in
2708 which they are written. All the @samp{-include} and @samp{-imacros}
2709 options are processed in the order in which they are written.
2711 @item -idirafter @var{dir}
2712 @cindex second include path
2713 Add the directory @var{dir} to the second include path. The directories
2714 on the second include path are searched when a header file is not found
2715 in any of the directories in the main include path (the one that
2718 @item -iprefix @var{prefix}
2719 Specify @var{prefix} as the prefix for subsequent @samp{-iwithprefix}
2722 @item -iwithprefix @var{dir}
2723 Add a directory to the second include path. The directory's name is
2724 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
2725 specified previously with @samp{-iprefix}. If you have not specified a
2726 prefix yet, the directory containing the installed passes of the
2727 compiler is used as the default.
2729 @item -iwithprefixbefore @var{dir}
2730 Add a directory to the main include path. The directory's name is made
2731 by concatenating @var{prefix} and @var{dir}, as in the case of
2732 @samp{-iwithprefix}.
2734 @item -isystem @var{dir}
2735 Add a directory to the beginning of the second include path, marking it
2736 as a system directory, so that it gets the same special treatment as
2737 is applied to the standard system directories.
2740 Do not search the standard system directories for header files. Only
2741 the directories you have specified with @samp{-I} options (and the
2742 current directory, if appropriate) are searched. @xref{Directory
2743 Options}, for information on @samp{-I}.
2745 By using both @samp{-nostdinc} and @samp{-I-}, you can limit the include-file
2746 search path to only those directories you specify explicitly.
2749 Do not predefine any nonstandard macros. (Including architecture flags).
2752 Run only the C preprocessor. Preprocess all the C source files
2753 specified and output the results to standard output or to the
2754 specified output file.
2757 Tell the preprocessor not to discard comments. Used with the
2761 Tell the preprocessor not to generate @samp{#line} directives.
2762 Used with the @samp{-E} option.
2765 @cindex dependencies, make
2767 Tell the preprocessor to output a rule suitable for @code{make}
2768 describing the dependencies of each object file. For each source file,
2769 the preprocessor outputs one @code{make}-rule whose target is the object
2770 file name for that source file and whose dependencies are all the
2771 @code{#include} header files it uses. This rule may be a single line or
2772 may be continued with @samp{\}-newline if it is long. The list of rules
2773 is printed on standard output instead of the preprocessed C program.
2775 @samp{-M} implies @samp{-E}.
2777 Another way to specify output of a @code{make} rule is by setting
2778 the environment variable @code{DEPENDENCIES_OUTPUT} (@pxref{Environment
2782 Like @samp{-M} but the output mentions only the user header files
2783 included with @samp{#include "@var{file}"}. System header files
2784 included with @samp{#include <@var{file}>} are omitted.
2787 Like @samp{-M} but the dependency information is written to a file made by
2788 replacing ".c" with ".d" at the end of the input file names.
2789 This is in addition to compiling the file as specified---@samp{-MD} does
2790 not inhibit ordinary compilation the way @samp{-M} does.
2792 In Mach, you can use the utility @code{md} to merge multiple dependency
2793 files into a single dependency file suitable for using with the @samp{make}
2797 Like @samp{-MD} except mention only user header files, not system
2801 Treat missing header files as generated files and assume they live in the
2802 same directory as the source file. If you specify @samp{-MG}, you
2803 must also specify either @samp{-M} or @samp{-MM}. @samp{-MG} is not
2804 supported with @samp{-MD} or @samp{-MMD}.
2807 Print the name of each header file used, in addition to other normal
2810 @item -A@var{question}(@var{answer})
2811 Assert the answer @var{answer} for @var{question}, in case it is tested
2812 with a preprocessing conditional such as @samp{#if
2813 #@var{question}(@var{answer})}. @samp{-A-} disables the standard
2814 assertions that normally describe the target machine.
2817 Define macro @var{macro} with the string @samp{1} as its definition.
2819 @item -D@var{macro}=@var{defn}
2820 Define macro @var{macro} as @var{defn}. All instances of @samp{-D} on
2821 the command line are processed before any @samp{-U} options.
2824 Undefine macro @var{macro}. @samp{-U} options are evaluated after all
2825 @samp{-D} options, but before any @samp{-include} and @samp{-imacros}
2829 Tell the preprocessor to output only a list of the macro definitions
2830 that are in effect at the end of preprocessing. Used with the @samp{-E}
2834 Tell the preprocessing to pass all macro definitions into the output, in
2835 their proper sequence in the rest of the output.
2838 Like @samp{-dD} except that the macro arguments and contents are omitted.
2839 Only @samp{#define @var{name}} is included in the output.
2842 Support ANSI C trigraphs. The @samp{-ansi} option also has this effect.
2844 @item -Wp,@var{option}
2845 Pass @var{option} as an option to the preprocessor. If @var{option}
2846 contains commas, it is split into multiple options at the commas.
2849 @node Assembler Options
2850 @section Passing Options to the Assembler
2852 @c prevent bad page break with this line
2853 You can pass options to the assembler.
2856 @item -Wa,@var{option}
2857 Pass @var{option} as an option to the assembler. If @var{option}
2858 contains commas, it is split into multiple options at the commas.
2862 @section Options for Linking
2863 @cindex link options
2864 @cindex options, linking
2866 These options come into play when the compiler links object files into
2867 an executable output file. They are meaningless if the compiler is
2868 not doing a link step.
2872 @item @var{object-file-name}
2873 A file name that does not end in a special recognized suffix is
2874 considered to name an object file or library. (Object files are
2875 distinguished from libraries by the linker according to the file
2876 contents.) If linking is done, these object files are used as input
2882 If any of these options is used, then the linker is not run, and
2883 object file names should not be used as arguments. @xref{Overall
2887 @item -l@var{library}
2888 Search the library named @var{library} when linking.
2890 It makes a difference where in the command you write this option; the
2891 linker searches processes libraries and object files in the order they
2892 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
2893 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
2894 to functions in @samp{z}, those functions may not be loaded.
2896 The linker searches a standard list of directories for the library,
2897 which is actually a file named @file{lib@var{library}.a}. The linker
2898 then uses this file as if it had been specified precisely by name.
2900 The directories searched include several standard system directories
2901 plus any that you specify with @samp{-L}.
2903 Normally the files found this way are library files---archive files
2904 whose members are object files. The linker handles an archive file by
2905 scanning through it for members which define symbols that have so far
2906 been referenced but not defined. But if the file that is found is an
2907 ordinary object file, it is linked in the usual fashion. The only
2908 difference between using an @samp{-l} option and specifying a file name
2909 is that @samp{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
2910 and searches several directories.
2913 You need this special case of the @samp{-l} option in order to
2914 link an Objective C program.
2917 Do not use the standard system startup files when linking.
2918 The standard system libraries are used normally, unless @code{-nostdlib}
2919 or @code{-nodefaultlibs} is used.
2921 @item -nodefaultlibs
2922 Do not use the standard system libraries when linking.
2923 Only the libraries you specify will be passed to the linker.
2924 The standard startup files are used normally, unless @code{-nostartfiles}
2925 is used. The compiler may generate calls to memcmp, memset, and memcpy
2926 for System V (and ANSI C) environments or to bcopy and bzero for
2927 BSD environments. These entries are usually resolved by entries in
2928 libc. These entry points should be supplied through some other
2929 mechanism when this option is specified.
2932 Do not use the standard system startup files or libraries when linking.
2933 No startup files and only the libraries you specify will be passed to
2934 the linker. The compiler may generate calls to memcmp, memset, and memcpy
2935 for System V (and ANSI C) environments or to bcopy and bzero for
2936 BSD environments. These entries are usually resolved by entries in
2937 libc. These entry points should be supplied through some other
2938 mechanism when this option is specified.
2940 @cindex @code{-lgcc}, use with @code{-nostdlib}
2941 @cindex @code{-nostdlib} and unresolved references
2942 @cindex unresolved references and @code{-nostdlib}
2943 @cindex @code{-lgcc}, use with @code{-nodefaultlibs}
2944 @cindex @code{-nodefaultlibs} and unresolved references
2945 @cindex unresolved references and @code{-nodefaultlibs}
2946 One of the standard libraries bypassed by @samp{-nostdlib} and
2947 @samp{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
2948 that GCC uses to overcome shortcomings of particular machines, or special
2949 needs for some languages.
2951 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
2955 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
2956 for more discussion of @file{libgcc.a}.)
2958 In most cases, you need @file{libgcc.a} even when you want to avoid
2959 other standard libraries. In other words, when you specify @samp{-nostdlib}
2960 or @samp{-nodefaultlibs} you should usually specify @samp{-lgcc} as well.
2961 This ensures that you have no unresolved references to internal GCC
2962 library subroutines. (For example, @samp{__main}, used to ensure C++
2963 constructors will be called; @pxref{Collect2,,@code{collect2}}.)
2966 Remove all symbol table and relocation information from the executable.
2969 On systems that support dynamic linking, this prevents linking with the shared
2970 libraries. On other systems, this option has no effect.
2973 Produce a shared object which can then be linked with other objects to
2974 form an executable. Not all systems support this option. You must
2975 also specify @samp{-fpic} or @samp{-fPIC} on some systems when
2976 you specify this option.
2979 Bind references to global symbols when building a shared object. Warn
2980 about any unresolved references (unless overridden by the link editor
2981 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
2984 @item -Xlinker @var{option}
2985 Pass @var{option} as an option to the linker. You can use this to
2986 supply system-specific linker options which GCC does not know how to
2989 If you want to pass an option that takes an argument, you must use
2990 @samp{-Xlinker} twice, once for the option and once for the argument.
2991 For example, to pass @samp{-assert definitions}, you must write
2992 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
2993 @samp{-Xlinker "-assert definitions"}, because this passes the entire
2994 string as a single argument, which is not what the linker expects.
2996 @item -Wl,@var{option}
2997 Pass @var{option} as an option to the linker. If @var{option} contains
2998 commas, it is split into multiple options at the commas.
3000 @item -u @var{symbol}
3001 Pretend the symbol @var{symbol} is undefined, to force linking of
3002 library modules to define it. You can use @samp{-u} multiple times with
3003 different symbols to force loading of additional library modules.
3006 @node Directory Options
3007 @section Options for Directory Search
3008 @cindex directory options
3009 @cindex options, directory search
3012 These options specify directories to search for header files, for
3013 libraries and for parts of the compiler:
3017 Add the directory @var{dir} to the head of the list of directories to be
3018 searched for header files. This can be used to override a system header
3019 file, substituting your own version, since these directories are
3020 searched before the system header file directories. If you use more
3021 than one @samp{-I} option, the directories are scanned in left-to-right
3022 order; the standard system directories come after.
3025 Any directories you specify with @samp{-I} options before the @samp{-I-}
3026 option are searched only for the case of @samp{#include "@var{file}"};
3027 they are not searched for @samp{#include <@var{file}>}.
3029 If additional directories are specified with @samp{-I} options after
3030 the @samp{-I-}, these directories are searched for all @samp{#include}
3031 directives. (Ordinarily @emph{all} @samp{-I} directories are used
3034 In addition, the @samp{-I-} option inhibits the use of the current
3035 directory (where the current input file came from) as the first search
3036 directory for @samp{#include "@var{file}"}. There is no way to
3037 override this effect of @samp{-I-}. With @samp{-I.} you can specify
3038 searching the directory which was current when the compiler was
3039 invoked. That is not exactly the same as what the preprocessor does
3040 by default, but it is often satisfactory.
3042 @samp{-I-} does not inhibit the use of the standard system directories
3043 for header files. Thus, @samp{-I-} and @samp{-nostdinc} are
3047 Add directory @var{dir} to the list of directories to be searched
3050 @item -B@var{prefix}
3051 This option specifies where to find the executables, libraries,
3052 include files, and data files of the compiler itself.
3054 The compiler driver program runs one or more of the subprograms
3055 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
3056 @var{prefix} as a prefix for each program it tries to run, both with and
3057 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
3059 For each subprogram to be run, the compiler driver first tries the
3060 @samp{-B} prefix, if any. If that name is not found, or if @samp{-B}
3061 was not specified, the driver tries two standard prefixes, which are
3062 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
3063 those results in a file name that is found, the unmodified program
3064 name is searched for using the directories specified in your
3065 @samp{PATH} environment variable.
3067 @samp{-B} prefixes that effectively specify directory names also apply
3068 to libraries in the linker, because the compiler translates these
3069 options into @samp{-L} options for the linker. They also apply to
3070 includes files in the preprocessor, because the compiler translates these
3071 options into @samp{-isystem} options for the preprocessor. In this case,
3072 the compiler appends @samp{include} to the prefix.
3074 The run-time support file @file{libgcc.a} can also be searched for using
3075 the @samp{-B} prefix, if needed. If it is not found there, the two
3076 standard prefixes above are tried, and that is all. The file is left
3077 out of the link if it is not found by those means.
3079 Another way to specify a prefix much like the @samp{-B} prefix is to use
3080 the environment variable @code{GCC_EXEC_PREFIX}. @xref{Environment
3083 @item -specs=@var{file}
3084 Process @var{file} after the compiler reads in the standard @file{specs}
3085 file, in order to override the defaults that the @file{gcc} driver
3086 program uses when determining what switches to pass to @file{cc1},
3087 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
3088 @samp{-specs=}@var{file} can be specified on the command line, and they
3089 are processed in order, from left to right.
3093 @section Specifying subprocesses and the switches to pass to them
3095 @code{GCC} is a driver program. It performs its job by invoking a
3096 sequence of other programs to do the work of compiling, assembling and
3097 linking. GCC interprets its command-line parameters and uses these to
3098 deduce which programs it should invoke, and which command-line options
3099 it ought to place on their command lines. This behaviour is controlled
3100 by @dfn{spec strings}. In most cases there is one spec string for each
3101 program that GCC can invoke, but a few programs have multiple spec
3102 strings to control their behaviour. The spec strings built into GCC can
3103 be overridden by using the @samp{-specs=} command-line switch to specify
3106 @dfn{Spec files} are plaintext files that are used to construct spec
3107 strings. They consist of a sequence of directives separated by blank
3108 lines. The type of directive is determined by the first non-whitespace
3109 character on the line and it can be one of the following:
3112 @item %@var{command}
3113 Issues a @var{command} to the spec file processor. The commands that can
3117 @item %include <@var{file}>
3119 Search for @var{file} and insert its text at the current point in the
3122 @item %include_noerr <@var{file}>
3123 @cindex %include_noerr
3124 Just like @samp{%include}, but do not generate an error message if the include
3125 file cannot be found.
3127 @item %rename @var{old_name} @var{new_name}
3129 Rename the spec string @var{old_name} to @var{new_name}.
3133 @item *[@var{spec_name}]:
3134 This tells the compiler to create, override or delete the named spec
3135 string. All lines after this directive up to the next directive or
3136 blank line are considered to be the text for the spec string. If this
3137 results in an empty string then the spec will be deleted. (Or, if the
3138 spec did not exist, then nothing will happened.) Otherwise, if the spec
3139 does not currently exist a new spec will be created. If the spec does
3140 exist then its contents will be overridden by the text of this
3141 directive, unless the first character of that text is the @samp{+}
3142 character, in which case the text will be appended to the spec.
3144 @item [@var{suffix}]:
3145 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
3146 and up to the next directive or blank line are considered to make up the
3147 spec string for the indicated suffix. When the compiler encounters an
3148 input file with the named suffix, it will processes the spec string in
3149 order to work out how to compile that file. For example:
3156 This says that any input file whose name ends in @samp{.ZZ} should be
3157 passed to the program @samp{z-compile}, which should be invoked with the
3158 command-line switch @samp{-input} and with the result of performing the
3159 @samp{%i} substitution. (See below.)
3161 As an alternative to providing a spec string, the text that follows a
3162 suffix directive can be one of the following:
3165 @item @@@var{language}
3166 This says that the suffix is an alias for a known @var{language}. This is
3167 similar to using the @code{-x} command-line switch to GCC to specify a
3168 language explicitly. For example:
3175 Says that .ZZ files are, in fact, C++ source files.
3178 This causes an error messages saying:
3181 @var{name} compiler not installed on this system.
3185 GCC already has an extensive list of suffixes built into it.
3186 This directive will add an entry to the end of the list of suffixes, but
3187 since the list is searched from the end backwards, it is effectively
3188 possible to override earlier entries using this technique.
3192 GCC has the following spec strings built into it. Spec files can
3193 override these strings or create their own. Note that individual
3194 targets can also add their own spec strings to this list.
3197 asm Options to pass to the assembler
3198 asm_final Options to pass to the assembler post-processor
3199 cpp Options to pass to the C preprocessor
3200 cc1 Options to pass to the C compiler
3201 cc1plus Options to pass to the C++ compiler
3202 endfile Object files to include at the end of the link
3203 link Options to pass to the linker
3204 lib Libraries to include on the command line to the linker
3205 libgcc Decides which GCC support library to pass to the linker
3206 linker Sets the name of the linker
3207 predefines Defines to be passed to the C preprocessor
3208 signed_char Defines to pass to CPP to say whether @code{char} is signed by default
3209 startfile Object files to include at the start of the link
3212 Here is a small example of a spec file:
3218 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
3221 This example renames the spec called @samp{lib} to @samp{old_lib} and
3222 then overrides the previous definition of @samp{lib} with a new one.
3223 The new definition adds in some extra command-line options before
3224 including the text of the old definition.
3226 @dfn{Spec strings} are a list of command-line options to be passed to their
3227 corresponding program. In addition, the spec strings can contain
3228 @samp{%}-prefixed sequences to substitute variable text or to
3229 conditionally insert text into the command line. Using these constructs
3230 it is possible to generate quite complex command lines.
3232 Here is a table of all defined @samp{%}-sequences for spec
3233 strings. Note that spaces are not generated automatically around the
3234 results of expanding these sequences. Therefore you can concatenate them
3235 together or combine them with constant text in a single argument.
3239 Substitute one @samp{%} into the program name or argument.
3242 Substitute the name of the input file being processed.
3245 Substitute the basename of the input file being processed.
3246 This is the substring up to (and not including) the last period
3247 and not including the directory.
3250 Marks the argument containing or following the @samp{%d} as a
3251 temporary file name, so that that file will be deleted if GCC exits
3252 successfully. Unlike @samp{%g}, this contributes no text to the
3255 @item %g@var{suffix}
3256 Substitute a file name that has suffix @var{suffix} and is chosen
3257 once per compilation, and mark the argument in the same way as
3258 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
3259 name is now chosen in a way that is hard to predict even when previously
3260 chosen file names are known. For example, @samp{%g.s ... %g.o ... %g.s}
3261 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
3262 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
3263 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
3264 was simply substituted with a file name chosen once per compilation,
3265 without regard to any appended suffix (which was therefore treated
3266 just like ordinary text), making such attacks more likely to succeed.
3268 @item %u@var{suffix}
3269 Like @samp{%g}, but generates a new temporary file name even if
3270 @samp{%u@var{suffix}} was already seen.
3272 @item %U@var{suffix}
3273 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
3274 new one if there is no such last file name. In the absence of any
3275 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
3276 the same suffix @emph{space}, so @samp{%g.s ... %U.s ... %g.s ... %U.s}
3277 would involve the generation of two distinct file names, one
3278 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
3279 simply substituted with a file name chosen for the previous @samp{%u},
3280 without regard to any appended suffix.
3283 Marks the argument containing or following the @samp{%w} as the
3284 designated output file of this compilation. This puts the argument
3285 into the sequence of arguments that @samp{%o} will substitute later.
3288 Substitutes the names of all the output files, with spaces
3289 automatically placed around them. You should write spaces
3290 around the @samp{%o} as well or the results are undefined.
3291 @samp{%o} is for use in the specs for running the linker.
3292 Input files whose names have no recognized suffix are not compiled
3293 at all, but they are included among the output files, so they will
3297 Substitutes the suffix for object files. Note that this is
3298 handled specially when it immediately follows @samp{%g, %u, or %U},
3299 because of the need for those to form complete file names. The
3300 handling is such that @samp{%O} is treated exactly as if it had already
3301 been substituted, except that @samp{%g, %u, and %U} do not currently
3302 support additional @var{suffix} characters following @samp{%O} as they would
3303 following, for example, @samp{.o}.
3306 Substitutes the standard macro predefinitions for the
3307 current target machine. Use this when running @code{cpp}.
3310 Like @samp{%p}, but puts @samp{__} before and after the name of each
3311 predefined macro, except for macros that start with @samp{__} or with
3312 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ANSI
3316 Substitute a @samp{-iprefix} option made from GCC_EXEC_PREFIX.
3319 Current argument is the name of a library or startup file of some sort.
3320 Search for that file in a standard list of directories and substitute
3321 the full name found.
3324 Print @var{str} as an error message. @var{str} is terminated by a newline.
3325 Use this when inconsistent options are detected.
3328 Output @samp{-} if the input for the current command is coming from a pipe.
3331 Substitute the contents of spec string @var{name} at this point.
3334 Like @samp{%(...)} but put @samp{__} around @samp{-D} arguments.
3336 @item %x@{@var{option}@}
3337 Accumulate an option for @samp{%X}.
3340 Output the accumulated linker options specified by @samp{-Wl} or a @samp{%x}
3344 Output the accumulated assembler options specified by @samp{-Wa}.
3347 Output the accumulated preprocessor options specified by @samp{-Wp}.
3350 Substitute the major version number of GCC.
3351 (For version 2.9.5, this is 2.)
3354 Substitute the minor version number of GCC.
3355 (For version 2.9.5, this is 9.)
3358 Process the @code{asm} spec. This is used to compute the
3359 switches to be passed to the assembler.
3362 Process the @code{asm_final} spec. This is a spec string for
3363 passing switches to an assembler post-processor, if such a program is
3367 Process the @code{link} spec. This is the spec for computing the
3368 command line passed to the linker. Typically it will make use of the
3369 @samp{%L %G %S %D and %E} sequences.
3372 Dump out a @samp{-L} option for each directory that GCC believes might
3373 contain startup files. If the target supports multilibs then the
3374 current multilib directory will be prepended to each of these paths.
3377 Process the @code{lib} spec. This is a spec string for deciding which
3378 libraries should be included on the command line to the linker.
3381 Process the @code{libgcc} spec. This is a spec string for deciding
3382 which GCC support library should be included on the command line to the linker.
3385 Process the @code{startfile} spec. This is a spec for deciding which
3386 object files should be the first ones passed to the linker. Typically
3387 this might be a file named @file{crt0.o}.
3390 Process the @code{endfile} spec. This is a spec string that specifies
3391 the last object files that will be passed to the linker.
3394 Process the @code{cpp} spec. This is used to construct the arguments
3395 to be passed to the C preprocessor.
3398 Process the @code{signed_char} spec. This is intended to be used
3399 to tell cpp whether a char is signed. It typically has the definition:
3401 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
3405 Process the @code{cc1} spec. This is used to construct the options to be
3406 passed to the actual C compiler (@samp{cc1}).
3409 Process the @code{cc1plus} spec. This is used to construct the options to be
3410 passed to the actual C++ compiler (@samp{cc1plus}).
3413 Substitute the variable part of a matched option. See below.
3414 Note that each comma in the substituted string is replaced by
3418 Substitutes the @code{-S} switch, if that switch was given to GCC.
3419 If that switch was not specified, this substitutes nothing. Note that
3420 the leading dash is omitted when specifying this option, and it is
3421 automatically inserted if the substitution is performed. Thus the spec
3422 string @samp{%@{foo@}} would match the command-line option @samp{-foo}
3423 and would output the command line option @samp{-foo}.
3425 @item %W@{@code{S}@}
3426 Like %@{@code{S}@} but mark last argument supplied within as a file to be
3429 @item %@{@code{S}*@}
3430 Substitutes all the switches specified to GCC whose names start
3431 with @code{-S}, but which also take an argument. This is used for
3432 switches like @samp{-o, -D, -I}, etc. GCC considers @samp{-o foo} as being
3433 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
3434 text, including the space. Thus two arguments would be generated.
3436 @item %@{^@code{S}*@}
3437 Like %@{@code{S}*@}, but don't put a blank between a switch and its
3438 argument. Thus %@{^o*@} would only generate one argument, not two.
3440 @item %@{@code{S}*:@code{X}@}
3441 Substitutes @code{X} if one or more switches whose names start with
3442 @code{-S} are specified to GCC. Note that the tail part of the
3443 @code{-S} option (i.e. the part matched by the @samp{*}) will be substituted
3444 for each occurrence of @samp{%*} within @code{X}.
3446 @item %@{@code{S}:@code{X}@}
3447 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC.
3449 @item %@{!@code{S}:@code{X}@}
3450 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC.
3452 @item %@{|@code{S}:@code{X}@}
3453 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
3455 @item %@{|!@code{S}:@code{X}@}
3456 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
3458 @item %@{.@code{S}:@code{X}@}
3459 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
3461 @item %@{!.@code{S}:@code{X}@}
3462 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
3464 @item %@{@code{S}|@code{P}:@code{X}@}
3465 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC. This may be
3466 combined with @samp{!} and @samp{.} sequences as well, although they
3467 have a stronger binding than the @samp{|}. For example a spec string
3471 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
3474 will output the following command-line options from the following input
3475 command-line options:
3480 -d fred.c -foo -baz -boggle
3481 -d jim.d -bar -baz -boggle
3486 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
3487 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
3488 or spaces, or even newlines. They are processed as usual, as described
3491 The @samp{-O, -f, -m, and -W} switches are handled specifically in these
3492 constructs. If another value of @samp{-O} or the negated form of a @samp{-f, -m, or
3493 -W} switch is found later in the command line, the earlier switch
3494 value is ignored, except with @{@code{S}*@} where @code{S} is just one
3495 letter, which passes all matching options.
3497 The character @samp{|} at the beginning of the predicate text is used to indicate
3498 that a command should be piped to the following command, but only if @samp{-pipe}
3501 It is built into GCC which switches take arguments and which do not.
3502 (You might think it would be useful to generalize this to allow each
3503 compiler's spec to say which switches take arguments. But this cannot
3504 be done in a consistent fashion. GCC cannot even decide which input
3505 files have been specified without knowing which switches take arguments,
3506 and it must know which input files to compile in order to tell which
3509 GCC also knows implicitly that arguments starting in @samp{-l} are to be
3510 treated as compiler output files, and passed to the linker in their
3511 proper position among the other output files.
3513 @node Target Options
3514 @section Specifying Target Machine and Compiler Version
3515 @cindex target options
3516 @cindex cross compiling
3517 @cindex specifying machine version
3518 @cindex specifying compiler version and target machine
3519 @cindex compiler version, specifying
3520 @cindex target machine, specifying
3522 By default, GCC compiles code for the same type of machine that you
3523 are using. However, it can also be installed as a cross-compiler, to
3524 compile for some other type of machine. In fact, several different
3525 configurations of GCC, for different target machines, can be
3526 installed side by side. Then you specify which one to use with the
3529 In addition, older and newer versions of GCC can be installed side
3530 by side. One of them (probably the newest) will be the default, but
3531 you may sometimes wish to use another.
3534 @item -b @var{machine}
3535 The argument @var{machine} specifies the target machine for compilation.
3536 This is useful when you have installed GCC as a cross-compiler.
3538 The value to use for @var{machine} is the same as was specified as the
3539 machine type when configuring GCC as a cross-compiler. For
3540 example, if a cross-compiler was configured with @samp{configure
3541 i386v}, meaning to compile for an 80386 running System V, then you
3542 would specify @samp{-b i386v} to run that cross compiler.
3544 When you do not specify @samp{-b}, it normally means to compile for
3545 the same type of machine that you are using.
3547 @item -V @var{version}
3548 The argument @var{version} specifies which version of GCC to run.
3549 This is useful when multiple versions are installed. For example,
3550 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
3552 The default version, when you do not specify @samp{-V}, is the last
3553 version of GCC that you installed.
3556 The @samp{-b} and @samp{-V} options actually work by controlling part of
3557 the file name used for the executable files and libraries used for
3558 compilation. A given version of GCC, for a given target machine, is
3559 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.@refill
3561 Thus, sites can customize the effect of @samp{-b} or @samp{-V} either by
3562 changing the names of these directories or adding alternate names (or
3563 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
3564 file @file{80386} is a link to the file @file{i386v}, then @samp{-b
3565 80386} becomes an alias for @samp{-b i386v}.
3567 In one respect, the @samp{-b} or @samp{-V} do not completely change
3568 to a different compiler: the top-level driver program @code{gcc}
3569 that you originally invoked continues to run and invoke the other
3570 executables (preprocessor, compiler per se, assembler and linker)
3571 that do the real work. However, since no real work is done in the
3572 driver program, it usually does not matter that the driver program
3573 in use is not the one for the specified target and version.
3575 The only way that the driver program depends on the target machine is
3576 in the parsing and handling of special machine-specific options.
3577 However, this is controlled by a file which is found, along with the
3578 other executables, in the directory for the specified version and
3579 target machine. As a result, a single installed driver program adapts
3580 to any specified target machine and compiler version.
3582 The driver program executable does control one significant thing,
3583 however: the default version and target machine. Therefore, you can
3584 install different instances of the driver program, compiled for
3585 different targets or versions, under different names.
3587 For example, if the driver for version 2.0 is installed as @code{ogcc}
3588 and that for version 2.1 is installed as @code{gcc}, then the command
3589 @code{gcc} will use version 2.1 by default, while @code{ogcc} will use
3590 2.0 by default. However, you can choose either version with either
3591 command with the @samp{-V} option.
3593 @node Submodel Options
3594 @section Hardware Models and Configurations
3595 @cindex submodel options
3596 @cindex specifying hardware config
3597 @cindex hardware models and configurations, specifying
3598 @cindex machine dependent options
3600 Earlier we discussed the standard option @samp{-b} which chooses among
3601 different installed compilers for completely different target
3602 machines, such as Vax vs. 68000 vs. 80386.
3604 In addition, each of these target machine types can have its own
3605 special options, starting with @samp{-m}, to choose among various
3606 hardware models or configurations---for example, 68010 vs 68020,
3607 floating coprocessor or none. A single installed version of the
3608 compiler can compile for any model or configuration, according to the
3611 Some configurations of the compiler also support additional special
3612 options, usually for compatibility with other compilers on the same
3616 These options are defined by the macro @code{TARGET_SWITCHES} in the
3617 machine description. The default for the options is also defined by
3618 that macro, which enables you to change the defaults.
3633 * RS/6000 and PowerPC Options::
3638 * Intel 960 Options::
3639 * DEC Alpha Options::
3643 * System V Options::
3644 * TMS320C3x/C4x Options::
3650 @node M680x0 Options
3651 @subsection M680x0 Options
3652 @cindex M680x0 options
3654 These are the @samp{-m} options defined for the 68000 series. The default
3655 values for these options depends on which style of 68000 was selected when
3656 the compiler was configured; the defaults for the most common choices are
3662 Generate output for a 68000. This is the default
3663 when the compiler is configured for 68000-based systems.
3665 Use this option for microcontrollers with a 68000 or EC000 core,
3666 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
3670 Generate output for a 68020. This is the default
3671 when the compiler is configured for 68020-based systems.
3674 Generate output containing 68881 instructions for floating point.
3675 This is the default for most 68020 systems unless @samp{-nfp} was
3676 specified when the compiler was configured.
3679 Generate output for a 68030. This is the default when the compiler is
3680 configured for 68030-based systems.
3683 Generate output for a 68040. This is the default when the compiler is
3684 configured for 68040-based systems.
3686 This option inhibits the use of 68881/68882 instructions that have to be
3687 emulated by software on the 68040. Use this option if your 68040 does not
3688 have code to emulate those instructions.
3691 Generate output for a 68060. This is the default when the compiler is
3692 configured for 68060-based systems.
3694 This option inhibits the use of 68020 and 68881/68882 instructions that
3695 have to be emulated by software on the 68060. Use this option if your 68060
3696 does not have code to emulate those instructions.
3699 Generate output for a CPU32. This is the default
3700 when the compiler is configured for CPU32-based systems.
3702 Use this option for microcontrollers with a
3703 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
3704 68336, 68340, 68341, 68349 and 68360.
3707 Generate output for a 520X "coldfire" family cpu. This is the default
3708 when the compiler is configured for 520X-based systems.
3710 Use this option for microcontroller with a 5200 core, including
3711 the MCF5202, MCF5203, MCF5204 and MCF5202.
3715 Generate output for a 68040, without using any of the new instructions.
3716 This results in code which can run relatively efficiently on either a
3717 68020/68881 or a 68030 or a 68040. The generated code does use the
3718 68881 instructions that are emulated on the 68040.
3721 Generate output for a 68060, without using any of the new instructions.
3722 This results in code which can run relatively efficiently on either a
3723 68020/68881 or a 68030 or a 68040. The generated code does use the
3724 68881 instructions that are emulated on the 68060.
3727 Generate output containing Sun FPA instructions for floating point.
3730 Generate output containing library calls for floating point.
3731 @strong{Warning:} the requisite libraries are not available for all m68k
3732 targets. Normally the facilities of the machine's usual C compiler are
3733 used, but this can't be done directly in cross-compilation. You must
3734 make your own arrangements to provide suitable library functions for
3735 cross-compilation. The embedded targets @samp{m68k-*-aout} and
3736 @samp{m68k-*-coff} do provide software floating point support.
3739 Consider type @code{int} to be 16 bits wide, like @code{short int}.
3742 Do not use the bit-field instructions. The @samp{-m68000}, @samp{-mcpu32}
3743 and @samp{-m5200} options imply @w{@samp{-mnobitfield}}.
3746 Do use the bit-field instructions. The @samp{-m68020} option implies
3747 @samp{-mbitfield}. This is the default if you use a configuration
3748 designed for a 68020.
3751 Use a different function-calling convention, in which functions
3752 that take a fixed number of arguments return with the @code{rtd}
3753 instruction, which pops their arguments while returning. This
3754 saves one instruction in the caller since there is no need to pop
3755 the arguments there.
3757 This calling convention is incompatible with the one normally
3758 used on Unix, so you cannot use it if you need to call libraries
3759 compiled with the Unix compiler.
3761 Also, you must provide function prototypes for all functions that
3762 take variable numbers of arguments (including @code{printf});
3763 otherwise incorrect code will be generated for calls to those
3766 In addition, seriously incorrect code will result if you call a
3767 function with too many arguments. (Normally, extra arguments are
3768 harmlessly ignored.)
3770 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
3771 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
3774 @itemx -mno-align-int
3775 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
3776 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
3777 boundary (@samp{-malign-int}) or a 16-bit boundary (@samp{-mno-align-int}).
3778 Aligning variables on 32-bit boundaries produces code that runs somewhat
3779 faster on processors with 32-bit busses at the expense of more memory.
3781 @strong{Warning:} if you use the @samp{-malign-int} switch, GCC will
3782 align structures containing the above types differently than
3783 most published application binary interface specifications for the m68k.
3786 Use the pc-relative addressing mode of the 68000 directly, instead of
3787 using a global offset table. At present, this option implies -fpic,
3788 allowing at most a 16-bit offset for pc-relative addressing. -fPIC is
3789 not presently supported with -mpcrel, though this could be supported for
3790 68020 and higher processors.
3795 @subsection VAX Options
3798 These @samp{-m} options are defined for the Vax:
3802 Do not output certain jump instructions (@code{aobleq} and so on)
3803 that the Unix assembler for the Vax cannot handle across long
3807 Do output those jump instructions, on the assumption that you
3808 will assemble with the GNU assembler.
3811 Output code for g-format floating point numbers instead of d-format.
3815 @subsection SPARC Options
3816 @cindex SPARC options
3818 These @samp{-m} switches are supported on the SPARC:
3823 Specify @samp{-mapp-regs} to generate output using the global registers
3824 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
3827 To be fully SVR4 ABI compliant at the cost of some performance loss,
3828 specify @samp{-mno-app-regs}. You should compile libraries and system
3829 software with this option.
3833 Generate output containing floating point instructions. This is the
3838 Generate output containing library calls for floating point.
3839 @strong{Warning:} the requisite libraries are not available for all SPARC
3840 targets. Normally the facilities of the machine's usual C compiler are
3841 used, but this cannot be done directly in cross-compilation. You must make
3842 your own arrangements to provide suitable library functions for
3843 cross-compilation. The embedded targets @samp{sparc-*-aout} and
3844 @samp{sparclite-*-*} do provide software floating point support.
3846 @samp{-msoft-float} changes the calling convention in the output file;
3847 therefore, it is only useful if you compile @emph{all} of a program with
3848 this option. In particular, you need to compile @file{libgcc.a}, the
3849 library that comes with GCC, with @samp{-msoft-float} in order for
3852 @item -mhard-quad-float
3853 Generate output containing quad-word (long double) floating point
3856 @item -msoft-quad-float
3857 Generate output containing library calls for quad-word (long double)
3858 floating point instructions. The functions called are those specified
3859 in the SPARC ABI. This is the default.
3861 As of this writing, there are no sparc implementations that have hardware
3862 support for the quad-word floating point instructions. They all invoke
3863 a trap handler for one of these instructions, and then the trap handler
3864 emulates the effect of the instruction. Because of the trap handler overhead,
3865 this is much slower than calling the ABI library routines. Thus the
3866 @samp{-msoft-quad-float} option is the default.
3870 With @samp{-mepilogue} (the default), the compiler always emits code for
3871 function exit at the end of each function. Any function exit in
3872 the middle of the function (such as a return statement in C) will
3873 generate a jump to the exit code at the end of the function.
3875 With @samp{-mno-epilogue}, the compiler tries to emit exit code inline
3876 at every function exit.
3880 With @samp{-mflat}, the compiler does not generate save/restore instructions
3881 and will use a "flat" or single register window calling convention.
3882 This model uses %i7 as the frame pointer and is compatible with the normal
3883 register window model. Code from either may be intermixed.
3884 The local registers and the input registers (0-5) are still treated as
3885 "call saved" registers and will be saved on the stack as necessary.
3887 With @samp{-mno-flat} (the default), the compiler emits save/restore
3888 instructions (except for leaf functions) and is the normal mode of operation.
3890 @item -mno-unaligned-doubles
3891 @itemx -munaligned-doubles
3892 Assume that doubles have 8 byte alignment. This is the default.
3894 With @samp{-munaligned-doubles}, GCC assumes that doubles have 8 byte
3895 alignment only if they are contained in another type, or if they have an
3896 absolute address. Otherwise, it assumes they have 4 byte alignment.
3897 Specifying this option avoids some rare compatibility problems with code
3898 generated by other compilers. It is not the default because it results
3899 in a performance loss, especially for floating point code.
3903 These two options select variations on the SPARC architecture.
3905 By default (unless specifically configured for the Fujitsu SPARClite),
3906 GCC generates code for the v7 variant of the SPARC architecture.
3908 @samp{-mv8} will give you SPARC v8 code. The only difference from v7
3909 code is that the compiler emits the integer multiply and integer
3910 divide instructions which exist in SPARC v8 but not in SPARC v7.
3912 @samp{-msparclite} will give you SPARClite code. This adds the integer
3913 multiply, integer divide step and scan (@code{ffs}) instructions which
3914 exist in SPARClite but not in SPARC v7.
3916 These options are deprecated and will be deleted in a future GCC release.
3917 They have been replaced with @samp{-mcpu=xxx}.
3921 These two options select the processor for which the code is optimised.
3923 With @samp{-mcypress} (the default), the compiler optimizes code for the
3924 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
3925 This is also appropriate for the older SparcStation 1, 2, IPX etc.
3927 With @samp{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
3928 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
3929 of the full SPARC v8 instruction set.
3931 These options are deprecated and will be deleted in a future GCC release.
3932 They have been replaced with @samp{-mcpu=xxx}.
3934 @item -mcpu=@var{cpu_type}
3935 Set the instruction set, register set, and instruction scheduling parameters
3936 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
3937 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
3938 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
3939 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
3941 Default instruction scheduling parameters are used for values that select
3942 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
3943 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
3945 Here is a list of each supported architecture and their supported
3950 v8: supersparc, hypersparc
3951 sparclite: f930, f934, sparclite86x
3956 @item -mtune=@var{cpu_type}
3957 Set the instruction scheduling parameters for machine type
3958 @var{cpu_type}, but do not set the instruction set or register set that the
3959 option @samp{-mcpu=}@var{cpu_type} would.
3961 The same values for @samp{-mcpu=}@var{cpu_type} are used for
3962 @samp{-mtune=}@*@var{cpu_type}, though the only useful values are those that
3963 select a particular cpu implementation: @samp{cypress}, @samp{supersparc},
3964 @samp{hypersparc}, @samp{f930}, @samp{f934}, @samp{sparclite86x},
3965 @samp{tsc701}, @samp{ultrasparc}.
3969 These @samp{-m} switches are supported in addition to the above
3970 on the SPARCLET processor.
3973 @item -mlittle-endian
3974 Generate code for a processor running in little-endian mode.
3977 Treat register @code{%g0} as a normal register.
3978 GCC will continue to clobber it as necessary but will not assume
3979 it always reads as 0.
3981 @item -mbroken-saverestore
3982 Generate code that does not use non-trivial forms of the @code{save} and
3983 @code{restore} instructions. Early versions of the SPARCLET processor do
3984 not correctly handle @code{save} and @code{restore} instructions used with
3985 arguments. They correctly handle them used without arguments. A @code{save}
3986 instruction used without arguments increments the current window pointer
3987 but does not allocate a new stack frame. It is assumed that the window
3988 overflow trap handler will properly handle this case as will interrupt
3992 These @samp{-m} switches are supported in addition to the above
3993 on SPARC V9 processors in 64 bit environments.
3996 @item -mlittle-endian
3997 Generate code for a processor running in little-endian mode.
4001 Generate code for a 32 bit or 64 bit environment.
4002 The 32 bit environment sets int, long and pointer to 32 bits.
4003 The 64 bit environment sets int to 32 bits and long and pointer
4006 @item -mcmodel=medlow
4007 Generate code for the Medium/Low code model: the program must be linked
4008 in the low 32 bits of the address space. Pointers are 64 bits.
4009 Programs can be statically or dynamically linked.
4011 @item -mcmodel=medmid
4012 Generate code for the Medium/Middle code model: the program must be linked
4013 in the low 44 bits of the address space, the text segment must be less than
4014 2G bytes, and data segment must be within 2G of the text segment.
4015 Pointers are 64 bits.
4017 @item -mcmodel=medany
4018 Generate code for the Medium/Anywhere code model: the program may be linked
4019 anywhere in the address space, the text segment must be less than
4020 2G bytes, and data segment must be within 2G of the text segment.
4021 Pointers are 64 bits.
4023 @item -mcmodel=embmedany
4024 Generate code for the Medium/Anywhere code model for embedded systems:
4025 assume a 32 bit text and a 32 bit data segment, both starting anywhere
4026 (determined at link time). Register %g4 points to the base of the
4027 data segment. Pointers still 64 bits.
4028 Programs are statically linked, PIC is not supported.
4031 @itemx -mno-stack-bias
4032 With @samp{-mstack-bias}, GCC assumes that the stack pointer, and
4033 frame pointer if present, are offset by -2047 which must be added back
4034 when making stack frame references.
4035 Otherwise, assume no such offset is present.
4038 @node Convex Options
4039 @subsection Convex Options
4040 @cindex Convex options
4042 These @samp{-m} options are defined for Convex:
4046 Generate output for C1. The code will run on any Convex machine.
4047 The preprocessor symbol @code{__convex__c1__} is defined.
4050 Generate output for C2. Uses instructions not available on C1.
4051 Scheduling and other optimizations are chosen for max performance on C2.
4052 The preprocessor symbol @code{__convex_c2__} is defined.
4055 Generate output for C32xx. Uses instructions not available on C1.
4056 Scheduling and other optimizations are chosen for max performance on C32.
4057 The preprocessor symbol @code{__convex_c32__} is defined.
4060 Generate output for C34xx. Uses instructions not available on C1.
4061 Scheduling and other optimizations are chosen for max performance on C34.
4062 The preprocessor symbol @code{__convex_c34__} is defined.
4065 Generate output for C38xx. Uses instructions not available on C1.
4066 Scheduling and other optimizations are chosen for max performance on C38.
4067 The preprocessor symbol @code{__convex_c38__} is defined.
4070 Generate code which puts an argument count in the word preceding each
4071 argument list. This is compatible with regular CC, and a few programs
4072 may need the argument count word. GDB and other source-level debuggers
4073 do not need it; this info is in the symbol table.
4076 Omit the argument count word. This is the default.
4078 @item -mvolatile-cache
4079 Allow volatile references to be cached. This is the default.
4081 @item -mvolatile-nocache
4082 Volatile references bypass the data cache, going all the way to memory.
4083 This is only needed for multi-processor code that does not use standard
4084 synchronization instructions. Making non-volatile references to volatile
4085 locations will not necessarily work.
4088 Type long is 32 bits, the same as type int. This is the default.
4091 Type long is 64 bits, the same as type long long. This option is useless,
4092 because no library support exists for it.
4095 @node AMD29K Options
4096 @subsection AMD29K Options
4097 @cindex AMD29K options
4099 These @samp{-m} options are defined for the AMD Am29000:
4104 @cindex DW bit (29k)
4105 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
4106 halfword operations are directly supported by the hardware. This is the
4111 Generate code that assumes the @code{DW} bit is not set.
4115 @cindex byte writes (29k)
4116 Generate code that assumes the system supports byte and halfword write
4117 operations. This is the default.
4121 Generate code that assumes the systems does not support byte and
4122 halfword write operations. @samp{-mnbw} implies @samp{-mndw}.
4126 @cindex memory model (29k)
4127 Use a small memory model that assumes that all function addresses are
4128 either within a single 256 KB segment or at an absolute address of less
4129 than 256k. This allows the @code{call} instruction to be used instead
4130 of a @code{const}, @code{consth}, @code{calli} sequence.
4134 Use the normal memory model: Generate @code{call} instructions only when
4135 calling functions in the same file and @code{calli} instructions
4136 otherwise. This works if each file occupies less than 256 KB but allows
4137 the entire executable to be larger than 256 KB. This is the default.
4140 Always use @code{calli} instructions. Specify this option if you expect
4141 a single file to compile into more than 256 KB of code.
4145 @cindex processor selection (29k)
4146 Generate code for the Am29050.
4150 Generate code for the Am29000. This is the default.
4152 @item -mkernel-registers
4153 @kindex -mkernel-registers
4154 @cindex kernel and user registers (29k)
4155 Generate references to registers @code{gr64-gr95} instead of to
4156 registers @code{gr96-gr127}. This option can be used when compiling
4157 kernel code that wants a set of global registers disjoint from that used
4160 Note that when this option is used, register names in @samp{-f} flags
4161 must use the normal, user-mode, names.
4163 @item -muser-registers
4164 @kindex -muser-registers
4165 Use the normal set of global registers, @code{gr96-gr127}. This is the
4169 @itemx -mno-stack-check
4170 @kindex -mstack-check
4171 @cindex stack checks (29k)
4172 Insert (or do not insert) a call to @code{__msp_check} after each stack
4173 adjustment. This is often used for kernel code.
4176 @itemx -mno-storem-bug
4177 @kindex -mstorem-bug
4178 @cindex storem bug (29k)
4179 @samp{-mstorem-bug} handles 29k processors which cannot handle the
4180 separation of a mtsrim insn and a storem instruction (most 29000 chips
4181 to date, but not the 29050).
4183 @item -mno-reuse-arg-regs
4184 @itemx -mreuse-arg-regs
4185 @kindex -mreuse-arg-regs
4186 @samp{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
4187 registers for copying out arguments. This helps detect calling a function
4188 with fewer arguments than it was declared with.
4190 @item -mno-impure-text
4191 @itemx -mimpure-text
4192 @kindex -mimpure-text
4193 @samp{-mimpure-text}, used in addition to @samp{-shared}, tells the compiler to
4194 not pass @samp{-assert pure-text} to the linker when linking a shared object.
4197 @kindex -msoft-float
4198 Generate output containing library calls for floating point.
4199 @strong{Warning:} the requisite libraries are not part of GCC.
4200 Normally the facilities of the machine's usual C compiler are used, but
4201 this can't be done directly in cross-compilation. You must make your
4202 own arrangements to provide suitable library functions for
4207 Do not generate multm or multmu instructions. This is useful for some embedded
4208 systems which do not have trap handlers for these instructions.
4212 @subsection ARM Options
4215 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
4220 @kindex -mapcs-frame
4221 Generate a stack frame that is compliant with the ARM Procedure Call
4222 Standard for all functions, even if this is not strictly necessary for
4223 correct execution of the code. Specifying @samp{-fomit-frame-pointer}
4224 with this option will cause the stack frames not to be generated for
4225 leaf functions. The default is @samp{-mno-apcs-frame}.
4229 This is a synonym for @samp{-mapcs-frame}.
4233 Generate code for a processor running with a 26-bit program counter,
4234 and conforming to the function calling standards for the APCS 26-bit
4235 option. This option replaces the @samp{-m2} and @samp{-m3} options
4236 of previous releases of the compiler.
4240 Generate code for a processor running with a 32-bit program counter,
4241 and conforming to the function calling standards for the APCS 32-bit
4242 option. This option replaces the @samp{-m6} option of previous releases
4245 @item -mapcs-stack-check
4246 @kindex -mapcs-stack-check
4247 @kindex -mno-apcs-stack-check
4248 Generate code to check the amount of stack space available upon entry to
4249 every function (that actually uses some stack space). If there is
4250 insufficient space available then either the function
4251 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
4252 called, depending upon the amount of stack space required. The run time
4253 system is required to provide these functions. The default is
4254 @samp{-mno-apcs-stack-check}, since this produces smaller code.
4257 @kindex -mapcs-float
4258 @kindex -mno-apcs-float
4259 Pass floating point arguments using the float point registers. This is
4260 one of the variants of the APCS. This option is recommended if the
4261 target hardware has a floating point unit or if a lot of floating point
4262 arithmetic is going to be performed by the code. The default is
4263 @samp{-mno-apcs-float}, since integer only code is slightly increased in
4264 size if @samp{-mapcs-float} is used.
4266 @item -mapcs-reentrant
4267 @kindex -mapcs-reentrant
4268 @kindex -mno-apcs-reentrant
4269 Generate reentrant, position independent code. This is the equivalent
4270 to specifying the @samp{-fpic} option. The default is
4271 @samp{-mno-apcs-reentrant}.
4273 @item -mthumb-interwork
4274 @kindex -mthumb-interwork
4275 @kindex -mno-thumb-interwork
4276 Generate code which supports calling between the ARM and THUMB
4277 instruction sets. Without this option the two instruction sets cannot
4278 be reliably used inside one program. The default is
4279 @samp{-mno-thumb-interwork}, since slightly larger code is generated
4280 when @samp{-mthumb-interwork} is specified.
4282 @item -mno-sched-prolog
4283 @kindex -mno-sched-prolog
4284 @kindex -msched-prolog
4285 Prevent the reordering of instructions in the function prolog, or the
4286 merging of those instruction with the instructions in the function's
4287 body. This means that all functions will start with a recognizable set
4288 of instructions (or in fact one of a choice from a small set of
4289 different function prologues), and this information can be used to
4290 locate the start if functions inside an executable piece of code. The
4291 default is @samp{-msched-prolog}.
4294 Generate output containing floating point instructions. This is the
4298 Generate output containing library calls for floating point.
4299 @strong{Warning:} the requisite libraries are not available for all ARM
4300 targets. Normally the facilities of the machine's usual C compiler are
4301 used, but this cannot be done directly in cross-compilation. You must make
4302 your own arrangements to provide suitable library functions for
4305 @samp{-msoft-float} changes the calling convention in the output file;
4306 therefore, it is only useful if you compile @emph{all} of a program with
4307 this option. In particular, you need to compile @file{libgcc.a}, the
4308 library that comes with GCC, with @samp{-msoft-float} in order for
4311 @item -mlittle-endian
4312 Generate code for a processor running in little-endian mode. This is
4313 the default for all standard configurations.
4316 Generate code for a processor running in big-endian mode; the default is
4317 to compile code for a little-endian processor.
4319 @item -mwords-little-endian
4320 This option only applies when generating code for big-endian processors.
4321 Generate code for a little-endian word order but a big-endian byte
4322 order. That is, a byte order of the form @samp{32107654}. Note: this
4323 option should only be used if you require compatibility with code for
4324 big-endian ARM processors generated by versions of the compiler prior to
4327 @item -mshort-load-bytes
4328 @kindex -mshort-load-bytes
4329 Do not try to load half-words (eg @samp{short}s) by loading a word from
4330 an unaligned address. For some targets the MMU is configured to trap
4331 unaligned loads; use this option to generate code that is safe in these
4334 @item -mno-short-load-bytes
4335 @kindex -mno-short-load-bytes
4336 Use unaligned word loads to load half-words (eg @samp{short}s). This
4337 option produces more efficient code, but the MMU is sometimes configured
4338 to trap these instructions.
4340 @item -mshort-load-words
4341 @kindex -mshort-load-words
4342 This is a synonym for @samp{-mno-short-load-bytes}.
4344 @item -mno-short-load-words
4345 @kindex -mno-short-load-words
4346 This is a synonym for @samp{-mshort-load-bytes}.
4350 This option only applies to RISC iX. Emulate the native BSD-mode
4351 compiler. This is the default if @samp{-ansi} is not specified.
4355 This option only applies to RISC iX. Emulate the native X/Open-mode
4358 @item -mno-symrename
4359 @kindex -mno-symrename
4360 This option only applies to RISC iX. Do not run the assembler
4361 post-processor, @samp{symrename}, after code has been assembled.
4362 Normally it is necessary to modify some of the standard symbols in
4363 preparation for linking with the RISC iX C library; this option
4364 suppresses this pass. The post-processor is never run when the
4365 compiler is built for cross-compilation.
4369 This specifies the name of the target ARM processor. GCC uses this name
4370 to determine what kind of instructions it can use when generating
4371 assembly code. Permissible names are: arm2, arm250, arm3, arm6, arm60,
4372 arm600, arm610, arm620, arm7, arm7m, arm7d, arm7dm, arm7di, arm7dmi,
4373 arm70, arm700, arm700i, arm710, arm710c, arm7100, arm7500, arm7500fe,
4374 arm7tdmi, arm8, strongarm, strongarm110, strongarm1100, arm8, arm810,
4375 arm9, arm920, arm920t, arm9tdmi.
4377 @itemx -mtune=<name>
4379 This option is very similar to the @samp{-mcpu=} option, except that
4380 instead of specifying the actual target processor type, and hence
4381 restricting which instructions can be used, it specifies that GCC should
4382 tune the performance of the code as if the target were of the type
4383 specified in this option, but still choosing the instructions that it
4384 will generate based on the cpu specified by a @samp{-mcpu=} option.
4385 For some arm implementations better performance can be obtained by using
4390 This specifies the name of the target ARM architecture. GCC uses this
4391 name to determine what kind of instructions it can use when generating
4392 assembly code. This option can be used in conjunction with or instead
4393 of the @samp{-mcpu=} option. Permissible names are: armv2, armv2a,
4394 armv3, armv3m, armv4, armv4t, armv5.
4396 @item -mfpe=<number>
4397 @itemx -mfp=<number>
4400 This specifes the version of the floating point emulation available on
4401 the target. Permissible values are 2 and 3. @samp{-mfp=} is a synonym
4402 for @samp{-mfpe=} to support older versions of GCC.
4404 @item -mstructure-size-boundary=<n>
4405 @kindex -mstructure-size-boundary
4406 The size of all structures and unions will be rounded up to a multiple
4407 of the number of bits set by this option. Permissible values are 8 and
4408 32. The default value varies for different toolchains. For the COFF
4409 targeted toolchain the default value is 8. Specifying the larger number
4410 can produce faster, more efficient code, but can also increase the size
4411 of the program. The two values are potentially incompatible. Code
4412 compiled with one value cannot necessarily expect to work with code or
4413 libraries compiled with the other value, if they exchange information
4414 using structures or unions. Programmers are encouraged to use the 32
4415 value as future versions of the toolchain may default to this value.
4417 @item -mabort-on-noreturn
4418 @kindex -mabort-on-noreturn
4419 @kindex -mnoabort-on-noreturn
4420 Generate a call to the function abort at the end of a noreturn function.
4421 It will be executed if the function tries to return.
4423 @item -mnop-fun-dllimport
4424 @kindex -mnop-fun-dllimport
4425 Disable the support for the @emph{dllimport} attribute.
4427 @item -msingle-pic-base
4428 @kindex -msingle-pic-base
4429 Treat the register used for PIC addressing as read-only, rather than
4430 loading it in the prologue for each function. The run-time system is
4431 responsible for initialising this register with an appropriate value
4432 before execution begins.
4434 @item -mpic-register=<reg>
4435 @kindex -mpic-register=
4436 Specify the register to be used for PIC addressing. The default is R10
4437 unless stack-checking is enabled, when R9 is used.
4442 @subsection Thumb Options
4443 @cindex Thumb Options
4447 @item -mthumb-interwork
4448 @kindex -mthumb-interwork
4449 @kindex -mno-thumb-interwork
4450 Generate code which supports calling between the THUMB and ARM
4451 instruction sets. Without this option the two instruction sets cannot
4452 be reliably used inside one program. The default is
4453 @samp{-mno-thumb-interwork}, since slightly smaller code is generated
4457 @kindex -mtpcs-frame
4458 @kindex -mno-tpcs-frame
4459 Generate a stack frame that is compliant with the Thumb Procedure Call
4460 Standard for all non-leaf functions. (A leaf function is one that does
4461 not call any other functions). The default is @samp{-mno-apcs-frame}.
4463 @item -mtpcs-leaf-frame
4464 @kindex -mtpcs-leaf-frame
4465 @kindex -mno-tpcs-leaf-frame
4466 Generate a stack frame that is compliant with the Thumb Procedure Call
4467 Standard for all leaf functions. (A leaf function is one that does
4468 not call any other functions). The default is @samp{-mno-apcs-leaf-frame}.
4470 @item -mlittle-endian
4471 @kindex -mlittle-endian
4472 Generate code for a processor running in little-endian mode. This is
4473 the default for all standard configurations.
4476 @kindex -mbig-endian
4477 Generate code for a processor running in big-endian mode.
4479 @item -mstructure-size-boundary=<n>
4480 @kindex -mstructure-size-boundary
4481 The size of all structures and unions will be rounded up to a multiple
4482 of the number of bits set by this option. Permissible values are 8 and
4483 32. The default value varies for different toolchains. For the COFF
4484 targeted toolchain the default value is 8. Specifying the larger number
4485 can produced faster, more efficient code, but can also increase the size
4486 of the program. The two values are potentially incompatible. Code
4487 compiled with one value cannot necessarily expect to work with code or
4488 libraries compiled with the other value, if they exchange information
4489 using structures or unions. Programmers are encouraged to use the 32
4490 value as future versions of the toolchain may default to this value.
4492 @item -mnop-fun-dllimport
4493 @kindex -mnop-fun-dllimport
4494 Disable the support for the @emph{dllimport} attribute.
4496 @item -mcallee-super-interworking
4497 @kindex -mcallee-super-interworking
4498 Gives all externally visible functions in the file being compiled an ARM
4499 instruction set header which switches to Thumb mode before executing the
4500 rest of the function. This allows these functions to be called from
4501 non-interworking code.
4503 @item -mcaller-super-interworking
4504 @kindex -mcaller-super-interworking
4505 Allows calls via function pointers (including virtual functions) to
4506 execute correctly regardless of whether the target code has been
4507 compiled for interworking or not. There is a small overhead in the cost
4508 of executing a function pointer if this option is enabled.
4510 @item -msingle-pic-base
4511 @kindex -msingle-pic-base
4512 Treat the register used for PIC addressing as read-only, rather than
4513 loading it in the prologue for each function. The run-time system is
4514 responsible for initialising this register with an appropriate value
4515 before execution begins.
4517 @item -mpic-register=<reg>
4518 @kindex -mpic-register=
4519 Specify the register to be used for PIC addressing. The default is R10.
4523 @node MN10200 Options
4524 @subsection MN10200 Options
4525 @cindex MN10200 options
4526 These @samp{-m} options are defined for Matsushita MN10200 architectures:
4530 Indicate to the linker that it should perform a relaxation optimization pass
4531 to shorten branches, calls and absolute memory addresses. This option only
4532 has an effect when used on the command line for the final link step.
4534 This option makes symbolic debugging impossible.
4537 @node MN10300 Options
4538 @subsection MN10300 Options
4539 @cindex MN10300 options
4540 These @samp{-m} options are defined for Matsushita MN10300 architectures:
4544 Generate code to avoid bugs in the multiply instructions for the MN10300
4545 processors. This is the default.
4548 Do not generate code to avoid bugs in the multiply instructions for the
4552 Indicate to the linker that it should perform a relaxation optimization pass
4553 to shorten branches, calls and absolute memory addresses. This option only
4554 has an effect when used on the command line for the final link step.
4556 This option makes symbolic debugging impossible.
4560 @node M32R/D Options
4561 @subsection M32R/D Options
4562 @cindex M32R/D options
4564 These @samp{-m} options are defined for Mitsubishi M32R/D architectures:
4567 @item -mcode-model=small
4568 Assume all objects live in the lower 16MB of memory (so that their addresses
4569 can be loaded with the @code{ld24} instruction), and assume all subroutines
4570 are reachable with the @code{bl} instruction.
4571 This is the default.
4573 The addressability of a particular object can be set with the
4574 @code{model} attribute.
4576 @item -mcode-model=medium
4577 Assume objects may be anywhere in the 32 bit address space (the compiler
4578 will generate @code{seth/add3} instructions to load their addresses), and
4579 assume all subroutines are reachable with the @code{bl} instruction.
4581 @item -mcode-model=large
4582 Assume objects may be anywhere in the 32 bit address space (the compiler
4583 will generate @code{seth/add3} instructions to load their addresses), and
4584 assume subroutines may not be reachable with the @code{bl} instruction
4585 (the compiler will generate the much slower @code{seth/add3/jl}
4586 instruction sequence).
4589 Disable use of the small data area. Variables will be put into
4590 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
4591 @code{section} attribute has been specified).
4592 This is the default.
4594 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
4595 Objects may be explicitly put in the small data area with the
4596 @code{section} attribute using one of these sections.
4599 Put small global and static data in the small data area, but do not
4600 generate special code to reference them.
4603 Put small global and static data in the small data area, and generate
4604 special instructions to reference them.
4607 @cindex smaller data references
4608 Put global and static objects less than or equal to @var{num} bytes
4609 into the small data or bss sections instead of the normal data or bss
4610 sections. The default value of @var{num} is 8.
4611 The @samp{-msdata} option must be set to one of @samp{sdata} or @samp{use}
4612 for this option to have any effect.
4614 All modules should be compiled with the same @samp{-G @var{num}} value.
4615 Compiling with different values of @var{num} may or may not work; if it
4616 doesn't the linker will give an error message - incorrect code will not be
4622 @subsection M88K Options
4623 @cindex M88k options
4625 These @samp{-m} options are defined for Motorola 88k architectures:
4630 Generate code that works well on both the m88100 and the
4635 Generate code that works best for the m88100, but that also
4640 Generate code that works best for the m88110, and may not run
4645 Obsolete option to be removed from the next revision.
4648 @item -midentify-revision
4649 @kindex -midentify-revision
4651 @cindex identifying source, compiler (88k)
4652 Include an @code{ident} directive in the assembler output recording the
4653 source file name, compiler name and version, timestamp, and compilation
4656 @item -mno-underscores
4657 @kindex -mno-underscores
4658 @cindex underscores, avoiding (88k)
4659 In assembler output, emit symbol names without adding an underscore
4660 character at the beginning of each name. The default is to use an
4661 underscore as prefix on each name.
4663 @item -mocs-debug-info
4664 @itemx -mno-ocs-debug-info
4665 @kindex -mocs-debug-info
4666 @kindex -mno-ocs-debug-info
4668 @cindex debugging, 88k OCS
4669 Include (or omit) additional debugging information (about registers used
4670 in each stack frame) as specified in the 88open Object Compatibility
4671 Standard, ``OCS''. This extra information allows debugging of code that
4672 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
4673 Delta 88 SVr3.2 is to include this information; other 88k configurations
4674 omit this information by default.
4676 @item -mocs-frame-position
4677 @kindex -mocs-frame-position
4678 @cindex register positions in frame (88k)
4679 When emitting COFF debugging information for automatic variables and
4680 parameters stored on the stack, use the offset from the canonical frame
4681 address, which is the stack pointer (register 31) on entry to the
4682 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
4683 @samp{-mocs-frame-position}; other 88k configurations have the default
4684 @samp{-mno-ocs-frame-position}.
4686 @item -mno-ocs-frame-position
4687 @kindex -mno-ocs-frame-position
4688 @cindex register positions in frame (88k)
4689 When emitting COFF debugging information for automatic variables and
4690 parameters stored on the stack, use the offset from the frame pointer
4691 register (register 30). When this option is in effect, the frame
4692 pointer is not eliminated when debugging information is selected by the
4695 @item -moptimize-arg-area
4696 @itemx -mno-optimize-arg-area
4697 @kindex -moptimize-arg-area
4698 @kindex -mno-optimize-arg-area
4699 @cindex arguments in frame (88k)
4700 Control how function arguments are stored in stack frames.
4701 @samp{-moptimize-arg-area} saves space by optimizing them, but this
4702 conflicts with the 88open specifications. The opposite alternative,
4703 @samp{-mno-optimize-arg-area}, agrees with 88open standards. By default
4704 GCC does not optimize the argument area.
4706 @item -mshort-data-@var{num}
4707 @kindex -mshort-data-@var{num}
4708 @cindex smaller data references (88k)
4709 @cindex r0-relative references (88k)
4710 Generate smaller data references by making them relative to @code{r0},
4711 which allows loading a value using a single instruction (rather than the
4712 usual two). You control which data references are affected by
4713 specifying @var{num} with this option. For example, if you specify
4714 @samp{-mshort-data-512}, then the data references affected are those
4715 involving displacements of less than 512 bytes.
4716 @samp{-mshort-data-@var{num}} is not effective for @var{num} greater
4719 @item -mserialize-volatile
4720 @kindex -mserialize-volatile
4721 @itemx -mno-serialize-volatile
4722 @kindex -mno-serialize-volatile
4723 @cindex sequential consistency on 88k
4724 Do, or don't, generate code to guarantee sequential consistency
4725 of volatile memory references. By default, consistency is
4728 The order of memory references made by the MC88110 processor does
4729 not always match the order of the instructions requesting those
4730 references. In particular, a load instruction may execute before
4731 a preceding store instruction. Such reordering violates
4732 sequential consistency of volatile memory references, when there
4733 are multiple processors. When consistency must be guaranteed,
4734 GNU C generates special instructions, as needed, to force
4735 execution in the proper order.
4737 The MC88100 processor does not reorder memory references and so
4738 always provides sequential consistency. However, by default, GNU
4739 C generates the special instructions to guarantee consistency
4740 even when you use @samp{-m88100}, so that the code may be run on an
4741 MC88110 processor. If you intend to run your code only on the
4742 MC88100 processor, you may use @samp{-mno-serialize-volatile}.
4744 The extra code generated to guarantee consistency may affect the
4745 performance of your application. If you know that you can safely
4746 forgo this guarantee, you may use @samp{-mno-serialize-volatile}.
4752 @cindex assembler syntax, 88k
4754 Turn on (@samp{-msvr4}) or off (@samp{-msvr3}) compiler extensions
4755 related to System V release 4 (SVr4). This controls the following:
4759 Which variant of the assembler syntax to emit.
4761 @samp{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
4762 that is used on System V release 4.
4764 @samp{-msvr4} makes GCC issue additional declaration directives used in
4768 @samp{-msvr4} is the default for the m88k-motorola-sysv4 and
4769 m88k-dg-dgux m88k configurations. @samp{-msvr3} is the default for all
4770 other m88k configurations.
4772 @item -mversion-03.00
4773 @kindex -mversion-03.00
4774 This option is obsolete, and is ignored.
4775 @c ??? which asm syntax better for GAS? option there too?
4777 @item -mno-check-zero-division
4778 @itemx -mcheck-zero-division
4779 @kindex -mno-check-zero-division
4780 @kindex -mcheck-zero-division
4781 @cindex zero division on 88k
4782 Do, or don't, generate code to guarantee that integer division by
4783 zero will be detected. By default, detection is guaranteed.
4785 Some models of the MC88100 processor fail to trap upon integer
4786 division by zero under certain conditions. By default, when
4787 compiling code that might be run on such a processor, GNU C
4788 generates code that explicitly checks for zero-valued divisors
4789 and traps with exception number 503 when one is detected. Use of
4790 mno-check-zero-division suppresses such checking for code
4791 generated to run on an MC88100 processor.
4793 GNU C assumes that the MC88110 processor correctly detects all
4794 instances of integer division by zero. When @samp{-m88110} is
4795 specified, both @samp{-mcheck-zero-division} and
4796 @samp{-mno-check-zero-division} are ignored, and no explicit checks for
4797 zero-valued divisors are generated.
4799 @item -muse-div-instruction
4800 @kindex -muse-div-instruction
4801 @cindex divide instruction, 88k
4802 Use the div instruction for signed integer division on the
4803 MC88100 processor. By default, the div instruction is not used.
4805 On the MC88100 processor the signed integer division instruction
4806 div) traps to the operating system on a negative operand. The
4807 operating system transparently completes the operation, but at a
4808 large cost in execution time. By default, when compiling code
4809 that might be run on an MC88100 processor, GNU C emulates signed
4810 integer division using the unsigned integer division instruction
4811 divu), thereby avoiding the large penalty of a trap to the
4812 operating system. Such emulation has its own, smaller, execution
4813 cost in both time and space. To the extent that your code's
4814 important signed integer division operations are performed on two
4815 nonnegative operands, it may be desirable to use the div
4816 instruction directly.
4818 On the MC88110 processor the div instruction (also known as the
4819 divs instruction) processes negative operands without trapping to
4820 the operating system. When @samp{-m88110} is specified,
4821 @samp{-muse-div-instruction} is ignored, and the div instruction is used
4822 for signed integer division.
4824 Note that the result of dividing INT_MIN by -1 is undefined. In
4825 particular, the behavior of such a division with and without
4826 @samp{-muse-div-instruction} may differ.
4828 @item -mtrap-large-shift
4829 @itemx -mhandle-large-shift
4830 @kindex -mtrap-large-shift
4831 @kindex -mhandle-large-shift
4832 @cindex bit shift overflow (88k)
4833 @cindex large bit shifts (88k)
4834 Include code to detect bit-shifts of more than 31 bits; respectively,
4835 trap such shifts or emit code to handle them properly. By default GCC
4836 makes no special provision for large bit shifts.
4838 @item -mwarn-passed-structs
4839 @kindex -mwarn-passed-structs
4840 @cindex structure passing (88k)
4841 Warn when a function passes a struct as an argument or result.
4842 Structure-passing conventions have changed during the evolution of the C
4843 language, and are often the source of portability problems. By default,
4844 GCC issues no such warning.
4847 @node RS/6000 and PowerPC Options
4848 @subsection IBM RS/6000 and PowerPC Options
4849 @cindex RS/6000 and PowerPC Options
4850 @cindex IBM RS/6000 and PowerPC Options
4852 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
4860 @itemx -mpowerpc-gpopt
4861 @itemx -mno-powerpc-gpopt
4862 @itemx -mpowerpc-gfxopt
4863 @itemx -mno-powerpc-gfxopt
4865 @itemx -mno-powerpc64
4869 @kindex -mpowerpc-gpopt
4870 @kindex -mpowerpc-gfxopt
4872 GCC supports two related instruction set architectures for the
4873 RS/6000 and PowerPC. The @dfn{POWER} instruction set are those
4874 instructions supported by the @samp{rios} chip set used in the original
4875 RS/6000 systems and the @dfn{PowerPC} instruction set is the
4876 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
4877 the IBM 4xx microprocessors.
4879 Neither architecture is a subset of the other. However there is a
4880 large common subset of instructions supported by both. An MQ
4881 register is included in processors supporting the POWER architecture.
4883 You use these options to specify which instructions are available on the
4884 processor you are using. The default value of these options is
4885 determined when configuring GCC. Specifying the
4886 @samp{-mcpu=@var{cpu_type}} overrides the specification of these
4887 options. We recommend you use the @samp{-mcpu=@var{cpu_type}} option
4888 rather than the options listed above.
4890 The @samp{-mpower} option allows GCC to generate instructions that
4891 are found only in the POWER architecture and to use the MQ register.
4892 Specifying @samp{-mpower2} implies @samp{-power} and also allows GCC
4893 to generate instructions that are present in the POWER2 architecture but
4894 not the original POWER architecture.
4896 The @samp{-mpowerpc} option allows GCC to generate instructions that
4897 are found only in the 32-bit subset of the PowerPC architecture.
4898 Specifying @samp{-mpowerpc-gpopt} implies @samp{-mpowerpc} and also allows
4899 GCC to use the optional PowerPC architecture instructions in the
4900 General Purpose group, including floating-point square root. Specifying
4901 @samp{-mpowerpc-gfxopt} implies @samp{-mpowerpc} and also allows GCC to
4902 use the optional PowerPC architecture instructions in the Graphics
4903 group, including floating-point select.
4905 The @samp{-mpowerpc64} option allows GCC to generate the additional
4906 64-bit instructions that are found in the full PowerPC64 architecture
4907 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
4908 @samp{-mno-powerpc64}.
4910 If you specify both @samp{-mno-power} and @samp{-mno-powerpc}, GCC
4911 will use only the instructions in the common subset of both
4912 architectures plus some special AIX common-mode calls, and will not use
4913 the MQ register. Specifying both @samp{-mpower} and @samp{-mpowerpc}
4914 permits GCC to use any instruction from either architecture and to
4915 allow use of the MQ register; specify this for the Motorola MPC601.
4917 @item -mnew-mnemonics
4918 @itemx -mold-mnemonics
4919 @kindex -mnew-mnemonics
4920 @kindex -mold-mnemonics
4921 Select which mnemonics to use in the generated assembler code.
4922 @samp{-mnew-mnemonics} requests output that uses the assembler mnemonics
4923 defined for the PowerPC architecture, while @samp{-mold-mnemonics}
4924 requests the assembler mnemonics defined for the POWER architecture.
4925 Instructions defined in only one architecture have only one mnemonic;
4926 GCC uses that mnemonic irrespective of which of these options is
4929 GCC defaults to the mnemonics appropriate for the architecture in
4930 use. Specifying @samp{-mcpu=@var{cpu_type}} sometimes overrides the
4931 value of these option. Unless you are building a cross-compiler, you
4932 should normally not specify either @samp{-mnew-mnemonics} or
4933 @samp{-mold-mnemonics}, but should instead accept the default.
4935 @item -mcpu=@var{cpu_type}
4937 Set architecture type, register usage, choice of mnemonics, and
4938 instruction scheduling parameters for machine type @var{cpu_type}.
4939 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
4940 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
4941 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
4942 @samp{630}, @samp{740}, @samp{750}, @samp{power}, @samp{power2},
4943 @samp{powerpc}, @samp{403}, @samp{505}, @samp{801}, @samp{821},
4944 @samp{823}, and @samp{860} and @samp{common}. @samp{-mcpu=power},
4945 @samp{-mcpu=power2}, @samp{-mcpu=powerpc}, and @samp{-mcpu=powerpc64}
4946 specify generic POWER, POWER2, pure 32-bit PowerPC (i.e., not MPC601),
4947 and 64-bit PowerPC architecture machine types, with an appropriate,
4948 generic processor model assumed for scheduling purposes.@refill
4950 Specifying any of the following options:
4951 @samp{-mcpu=rios1}, @samp{-mcpu=rios2}, @samp{-mcpu=rsc},
4952 @samp{-mcpu=power}, or @samp{-mcpu=power2}
4953 enables the @samp{-mpower} option and disables the @samp{-mpowerpc} option;
4954 @samp{-mcpu=601} enables both the @samp{-mpower} and @samp{-mpowerpc} options.
4955 All of @samp{-mcpu=rs64a}, @samp{-mcpu=602}, @samp{-mcpu=603},
4956 @samp{-mcpu=603e}, @samp{-mcpu=604}, @samp{-mcpu=620}, @samp{-mcpu=630},
4957 @samp{-mcpu=740}, and @samp{-mcpu=750}
4958 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
4959 Exactly similarly, all of @samp{-mcpu=403},
4960 @samp{-mcpu=505}, @samp{-mcpu=821}, @samp{-mcpu=860} and @samp{-mcpu=powerpc}
4961 enable the @samp{-mpowerpc} option and disable the @samp{-mpower} option.
4962 @samp{-mcpu=common} disables both the
4963 @samp{-mpower} and @samp{-mpowerpc} options.@refill
4965 AIX versions 4 or greater selects @samp{-mcpu=common} by default, so
4966 that code will operate on all members of the RS/6000 POWER and PowerPC
4967 families. In that case, GCC will use only the instructions in the
4968 common subset of both architectures plus some special AIX common-mode
4969 calls, and will not use the MQ register. GCC assumes a generic
4970 processor model for scheduling purposes.
4972 Specifying any of the options @samp{-mcpu=rios1}, @samp{-mcpu=rios2},
4973 @samp{-mcpu=rsc}, @samp{-mcpu=power}, or @samp{-mcpu=power2} also
4974 disables the @samp{new-mnemonics} option. Specifying @samp{-mcpu=601},
4975 @samp{-mcpu=602}, @samp{-mcpu=603}, @samp{-mcpu=603e}, @samp{-mcpu=604},
4976 @samp{-mcpu=620}, @samp{-mcpu=630}, @samp{-mcpu=403}, @samp{-mcpu=505},
4977 @samp{-mcpu=821}, @samp{-mcpu=860} or @samp{-mcpu=powerpc} also enables
4978 the @samp{new-mnemonics} option.@refill
4980 Specifying @samp{-mcpu=403}, @samp{-mcpu=821}, or @samp{-mcpu=860} also
4981 enables the @samp{-msoft-float} option.
4983 @item -mtune=@var{cpu_type}
4984 Set the instruction scheduling parameters for machine type
4985 @var{cpu_type}, but do not set the architecture type, register usage,
4986 choice of mnemonics like @samp{-mcpu=}@var{cpu_type} would. The same
4987 values for @var{cpu_type} are used for @samp{-mtune=}@var{cpu_type} as
4988 for @samp{-mcpu=}@var{cpu_type}. The @samp{-mtune=}@var{cpu_type}
4989 option overrides the @samp{-mcpu=}@var{cpu_type} option in terms of
4990 instruction scheduling parameters.
4993 @itemx -mno-fp-in-toc
4994 @itemx -mno-sum-in-toc
4995 @itemx -mminimal-toc
4996 @kindex -mminimal-toc
4997 Modify generation of the TOC (Table Of Contents), which is created for
4998 every executable file. The @samp{-mfull-toc} option is selected by
4999 default. In that case, GCC will allocate at least one TOC entry for
5000 each unique non-automatic variable reference in your program. GCC
5001 will also place floating-point constants in the TOC. However, only
5002 16,384 entries are available in the TOC.
5004 If you receive a linker error message that saying you have overflowed
5005 the available TOC space, you can reduce the amount of TOC space used
5006 with the @samp{-mno-fp-in-toc} and @samp{-mno-sum-in-toc} options.
5007 @samp{-mno-fp-in-toc} prevents GCC from putting floating-point
5008 constants in the TOC and @samp{-mno-sum-in-toc} forces GCC to
5009 generate code to calculate the sum of an address and a constant at
5010 run-time instead of putting that sum into the TOC. You may specify one
5011 or both of these options. Each causes GCC to produce very slightly
5012 slower and larger code at the expense of conserving TOC space.
5014 If you still run out of space in the TOC even when you specify both of
5015 these options, specify @samp{-mminimal-toc} instead. This option causes
5016 GCC to make only one TOC entry for every file. When you specify this
5017 option, GCC will produce code that is slower and larger but which
5018 uses extremely little TOC space. You may wish to use this option
5019 only on files that contain less frequently executed code. @refill
5025 Enable 64-bit PowerPC ABI and calling convention: 64-bit pointers, 64-bit
5026 @code{long} type, and the infrastructure needed to support them.
5027 Specifying @samp{-m64} implies @samp{-mpowerpc64} and
5028 @samp{-mpowerpc}, while @samp{-m32} disables the 64-bit ABI and
5029 implies @samp{-mno-powerpc64}. GCC defaults to @samp{-m32}.
5034 On AIX, pass floating-point arguments to prototyped functions beyond the
5035 register save area (RSA) on the stack in addition to argument FPRs. The
5036 AIX calling convention was extended but not initially documented to
5037 handle an obscure K&R C case of calling a function that takes the
5038 address of its arguments with fewer arguments than declared. AIX XL
5039 compilers access floating point arguments which do not fit in the
5040 RSA from the stack when a subroutine is compiled without
5041 optimization. Because always storing floating-point arguments on the
5042 stack is inefficient and rarely needed, this option is not enabled by
5043 default and only is necessary when calling subroutines compiled by AIX
5044 XL compilers without optimization.
5048 Support @dfn{AIX Threads}. Link an application written to use
5049 @dfn{pthreads} with special libraries and startup code to enable the
5054 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE). Link an
5055 application written to use message passing with special startup code to
5056 enable the application to run. The system must have PE installed in the
5057 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
5058 must be overridden with the @samp{-specs=} option to specify the
5059 appropriate directory location. The Parallel Environment does not
5060 support threads, so the @samp{-mpe} option and the @samp{-mthreads}
5061 option are incompatible.
5065 @kindex -msoft-float
5066 Generate code that does not use (uses) the floating-point register set.
5067 Software floating point emulation is provided if you use the
5068 @samp{-msoft-float} option, and pass the option to GCC when linking.
5071 @itemx -mno-multiple
5072 Generate code that uses (does not use) the load multiple word
5073 instructions and the store multiple word instructions. These
5074 instructions are generated by default on POWER systems, and not
5075 generated on PowerPC systems. Do not use @samp{-mmultiple} on little
5076 endian PowerPC systems, since those instructions do not work when the
5077 processor is in little endian mode. The exceptions are PPC740 and
5078 PPC750 which permit the instructions usage in little endian mode.
5083 Generate code that uses (does not use) the load string instructions
5084 and the store string word instructions to save multiple registers and
5085 do small block moves. These instructions are generated by default on
5086 POWER systems, and not generated on PowerPC systems. Do not use
5087 @samp{-mstring} on little endian PowerPC systems, since those
5088 instructions do not work when the processor is in little endian mode.
5089 The exceptions are PPC740 and PPC750 which permit the instructions
5090 usage in little endian mode.
5095 Generate code that uses (does not use) the load or store instructions
5096 that update the base register to the address of the calculated memory
5097 location. These instructions are generated by default. If you use
5098 @samp{-mno-update}, there is a small window between the time that the
5099 stack pointer is updated and the address of the previous frame is
5100 stored, which means code that walks the stack frame across interrupts or
5101 signals may get corrupted data.
5104 @itemx -mno-fused-madd
5105 @kindex -mfused-madd
5106 Generate code that uses (does not use) the floating point multiply and
5107 accumulate instructions. These instructions are generated by default if
5108 hardware floating is used.
5110 @item -mno-bit-align
5113 On System V.4 and embedded PowerPC systems do not (do) force structures
5114 and unions that contain bit fields to be aligned to the base type of the
5117 For example, by default a structure containing nothing but 8
5118 @code{unsigned} bitfields of length 1 would be aligned to a 4 byte
5119 boundary and have a size of 4 bytes. By using @samp{-mno-bit-align},
5120 the structure would be aligned to a 1 byte boundary and be one byte in
5123 @item -mno-strict-align
5124 @itemx -mstrict-align
5125 @kindex -mstrict-align
5126 On System V.4 and embedded PowerPC systems do not (do) assume that
5127 unaligned memory references will be handled by the system.
5130 @itemx -mno-relocatable
5131 @kindex -mrelocatable
5132 On embedded PowerPC systems generate code that allows (does not allow)
5133 the program to be relocated to a different address at runtime. If you
5134 use @samp{-mrelocatable} on any module, all objects linked together must
5135 be compiled with @samp{-mrelocatable} or @samp{-mrelocatable-lib}.
5137 @item -mrelocatable-lib
5138 @itemx -mno-relocatable-lib
5139 On embedded PowerPC systems generate code that allows (does not allow)
5140 the program to be relocated to a different address at runtime. Modules
5141 compiled with @samp{-mrelocatable-lib} can be linked with either modules
5142 compiled without @samp{-mrelocatable} and @samp{-mrelocatable-lib} or
5143 with modules compiled with the @samp{-mrelocatable} options.
5147 On System V.4 and embedded PowerPC systems do not (do) assume that
5148 register 2 contains a pointer to a global area pointing to the addresses
5149 used in the program.
5152 @itemx -mlittle-endian
5153 On System V.4 and embedded PowerPC systems compile code for the
5154 processor in little endian mode. The @samp{-mlittle-endian} option is
5155 the same as @samp{-mlittle}.
5159 On System V.4 and embedded PowerPC systems compile code for the
5160 processor in big endian mode. The @samp{-mbig-endian} option is
5161 the same as @samp{-mbig}.
5164 On System V.4 and embedded PowerPC systems compile code using calling
5165 conventions that adheres to the March 1995 draft of the System V
5166 Application Binary Interface, PowerPC processor supplement. This is the
5167 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
5169 @item -mcall-sysv-eabi
5170 Specify both @samp{-mcall-sysv} and @samp{-meabi} options.
5172 @item -mcall-sysv-noeabi
5173 Specify both @samp{-mcall-sysv} and @samp{-mno-eabi} options.
5176 On System V.4 and embedded PowerPC systems compile code using calling
5177 conventions that are similar to those used on AIX. This is the
5178 default if you configured GCC using @samp{powerpc-*-eabiaix}.
5180 @item -mcall-solaris
5181 On System V.4 and embedded PowerPC systems compile code for the Solaris
5185 On System V.4 and embedded PowerPC systems compile code for the
5186 Linux-based GNU system.
5189 @itemx -mno-prototype
5190 On System V.4 and embedded PowerPC systems assume that all calls to
5191 variable argument functions are properly prototyped. Otherwise, the
5192 compiler must insert an instruction before every non prototyped call to
5193 set or clear bit 6 of the condition code register (@var{CR}) to
5194 indicate whether floating point values were passed in the floating point
5195 registers in case the function takes a variable arguments. With
5196 @samp{-mprototype}, only calls to prototyped variable argument functions
5197 will set or clear the bit.
5200 On embedded PowerPC systems, assume that the startup module is called
5201 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
5202 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
5206 On embedded PowerPC systems, assume that the startup module is called
5207 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
5211 On embedded PowerPC systems, assume that the startup module is called
5212 @file{crt0.o} and the standard C libraries are @file{libads.a} and
5216 On embedded PowerPC systems, assume that the startup module is called
5217 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
5221 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
5222 header to indicate that @samp{eabi} extended relocations are used.
5226 On System V.4 and embedded PowerPC systems do (do not) adhere to the
5227 Embedded Applications Binary Interface (eabi) which is a set of
5228 modifications to the System V.4 specifications. Selecting @code{-meabi}
5229 means that the stack is aligned to an 8 byte boundary, a function
5230 @code{__eabi} is called to from @code{main} to set up the eabi
5231 environment, and the @samp{-msdata} option can use both @code{r2} and
5232 @code{r13} to point to two separate small data areas. Selecting
5233 @code{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
5234 do not call an initialization function from @code{main}, and the
5235 @samp{-msdata} option will only use @code{r13} to point to a single
5236 small data area. The @samp{-meabi} option is on by default if you
5237 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
5240 On System V.4 and embedded PowerPC systems, put small initialized
5241 @code{const} global and static data in the @samp{.sdata2} section, which
5242 is pointed to by register @code{r2}. Put small initialized
5243 non-@code{const} global and static data in the @samp{.sdata} section,
5244 which is pointed to by register @code{r13}. Put small uninitialized
5245 global and static data in the @samp{.sbss} section, which is adjacent to
5246 the @samp{.sdata} section. The @samp{-msdata=eabi} option is
5247 incompatible with the @samp{-mrelocatable} option. The
5248 @samp{-msdata=eabi} option also sets the @samp{-memb} option.
5251 On System V.4 and embedded PowerPC systems, put small global and static
5252 data in the @samp{.sdata} section, which is pointed to by register
5253 @code{r13}. Put small uninitialized global and static data in the
5254 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
5255 The @samp{-msdata=sysv} option is incompatible with the
5256 @samp{-mrelocatable} option.
5258 @item -msdata=default
5260 On System V.4 and embedded PowerPC systems, if @samp{-meabi} is used,
5261 compile code the same as @samp{-msdata=eabi}, otherwise compile code the
5262 same as @samp{-msdata=sysv}.
5265 On System V.4 and embedded PowerPC systems, put small global and static
5266 data in the @samp{.sdata} section. Put small uninitialized global and
5267 static data in the @samp{.sbss} section. Do not use register @code{r13}
5268 to address small data however. This is the default behavior unless
5269 other @samp{-msdata} options are used.
5273 On embedded PowerPC systems, put all initialized global and static data
5274 in the @samp{.data} section, and all uninitialized data in the
5275 @samp{.bss} section.
5278 @cindex smaller data references (PowerPC)
5279 @cindex .sdata/.sdata2 references (PowerPC)
5280 On embedded PowerPC systems, put global and static items less than or
5281 equal to @var{num} bytes into the small data or bss sections instead of
5282 the normal data or bss section. By default, @var{num} is 8. The
5283 @samp{-G @var{num}} switch is also passed to the linker.
5284 All modules should be compiled with the same @samp{-G @var{num}} value.
5287 @itemx -mno-regnames
5288 On System V.4 and embedded PowerPC systems do (do not) emit register
5289 names in the assembly language output using symbolic forms.
5294 @subsection IBM RT Options
5296 @cindex IBM RT options
5298 These @samp{-m} options are defined for the IBM RT PC:
5302 Use an in-line code sequence for integer multiplies. This is the
5305 @item -mcall-lib-mul
5306 Call @code{lmul$$} for integer multiples.
5308 @item -mfull-fp-blocks
5309 Generate full-size floating point data blocks, including the minimum
5310 amount of scratch space recommended by IBM. This is the default.
5312 @item -mminimum-fp-blocks
5313 Do not include extra scratch space in floating point data blocks. This
5314 results in smaller code, but slower execution, since scratch space must
5315 be allocated dynamically.
5317 @cindex @file{varargs.h} and RT PC
5318 @cindex @file{stdarg.h} and RT PC
5319 @item -mfp-arg-in-fpregs
5320 Use a calling sequence incompatible with the IBM calling convention in
5321 which floating point arguments are passed in floating point registers.
5322 Note that @code{varargs.h} and @code{stdargs.h} will not work with
5323 floating point operands if this option is specified.
5325 @item -mfp-arg-in-gregs
5326 Use the normal calling convention for floating point arguments. This is
5329 @item -mhc-struct-return
5330 Return structures of more than one word in memory, rather than in a
5331 register. This provides compatibility with the MetaWare HighC (hc)
5332 compiler. Use the option @samp{-fpcc-struct-return} for compatibility
5333 with the Portable C Compiler (pcc).
5335 @item -mnohc-struct-return
5336 Return some structures of more than one word in registers, when
5337 convenient. This is the default. For compatibility with the
5338 IBM-supplied compilers, use the option @samp{-fpcc-struct-return} or the
5339 option @samp{-mhc-struct-return}.
5343 @subsection MIPS Options
5344 @cindex MIPS options
5346 These @samp{-m} options are defined for the MIPS family of computers:
5349 @item -mcpu=@var{cpu type}
5350 Assume the defaults for the machine type @var{cpu type} when scheduling
5351 instructions. The choices for @var{cpu type} are @samp{r2000}, @samp{r3000},
5352 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
5353 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
5354 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
5355 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
5356 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
5357 @var{cpu type} will schedule things appropriately for that particular
5358 chip, the compiler will not generate any code that does not meet level 1
5359 of the MIPS ISA (instruction set architecture) without a @samp{-mipsX}
5360 or @samp{-mabi} switch being used.
5363 Issue instructions from level 1 of the MIPS ISA. This is the default.
5364 @samp{r3000} is the default @var{cpu type} at this ISA level.
5367 Issue instructions from level 2 of the MIPS ISA (branch likely, square
5368 root instructions). @samp{r6000} is the default @var{cpu type} at this
5372 Issue instructions from level 3 of the MIPS ISA (64 bit instructions).
5373 @samp{r4000} is the default @var{cpu type} at this ISA level.
5376 Issue instructions from level 4 of the MIPS ISA (conditional move,
5377 prefetch, enhanced FPU instructions). @samp{r8000} is the default
5378 @var{cpu type} at this ISA level.
5381 Assume that 32 32-bit floating point registers are available. This is
5385 Assume that 32 64-bit floating point registers are available. This is
5386 the default when the @samp{-mips3} option is used.
5389 Assume that 32 32-bit general purpose registers are available. This is
5393 Assume that 32 64-bit general purpose registers are available. This is
5394 the default when the @samp{-mips3} option is used.
5397 Force int and long types to be 64 bits wide. See @samp{-mlong32} for an
5398 explanation of the default, and the width of pointers.
5401 Force long types to be 64 bits wide. See @samp{-mlong32} for an
5402 explanation of the default, and the width of pointers.
5405 Force long, int, and pointer types to be 32 bits wide.
5407 If none of @samp{-mlong32}, @samp{-mlong64}, or @samp{-mint64} are set,
5408 the size of ints, longs, and pointers depends on the ABI and ISA choosen.
5409 For @samp{-mabi=32}, and @samp{-mabi=n32}, ints and longs are 32 bits
5410 wide. For @samp{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
5411 For @samp{-mabi=eabi} and either @samp{-mips1} or @samp{-mips2}, ints
5412 and longs are 32 bits wide. For @samp{-mabi=eabi} and higher ISAs, ints
5413 are 32 bits, and longs are 64 bits wide. The width of pointer types is
5414 the smaller of the width of longs or the width of general purpose
5415 registers (which in turn depends on the ISA).
5422 Generate code for the indicated ABI. The default instruction level is
5423 @samp{-mips1} for @samp{32}, @samp{-mips3} for @samp{n32}, and
5424 @samp{-mips4} otherwise. Conversely, with @samp{-mips1} or
5425 @samp{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
5429 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
5430 add normal debug information. This is the default for all
5431 platforms except for the OSF/1 reference platform, using the OSF/rose
5432 object format. If the either of the @samp{-gstabs} or @samp{-gstabs+}
5433 switches are used, the @file{mips-tfile} program will encapsulate the
5434 stabs within MIPS ECOFF.
5437 Generate code for the GNU assembler. This is the default on the OSF/1
5438 reference platform, using the OSF/rose object format. Also, this is
5439 the default if the configure option @samp{--with-gnu-as} is used.
5441 @item -msplit-addresses
5442 @itemx -mno-split-addresses
5443 Generate code to load the high and low parts of address constants separately.
5444 This allows @code{gcc} to optimize away redundant loads of the high order
5445 bits of addresses. This optimization requires GNU as and GNU ld.
5446 This optimization is enabled by default for some embedded targets where
5447 GNU as and GNU ld are standard.
5451 The @samp{-mrnames} switch says to output code using the MIPS software
5452 names for the registers, instead of the hardware names (ie, @var{a0}
5453 instead of @var{$4}). The only known assembler that supports this option
5454 is the Algorithmics assembler.
5458 The @samp{-mgpopt} switch says to write all of the data declarations
5459 before the instructions in the text section, this allows the MIPS
5460 assembler to generate one word memory references instead of using two
5461 words for short global or static data items. This is on by default if
5462 optimization is selected.
5466 For each non-inline function processed, the @samp{-mstats} switch
5467 causes the compiler to emit one line to the standard error file to
5468 print statistics about the program (number of registers saved, stack
5473 The @samp{-mmemcpy} switch makes all block moves call the appropriate
5474 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
5475 generating inline code.
5478 @itemx -mno-mips-tfile
5479 The @samp{-mno-mips-tfile} switch causes the compiler not
5480 postprocess the object file with the @file{mips-tfile} program,
5481 after the MIPS assembler has generated it to add debug support. If
5482 @file{mips-tfile} is not run, then no local variables will be
5483 available to the debugger. In addition, @file{stage2} and
5484 @file{stage3} objects will have the temporary file names passed to the
5485 assembler embedded in the object file, which means the objects will
5486 not compare the same. The @samp{-mno-mips-tfile} switch should only
5487 be used when there are bugs in the @file{mips-tfile} program that
5488 prevents compilation.
5491 Generate output containing library calls for floating point.
5492 @strong{Warning:} the requisite libraries are not part of GCC.
5493 Normally the facilities of the machine's usual C compiler are used, but
5494 this can't be done directly in cross-compilation. You must make your
5495 own arrangements to provide suitable library functions for
5499 Generate output containing floating point instructions. This is the
5500 default if you use the unmodified sources.
5503 @itemx -mno-abicalls
5504 Emit (or do not emit) the pseudo operations @samp{.abicalls},
5505 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
5506 position independent code.
5509 @itemx -mno-long-calls
5510 Do all calls with the @samp{JALR} instruction, which requires
5511 loading up a function's address into a register before the call.
5512 You need to use this switch, if you call outside of the current
5513 512 megabyte segment to functions that are not through pointers.
5516 @itemx -mno-half-pic
5517 Put pointers to extern references into the data section and load them
5518 up, rather than put the references in the text section.
5520 @item -membedded-pic
5521 @itemx -mno-embedded-pic
5522 Generate PIC code suitable for some embedded systems. All calls are
5523 made using PC relative address, and all data is addressed using the $gp
5524 register. No more than 65536 bytes of global data may be used. This
5525 requires GNU as and GNU ld which do most of the work. This currently
5526 only works on targets which use ECOFF; it does not work with ELF.
5528 @item -membedded-data
5529 @itemx -mno-embedded-data
5530 Allocate variables to the read-only data section first if possible, then
5531 next in the small data section if possible, otherwise in data. This gives
5532 slightly slower code than the default, but reduces the amount of RAM required
5533 when executing, and thus may be preferred for some embedded systems.
5535 @item -msingle-float
5536 @itemx -mdouble-float
5537 The @samp{-msingle-float} switch tells gcc to assume that the floating
5538 point coprocessor only supports single precision operations, as on the
5539 @samp{r4650} chip. The @samp{-mdouble-float} switch permits gcc to use
5540 double precision operations. This is the default.
5544 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
5545 as on the @samp{r4650} chip.
5548 Turns on @samp{-msingle-float}, @samp{-mmad}, and, at least for now,
5553 Enable 16-bit instructions.
5556 Use the entry and exit pseudo ops. This option can only be used with
5560 Compile code for the processor in little endian mode.
5561 The requisite libraries are assumed to exist.
5564 Compile code for the processor in big endian mode.
5565 The requisite libraries are assumed to exist.
5568 @cindex smaller data references (MIPS)
5569 @cindex gp-relative references (MIPS)
5570 Put global and static items less than or equal to @var{num} bytes into
5571 the small data or bss sections instead of the normal data or bss
5572 section. This allows the assembler to emit one word memory reference
5573 instructions based on the global pointer (@var{gp} or @var{$28}),
5574 instead of the normal two words used. By default, @var{num} is 8 when
5575 the MIPS assembler is used, and 0 when the GNU assembler is used. The
5576 @samp{-G @var{num}} switch is also passed to the assembler and linker.
5577 All modules should be compiled with the same @samp{-G @var{num}}
5581 Tell the MIPS assembler to not run its preprocessor over user
5582 assembler files (with a @samp{.s} suffix) when assembling them.
5586 These options are defined by the macro
5587 @code{TARGET_SWITCHES} in the machine description. The default for the
5588 options is also defined by that macro, which enables you to change the
5593 @subsection Intel 386 Options
5594 @cindex i386 Options
5595 @cindex Intel 386 Options
5597 These @samp{-m} options are defined for the i386 family of computers:
5600 @item -mcpu=@var{cpu type}
5601 Assume the defaults for the machine type @var{cpu type} when scheduling
5602 instructions. The choices for @var{cpu type} are:
5604 @multitable @columnfractions .20 .20 .20 .20
5605 @item @samp{i386} @tab @samp{i486} @tab @samp{i586} @tab @samp{i686}
5606 @item @samp{pentium} @tab @samp{pentiumpro} @tab @samp{k6}
5609 While picking a specific @var{cpu type} will schedule things appropriately
5610 for that particular chip, the compiler will not generate any code that
5611 does not run on the i386 without the @samp{-march=@var{cpu type}} option
5612 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
5613 is equivalent to @samp{pentiumpro}. @samp{k6} is the AMD chip as
5614 opposed to the Intel ones.
5616 @item -march=@var{cpu type}
5617 Generate instructions for the machine type @var{cpu type}. The choices
5618 for @var{cpu type} are the same as for @samp{-mcpu}. Moreover,
5619 specifying @samp{-march=@var{cpu type}} implies @samp{-mcpu=@var{cpu type}}.
5625 Synonyms for -mcpu=i386, -mcpu=i486, -mcpu=pentium, and -mcpu=pentiumpro
5626 respectively. These synonyms are deprecated.
5630 Control whether or not the compiler uses IEEE floating point
5631 comparisons. These handle correctly the case where the result of a
5632 comparison is unordered.
5635 Generate output containing library calls for floating point.
5636 @strong{Warning:} the requisite libraries are not part of GCC.
5637 Normally the facilities of the machine's usual C compiler are used, but
5638 this can't be done directly in cross-compilation. You must make your
5639 own arrangements to provide suitable library functions for
5642 On machines where a function returns floating point results in the 80387
5643 register stack, some floating point opcodes may be emitted even if
5644 @samp{-msoft-float} is used.
5646 @item -mno-fp-ret-in-387
5647 Do not use the FPU registers for return values of functions.
5649 The usual calling convention has functions return values of types
5650 @code{float} and @code{double} in an FPU register, even if there
5651 is no FPU. The idea is that the operating system should emulate
5654 The option @samp{-mno-fp-ret-in-387} causes such values to be returned
5655 in ordinary CPU registers instead.
5657 @item -mno-fancy-math-387
5658 Some 387 emulators do not support the @code{sin}, @code{cos} and
5659 @code{sqrt} instructions for the 387. Specify this option to avoid
5660 generating those instructions. This option is the default on FreeBSD.
5661 As of revision 2.6.1, these instructions are not generated unless you
5662 also use the @samp{-ffast-math} switch.
5664 @item -malign-double
5665 @itemx -mno-align-double
5666 Control whether GCC aligns @code{double}, @code{long double}, and
5667 @code{long long} variables on a two word boundary or a one word
5668 boundary. Aligning @code{double} variables on a two word boundary will
5669 produce code that runs somewhat faster on a @samp{Pentium} at the
5670 expense of more memory.
5672 @strong{Warning:} if you use the @samp{-malign-double} switch,
5673 structures containing the above types will be aligned differently than
5674 the published application binary interface specifications for the 386.
5677 @itemx -mno-svr3-shlib
5678 Control whether GCC places uninitialized locals into @code{bss} or
5679 @code{data}. @samp{-msvr3-shlib} places these locals into @code{bss}.
5680 These options are meaningful only on System V Release 3.
5682 @item -mno-wide-multiply
5683 @itemx -mwide-multiply
5684 Control whether GCC uses the @code{mul} and @code{imul} that produce
5685 64 bit results in @code{eax:edx} from 32 bit operands to do @code{long
5686 long} multiplies and 32-bit division by constants.
5689 Use a different function-calling convention, in which functions that
5690 take a fixed number of arguments return with the @code{ret} @var{num}
5691 instruction, which pops their arguments while returning. This saves one
5692 instruction in the caller since there is no need to pop the arguments
5695 You can specify that an individual function is called with this calling
5696 sequence with the function attribute @samp{stdcall}. You can also
5697 override the @samp{-mrtd} option by using the function attribute
5698 @samp{cdecl}. @xref{Function Attributes}.
5700 @strong{Warning:} this calling convention is incompatible with the one
5701 normally used on Unix, so you cannot use it if you need to call
5702 libraries compiled with the Unix compiler.
5704 Also, you must provide function prototypes for all functions that
5705 take variable numbers of arguments (including @code{printf});
5706 otherwise incorrect code will be generated for calls to those
5709 In addition, seriously incorrect code will result if you call a
5710 function with too many arguments. (Normally, extra arguments are
5711 harmlessly ignored.)
5713 @item -mreg-alloc=@var{regs}
5714 Control the default allocation order of integer registers. The
5715 string @var{regs} is a series of letters specifying a register. The
5716 supported letters are: @code{a} allocate EAX; @code{b} allocate EBX;
5717 @code{c} allocate ECX; @code{d} allocate EDX; @code{S} allocate ESI;
5718 @code{D} allocate EDI; @code{B} allocate EBP.
5720 @item -mregparm=@var{num}
5721 Control how many registers are used to pass integer arguments. By
5722 default, no registers are used to pass arguments, and at most 3
5723 registers can be used. You can control this behavior for a specific
5724 function by using the function attribute @samp{regparm}.
5725 @xref{Function Attributes}.
5727 @strong{Warning:} if you use this switch, and
5728 @var{num} is nonzero, then you must build all modules with the same
5729 value, including any libraries. This includes the system libraries and
5732 @item -malign-loops=@var{num}
5733 Align loops to a 2 raised to a @var{num} byte boundary. If
5734 @samp{-malign-loops} is not specified, the default is 2 unless
5735 gas 2.8 (or later) is being used in which case the default is
5736 to align the loop on a 16 byte boundary if it is less than 8
5739 @item -malign-jumps=@var{num}
5740 Align instructions that are only jumped to to a 2 raised to a @var{num}
5741 byte boundary. If @samp{-malign-jumps} is not specified, the default is
5742 2 if optimizing for a 386, and 4 if optimizing for a 486 unless
5743 gas 2.8 (or later) is being used in which case the default is
5744 to align the instruction on a 16 byte boundary if it is less
5747 @item -malign-functions=@var{num}
5748 Align the start of functions to a 2 raised to @var{num} byte boundary.
5749 If @samp{-malign-functions} is not specified, the default is 2 if optimizing
5750 for a 386, and 4 if optimizing for a 486.
5752 @item -mpreferred-stack-boundary=@var{num}
5753 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
5754 byte boundary. If @samp{-mpreferred-stack-boundary} is not specified,
5755 the default is 4 (16 bytes or 128 bits).
5757 The stack is required to be aligned on a 4 byte boundary. On Pentium
5758 and PentiumPro, @code{double} and @code{long double} values should be
5759 aligned to an 8 byte boundary (see @samp{-malign-double}) or suffer
5760 significant run time performance penalties. On Pentium III, the
5761 Streaming SIMD Extention (SSE) data type @code{__m128} suffers similar
5762 penalties if it is not 16 byte aligned.
5764 To ensure proper alignment of this values on the stack, the stack boundary
5765 must be as aligned as that required by any value stored on the stack.
5766 Further, every function must be generated such that it keeps the stack
5767 aligned. Thus calling a function compiled with a higher preferred
5768 stack boundary from a function compiled with a lower preferred stack
5769 boundary will most likely misalign the stack. It is recommended that
5770 libraries that use callbacks always use the default setting.
5772 This extra alignment does consume extra stack space. Code that is sensitive
5773 to stack space usage, such as embedded systems and operating system kernels,
5774 may want to reduce the preferred alignment to
5775 @samp{-mpreferred-stack-boundary=2}.
5779 @subsection HPPA Options
5780 @cindex HPPA Options
5782 These @samp{-m} options are defined for the HPPA family of computers:
5785 @item -march=@var{architecture type}
5786 Generate code for the specified architecture. The choices for
5787 @var{architecture type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
5788 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
5789 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
5790 architecture option for your machine. Code compiled for lower numbered
5791 architectures will run on higher numbered architectures, but not the
5794 PA 2.0 support currently requires gas snapshot 19990413 or later. The
5795 next release of binutils (current is 2.9.1) will probably contain PA 2.0
5799 @itemx -mpa-risc-1-1
5800 @itemx -mpa-risc-2-0
5801 Synonyms for -march=1.0, -march=1.1, and -march=2.0 respectively.
5804 Generate code suitable for big switch tables. Use this option only if
5805 the assembler/linker complain about out of range branches within a switch
5808 @item -mjump-in-delay
5809 Fill delay slots of function calls with unconditional jump instructions
5810 by modifying the return pointer for the function call to be the target
5811 of the conditional jump.
5813 @item -mdisable-fpregs
5814 Prevent floating point registers from being used in any manner. This is
5815 necessary for compiling kernels which perform lazy context switching of
5816 floating point registers. If you use this option and attempt to perform
5817 floating point operations, the compiler will abort.
5819 @item -mdisable-indexing
5820 Prevent the compiler from using indexing address modes. This avoids some
5821 rather obscure problems when compiling MIG generated code under MACH.
5823 @item -mno-space-regs
5824 Generate code that assumes the target has no space registers. This allows
5825 GCC to generate faster indirect calls and use unscaled index address modes.
5827 Such code is suitable for level 0 PA systems and kernels.
5829 @item -mfast-indirect-calls
5830 Generate code that assumes calls never cross space boundaries. This
5831 allows GCC to emit code which performs faster indirect calls.
5833 This option will not work in the presense of shared libraries or nested
5836 @item -mlong-load-store
5837 Generate 3-instruction load and store sequences as sometimes required by
5838 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
5841 @item -mportable-runtime
5842 Use the portable calling conventions proposed by HP for ELF systems.
5845 Enable the use of assembler directives only GAS understands.
5847 @item -mschedule=@var{cpu type}
5848 Schedule code according to the constraints for the machine type
5849 @var{cpu type}. The choices for @var{cpu type} are @samp{700}
5850 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
5851 @file{/usr/lib/sched.models} on an HP-UX system to determine the
5852 proper scheduling option for your machine.
5855 Enable the optimization pass in the HPUX linker. Note this makes symbolic
5856 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
5857 in which they give bogus error messages when linking some programs.
5860 Generate output containing library calls for floating point.
5861 @strong{Warning:} the requisite libraries are not available for all HPPA
5862 targets. Normally the facilities of the machine's usual C compiler are
5863 used, but this cannot be done directly in cross-compilation. You must make
5864 your own arrangements to provide suitable library functions for
5865 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
5866 does provide software floating point support.
5868 @samp{-msoft-float} changes the calling convention in the output file;
5869 therefore, it is only useful if you compile @emph{all} of a program with
5870 this option. In particular, you need to compile @file{libgcc.a}, the
5871 library that comes with GCC, with @samp{-msoft-float} in order for
5875 @node Intel 960 Options
5876 @subsection Intel 960 Options
5878 These @samp{-m} options are defined for the Intel 960 implementations:
5881 @item -m@var{cpu type}
5882 Assume the defaults for the machine type @var{cpu type} for some of
5883 the other options, including instruction scheduling, floating point
5884 support, and addressing modes. The choices for @var{cpu type} are
5885 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
5886 @samp{sa}, and @samp{sb}.
5892 The @samp{-mnumerics} option indicates that the processor does support
5893 floating-point instructions. The @samp{-msoft-float} option indicates
5894 that floating-point support should not be assumed.
5896 @item -mleaf-procedures
5897 @itemx -mno-leaf-procedures
5898 Do (or do not) attempt to alter leaf procedures to be callable with the
5899 @code{bal} instruction as well as @code{call}. This will result in more
5900 efficient code for explicit calls when the @code{bal} instruction can be
5901 substituted by the assembler or linker, but less efficient code in other
5902 cases, such as calls via function pointers, or using a linker that doesn't
5903 support this optimization.
5906 @itemx -mno-tail-call
5907 Do (or do not) make additional attempts (beyond those of the
5908 machine-independent portions of the compiler) to optimize tail-recursive
5909 calls into branches. You may not want to do this because the detection of
5910 cases where this is not valid is not totally complete. The default is
5911 @samp{-mno-tail-call}.
5913 @item -mcomplex-addr
5914 @itemx -mno-complex-addr
5915 Assume (or do not assume) that the use of a complex addressing mode is a
5916 win on this implementation of the i960. Complex addressing modes may not
5917 be worthwhile on the K-series, but they definitely are on the C-series.
5918 The default is currently @samp{-mcomplex-addr} for all processors except
5922 @itemx -mno-code-align
5923 Align code to 8-byte boundaries for faster fetching (or don't bother).
5924 Currently turned on by default for C-series implementations only.
5927 @item -mclean-linkage
5928 @itemx -mno-clean-linkage
5929 These options are not fully implemented.
5933 @itemx -mic2.0-compat
5934 @itemx -mic3.0-compat
5935 Enable compatibility with iC960 v2.0 or v3.0.
5939 Enable compatibility with the iC960 assembler.
5941 @item -mstrict-align
5942 @itemx -mno-strict-align
5943 Do not permit (do permit) unaligned accesses.
5946 Enable structure-alignment compatibility with Intel's gcc release version
5947 1.3 (based on gcc 1.37). This option implies @samp{-mstrict-align}.
5949 @item -mlong-double-64
5950 Implement type @samp{long double} as 64-bit floating point numbers.
5951 Without the option @samp{long double} is implemented by 80-bit
5952 floating point numbers. The only reason we have it because there is
5953 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
5954 is only useful for people using soft-float targets. Otherwise, we
5955 should recommend against use of it.
5959 @node DEC Alpha Options
5960 @subsection DEC Alpha Options
5962 These @samp{-m} options are defined for the DEC Alpha implementations:
5965 @item -mno-soft-float
5967 Use (do not use) the hardware floating-point instructions for
5968 floating-point operations. When @code{-msoft-float} is specified,
5969 functions in @file{libgcc1.c} will be used to perform floating-point
5970 operations. Unless they are replaced by routines that emulate the
5971 floating-point operations, or compiled in such a way as to call such
5972 emulations routines, these routines will issue floating-point
5973 operations. If you are compiling for an Alpha without floating-point
5974 operations, you must ensure that the library is built so as not to call
5977 Note that Alpha implementations without floating-point operations are
5978 required to have floating-point registers.
5982 Generate code that uses (does not use) the floating-point register set.
5983 @code{-mno-fp-regs} implies @code{-msoft-float}. If the floating-point
5984 register set is not used, floating point operands are passed in integer
5985 registers as if they were integers and floating-point results are passed
5986 in $0 instead of $f0. This is a non-standard calling sequence, so any
5987 function with a floating-point argument or return value called by code
5988 compiled with @code{-mno-fp-regs} must also be compiled with that
5991 A typical use of this option is building a kernel that does not use,
5992 and hence need not save and restore, any floating-point registers.
5995 The Alpha architecture implements floating-point hardware optimized for
5996 maximum performance. It is mostly compliant with the IEEE floating
5997 point standard. However, for full compliance, software assistance is
5998 required. This option generates code fully IEEE compliant code
5999 @emph{except} that the @var{inexact flag} is not maintained (see below).
6000 If this option is turned on, the CPP macro @code{_IEEE_FP} is defined
6001 during compilation. The option is a shorthand for: @samp{-D_IEEE_FP
6002 -mfp-trap-mode=su -mtrap-precision=i -mieee-conformant}. The resulting
6003 code is less efficient but is able to correctly support denormalized
6004 numbers and exceptional IEEE values such as not-a-number and plus/minus
6005 infinity. Other Alpha compilers call this option
6006 @code{-ieee_with_no_inexact}.
6008 @item -mieee-with-inexact
6009 @c overfull hbox here --bob 22 jul96
6010 @c original text between ignore ... end ignore
6012 This is like @samp{-mieee} except the generated code also maintains the
6013 IEEE @var{inexact flag}. Turning on this option causes the generated
6014 code to implement fully-compliant IEEE math. The option is a shorthand
6015 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus @samp{-mieee-conformant},
6016 @samp{-mfp-trap-mode=sui}, and @samp{-mtrap-precision=i}. On some Alpha
6017 implementations the resulting code may execute significantly slower than
6018 the code generated by default. Since there is very little code that
6019 depends on the @var{inexact flag}, you should normally not specify this
6020 option. Other Alpha compilers call this option
6021 @samp{-ieee_with_inexact}.
6023 @c changed paragraph
6024 This is like @samp{-mieee} except the generated code also maintains the
6025 IEEE @var{inexact flag}. Turning on this option causes the generated
6026 code to implement fully-compliant IEEE math. The option is a shorthand
6027 for @samp{-D_IEEE_FP -D_IEEE_FP_INEXACT} plus the three following:
6028 @samp{-mieee-conformant},
6029 @samp{-mfp-trap-mode=sui},
6030 and @samp{-mtrap-precision=i}.
6031 On some Alpha implementations the resulting code may execute
6032 significantly slower than the code generated by default. Since there
6033 is very little code that depends on the @var{inexact flag}, you should
6034 normally not specify this option. Other Alpha compilers call this
6035 option @samp{-ieee_with_inexact}.
6036 @c end changes to prevent overfull hboxes
6038 @item -mfp-trap-mode=@var{trap mode}
6039 This option controls what floating-point related traps are enabled.
6040 Other Alpha compilers call this option @samp{-fptm }@var{trap mode}.
6041 The trap mode can be set to one of four values:
6045 This is the default (normal) setting. The only traps that are enabled
6046 are the ones that cannot be disabled in software (e.g., division by zero
6050 In addition to the traps enabled by @samp{n}, underflow traps are enabled
6054 Like @samp{su}, but the instructions are marked to be safe for software
6055 completion (see Alpha architecture manual for details).
6058 Like @samp{su}, but inexact traps are enabled as well.
6061 @item -mfp-rounding-mode=@var{rounding mode}
6062 Selects the IEEE rounding mode. Other Alpha compilers call this option
6063 @samp{-fprm }@var{rounding mode}. The @var{rounding mode} can be one
6068 Normal IEEE rounding mode. Floating point numbers are rounded towards
6069 the nearest machine number or towards the even machine number in case
6073 Round towards minus infinity.
6076 Chopped rounding mode. Floating point numbers are rounded towards zero.
6079 Dynamic rounding mode. A field in the floating point control register
6080 (@var{fpcr}, see Alpha architecture reference manual) controls the
6081 rounding mode in effect. The C library initializes this register for
6082 rounding towards plus infinity. Thus, unless your program modifies the
6083 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
6086 @item -mtrap-precision=@var{trap precision}
6087 In the Alpha architecture, floating point traps are imprecise. This
6088 means without software assistance it is impossible to recover from a
6089 floating trap and program execution normally needs to be terminated.
6090 GCC can generate code that can assist operating system trap handlers
6091 in determining the exact location that caused a floating point trap.
6092 Depending on the requirements of an application, different levels of
6093 precisions can be selected:
6097 Program precision. This option is the default and means a trap handler
6098 can only identify which program caused a floating point exception.
6101 Function precision. The trap handler can determine the function that
6102 caused a floating point exception.
6105 Instruction precision. The trap handler can determine the exact
6106 instruction that caused a floating point exception.
6109 Other Alpha compilers provide the equivalent options called
6110 @samp{-scope_safe} and @samp{-resumption_safe}.
6112 @item -mieee-conformant
6113 This option marks the generated code as IEEE conformant. You must not
6114 use this option unless you also specify @samp{-mtrap-precision=i} and either
6115 @samp{-mfp-trap-mode=su} or @samp{-mfp-trap-mode=sui}. Its only effect
6116 is to emit the line @samp{.eflag 48} in the function prologue of the
6117 generated assembly file. Under DEC Unix, this has the effect that
6118 IEEE-conformant math library routines will be linked in.
6120 @item -mbuild-constants
6121 Normally GCC examines a 32- or 64-bit integer constant to
6122 see if it can construct it from smaller constants in two or three
6123 instructions. If it cannot, it will output the constant as a literal and
6124 generate code to load it from the data segment at runtime.
6126 Use this option to require GCC to construct @emph{all} integer constants
6127 using code, even if it takes more instructions (the maximum is six).
6129 You would typically use this option to build a shared library dynamic
6130 loader. Itself a shared library, it must relocate itself in memory
6131 before it can find the variables and constants in its own data segment.
6135 Select whether to generate code to be assembled by the vendor-supplied
6136 assembler (@samp{-malpha-as}) or by the GNU assembler @samp{-mgas}.
6144 Indicate whether GCC should generate code to use the optional BWX,
6145 CIX, and MAX instruction sets. The default is to use the instruction sets
6146 supported by the CPU type specified via @samp{-mcpu=} option or that
6147 of the CPU on which GCC was built if none was specified.
6149 @item -mcpu=@var{cpu_type}
6150 Set the instruction set, register set, and instruction scheduling
6151 parameters for machine type @var{cpu_type}. You can specify either the
6152 @samp{EV} style name or the corresponding chip number. GCC
6153 supports scheduling parameters for the EV4 and EV5 family of processors
6154 and will choose the default values for the instruction set from
6155 the processor you specify. If you do not specify a processor type,
6156 GCC will default to the processor on which the compiler was built.
6158 Supported values for @var{cpu_type} are
6163 Schedules as an EV4 and has no instruction set extensions.
6167 Schedules as an EV5 and has no instruction set extensions.
6171 Schedules as an EV5 and supports the BWX extension.
6176 Schedules as an EV5 and supports the BWX and MAX extensions.
6180 Schedules as an EV5 (until Digital releases the scheduling parameters
6181 for the EV6) and supports the BWX, CIX, and MAX extensions.
6184 @item -mmemory-latency=@var{time}
6185 Sets the latency the scheduler should assume for typical memory
6186 references as seen by the application. This number is highly
6187 dependant on the memory access patterns used by the application
6188 and the size of the external cache on the machine.
6190 Valid options for @var{time} are
6194 A decimal number representing clock cycles.
6200 The compiler contains estimates of the number of clock cycles for
6201 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
6202 (also called Dcache, Scache, and Bcache), as well as to main memory.
6203 Note that L3 is only valid for EV5.
6208 @node Clipper Options
6209 @subsection Clipper Options
6211 These @samp{-m} options are defined for the Clipper implementations:
6215 Produce code for a C300 Clipper processor. This is the default.
6218 Produce code for a C400 Clipper processor i.e. use floating point
6222 @node H8/300 Options
6223 @subsection H8/300 Options
6225 These @samp{-m} options are defined for the H8/300 implementations:
6229 Shorten some address references at link time, when possible; uses the
6230 linker option @samp{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
6231 ld.info, Using ld}, for a fuller description.
6234 Generate code for the H8/300H.
6237 Generate code for the H8/S.
6240 Make @code{int} data 32 bits by default.
6243 On the h8/300h, use the same alignment rules as for the h8/300.
6244 The default for the h8/300h is to align longs and floats on 4 byte boundaries.
6245 @samp{-malign-300} causes them to be aligned on 2 byte boundaries.
6246 This option has no effect on the h8/300.
6250 @subsection SH Options
6252 These @samp{-m} options are defined for the SH implementations:
6256 Generate code for the SH1.
6259 Generate code for the SH2.
6262 Generate code for the SH3.
6265 Generate code for the SH3e.
6268 Compile code for the processor in big endian mode.
6271 Compile code for the processor in little endian mode.
6274 Align doubles at 64 bit boundaries. Note that this changes the calling
6275 conventions, and thus some functions from the standard C library will
6276 not work unless you recompile it first with -mdalign.
6279 Shorten some address references at link time, when possible; uses the
6280 linker option @samp{-relax}.
6283 @node System V Options
6284 @subsection Options for System V
6286 These additional options are available on System V Release 4 for
6287 compatibility with other compilers on those systems:
6291 Create a shared object.
6292 It is recommended that @samp{-symbolic} or @samp{-shared} be used instead.
6295 Identify the versions of each tool used by the compiler, in a
6296 @code{.ident} assembler directive in the output.
6299 Refrain from adding @code{.ident} directives to the output file (this is
6302 @item -YP,@var{dirs}
6303 Search the directories @var{dirs}, and no others, for libraries
6304 specified with @samp{-l}.
6307 Look in the directory @var{dir} to find the M4 preprocessor.
6308 The assembler uses this option.
6309 @c This is supposed to go with a -Yd for predefined M4 macro files, but
6310 @c the generic assembler that comes with Solaris takes just -Ym.
6313 @node TMS320C3x/C4x Options
6314 @subsection TMS320C3x/C4x Options
6315 @cindex TMS320C3x/C4x Options
6317 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
6321 @item -mcpu=@var{cpu_type}
6322 Set the instruction set, register set, and instruction scheduling
6323 parameters for machine type @var{cpu_type}. Supported values for
6324 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
6325 @samp{c44}. The default is @samp{c40} to generate code for the
6330 @itemx -msmall-memory
6332 Generates code for the big or small memory model. The small memory
6333 model assumed that all data fits into one 64K word page. At run-time
6334 the data page (DP) register must be set to point to the 64K page
6335 containing the .bss and .data program sections. The big memory model is
6336 the default and requires reloading of the DP register for every direct
6341 Allow (disallow) allocation of general integer operands into the block
6346 Enable (disable) generation of code using decrement and branch,
6347 DBcond(D), instructions. This is enabled by default for the C4x. To be
6348 on the safe side, this is disabled for the C3x, since the maximum
6349 iteration count on the C3x is 2^23 + 1 (but who iterates loops more than
6350 2^23 times on the C3x?). Note that GCC will try to reverse a loop so
6351 that it can utilise the decrement and branch instruction, but will give
6352 up if there is more than one memory reference in the loop. Thus a loop
6353 where the loop counter is decremented can generate slightly more
6354 efficient code, in cases where the RPTB instruction cannot be utilised.
6356 @item -mdp-isr-reload
6358 Force the DP register to be saved on entry to an interrupt service
6359 routine (ISR), reloaded to point to the data section, and restored on
6360 exit from the ISR. This should not be required unless someone has
6361 violated the small memory model by modifying the DP register, say within
6366 For the C3x use the 24-bit MPYI instruction for integer multiplies
6367 instead of a library call to guarantee 32-bit results. Note that if one
6368 of the operands is a constant, then the multiplication will be performed
6369 using shifts and adds. If the -mmpyi option is not specified for the C3x,
6370 then squaring operations are performed inline instead of a library call.
6373 @itemx -mno-fast-fix
6374 The C3x/C4x FIX instruction to convert a floating point value to an
6375 integer value chooses the nearest integer less than or equal to the
6376 floating point value rather than to the nearest integer. Thus if the
6377 floating point number is negative, the result will be incorrectly
6378 truncated an additional code is necessary to detect and correct this
6379 case. This option can be used to disable generation of the additional
6380 code required to correct the result.
6384 Enable (disable) generation of repeat block sequences using the RPTB
6385 instruction for zero overhead looping. The RPTB construct is only used
6386 for innermost loops that do not call functions or jump across the loop
6387 boundaries. There is no advantage having nested RPTB loops due to the
6388 overhead required to save and restore the RC, RS, and RE registers.
6389 This is enabled by default with -O2.
6391 @item -mrpts=@var{count}
6393 Enable (disable) the use of the single instruction repeat instruction
6394 RPTS. If a repeat block contains a single instruction, and the loop
6395 count can be guaranteed to be less than the value @var{count}, GCC will
6396 emit a RPTS instruction instead of a RPTB. If no value is specified,
6397 then a RPTS will be emitted even if the loop count cannot be determined
6398 at compile time. Note that the repeated instruction following RPTS does
6399 not have to be reloaded from memory each iteration, thus freeing up the
6400 CPU buses for oeprands. However, since interrupts are blocked by this
6401 instruction, it is disabled by default.
6403 @item -mloop-unsigned
6404 @itemx -mno-loop-unsigned
6405 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
6406 is 2^31 + 1 since these instructions test if the iteration count is
6407 negative to terminate the loop. If the iteration count is unsigned
6408 there is a possibility than the 2^31 + 1 maximum iteration count may be
6409 exceeded. This switch allows an unsigned iteration count.
6412 Try to emit an assembler syntax that the TI assembler (asm30) is happy
6413 with. This also enforces compatibility with the API employed by the TI
6414 C3x C compiler. For example, long doubles are passed as structures
6415 rather than in floating point registers.
6419 Generate code that uses registers (stack) for passing arguments to functions.
6420 By default, arguments are passed in registers where possible rather
6421 than by pushing arguments on to the stack.
6423 @item -mparallel-insns
6424 @itemx -mno-parallel-insns
6425 Allow the generation of parallel instructions. This is enabled by
6428 @item -mparallel-mpy
6429 @itemx -mno-parallel-mpy
6430 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
6431 provided -mparallel-insns is also specified. These instructions have
6432 tight register constraints which can pessimize the code generation
6438 @subsection V850 Options
6439 @cindex V850 Options
6441 These @samp{-m} options are defined for V850 implementations:
6445 @itemx -mno-long-calls
6446 Treat all calls as being far away (near). If calls are assumed to be
6447 far away, the compiler will always load the functions address up into a
6448 register, and call indirect through the pointer.
6452 Do not optimize (do optimize) basic blocks that use the same index
6453 pointer 4 or more times to copy pointer into the @code{ep} register, and
6454 use the shorter @code{sld} and @code{sst} instructions. The @samp{-mep}
6455 option is on by default if you optimize.
6457 @item -mno-prolog-function
6458 @itemx -mprolog-function
6459 Do not use (do use) external functions to save and restore registers at
6460 the prolog and epilog of a function. The external functions are slower,
6461 but use less code space if more than one function saves the same number
6462 of registers. The @samp{-mprolog-function} option is on by default if
6466 Try to make the code as small as possible. At present, this just turns
6467 on the @samp{-mep} and @samp{-mprolog-function} options.
6470 Put static or global variables whose size is @var{n} bytes or less into
6471 the tiny data area that register @code{ep} points to. The tiny data
6472 area can hold up to 256 bytes in total (128 bytes for byte references).
6475 Put static or global variables whose size is @var{n} bytes or less into
6476 the small data area that register @code{gp} points to. The small data
6477 area can hold up to 64 kilobytes.
6480 Put static or global variables whose size is @var{n} bytes or less into
6481 the first 32 kilobytes of memory.
6484 Specify that the target processor is the V850.
6487 Generate code suitable for big switch tables. Use this option only if
6488 the assembler/linker complain about out of range branches within a switch
6493 @subsection ARC Options
6496 These options are defined for ARC implementations:
6500 Compile code for little endian mode. This is the default.
6503 Compile code for big endian mode.
6506 Prepend the name of the cpu to all public symbol names.
6507 In multiple-processor systems, there are many ARC variants with different
6508 instruction and register set characteristics. This flag prevents code
6509 compiled for one cpu to be linked with code compiled for another.
6510 No facility exists for handling variants that are "almost identical".
6511 This is an all or nothing option.
6513 @item -mcpu=@var{cpu}
6514 Compile code for ARC variant @var{cpu}.
6515 Which variants are supported depend on the configuration.
6516 All variants support @samp{-mcpu=base}, this is the default.
6518 @item -mtext=@var{text section}
6519 @itemx -mdata=@var{data section}
6520 @itemx -mrodata=@var{readonly data section}
6521 Put functions, data, and readonly data in @var{text section},
6522 @var{data section}, and @var{readonly data section} respectively
6523 by default. This can be overridden with the @code{section} attribute.
6524 @xref{Variable Attributes}.
6529 @subsection NS32K Options
6530 @cindex NS32K options
6532 These are the @samp{-m} options defined for the 32000 series. The default
6533 values for these options depends on which style of 32000 was selected when
6534 the compiler was configured; the defaults for the most common choices are
6540 Generate output for a 32032. This is the default
6541 when the compiler is configured for 32032 and 32016 based systems.
6545 Generate output for a 32332. This is the default
6546 when the compiler is configured for 32332-based systems.
6550 Generate output for a 32532. This is the default
6551 when the compiler is configured for 32532-based systems.
6554 Generate output containing 32081 instructions for floating point.
6555 This is the default for all systems.
6558 Generate output containing 32381 instructions for floating point. This
6559 also implies @samp{-m32081}. The 32381 is only compatible with the 32332
6560 and 32532 cpus. This is the default for the pc532-netbsd configuration.
6563 Try and generate multiply-add floating point instructions @code{polyF}
6564 and @code{dotF}. This option is only available if the @samp{-m32381}
6565 option is in effect. Using these instructions requires changes to to
6566 register allocation which generally has a negative impact on
6567 performance. This option should only be enabled when compiling code
6568 particularly likely to make heavy use of multiply-add instructions.
6571 Do not try and generate multiply-add floating point instructions
6572 @code{polyF} and @code{dotF}. This is the default on all platforms.
6575 Generate output containing library calls for floating point.
6576 @strong{Warning:} the requisite libraries may not be available.
6579 Do not use the bit-field instructions. On some machines it is faster to
6580 use shifting and masking operations. This is the default for the pc532.
6583 Do use the bit-field instructions. This is the default for all platforms
6587 Use a different function-calling convention, in which functions
6588 that take a fixed number of arguments return pop their
6589 arguments on return with the @code{ret} instruction.
6591 This calling convention is incompatible with the one normally
6592 used on Unix, so you cannot use it if you need to call libraries
6593 compiled with the Unix compiler.
6595 Also, you must provide function prototypes for all functions that
6596 take variable numbers of arguments (including @code{printf});
6597 otherwise incorrect code will be generated for calls to those
6600 In addition, seriously incorrect code will result if you call a
6601 function with too many arguments. (Normally, extra arguments are
6602 harmlessly ignored.)
6604 This option takes its name from the 680x0 @code{rtd} instruction.
6608 Use a different function-calling convention where the first two arguments
6609 are passed in registers.
6611 This calling convention is incompatible with the one normally
6612 used on Unix, so you cannot use it if you need to call libraries
6613 compiled with the Unix compiler.
6616 Do not pass any arguments in registers. This is the default for all
6620 It is OK to use the sb as an index register which is always loaded with
6621 zero. This is the default for the pc532-netbsd target.
6624 The sb register is not available for use or has not been initialized to
6625 zero by the run time system. This is the default for all targets except
6626 the pc532-netbsd. It is also implied whenever @samp{-mhimem} or
6627 @samp{-fpic} is set.
6630 Many ns32000 series addressing modes use displacements of up to 512MB.
6631 If an address is above 512MB then displacements from zero can not be used.
6632 This option causes code to be generated which can be loaded above 512MB.
6633 This may be useful for operating systems or ROM code.
6636 Assume code will be loaded in the first 512MB of virtual address space.
6637 This is the default for all platforms.
6644 @node Code Gen Options
6645 @section Options for Code Generation Conventions
6646 @cindex code generation conventions
6647 @cindex options, code generation
6648 @cindex run-time options
6650 These machine-independent options control the interface conventions
6651 used in code generation.
6653 Most of them have both positive and negative forms; the negative form
6654 of @samp{-ffoo} would be @samp{-fno-foo}. In the table below, only
6655 one of the forms is listed---the one which is not the default. You
6656 can figure out the other form by either removing @samp{no-} or adding
6661 Enable exception handling. Generates extra code needed to propagate
6662 exceptions. For some targets, this implies GNU CC will generate frame
6663 unwind information for all functions, which can produce significant data
6664 size overhead, although it does not affect execution. If you do not
6665 specify this option, GNU CC will enable it by default for languages like
6666 C++ which normally require exception handling, and disable itfor
6667 languages like C that do not normally require it. However, you may need
6668 to enable this option when compiling C code that needs to interoperate
6669 properly with exception handlers written in C++. You may also wish to
6670 disable this option if you are compiling older C++ programs that don't
6671 use exception handling.
6673 @item -fpcc-struct-return
6674 Return ``short'' @code{struct} and @code{union} values in memory like
6675 longer ones, rather than in registers. This convention is less
6676 efficient, but it has the advantage of allowing intercallability between
6677 GCC-compiled files and files compiled with other compilers.
6679 The precise convention for returning structures in memory depends
6680 on the target configuration macros.
6682 Short structures and unions are those whose size and alignment match
6683 that of some integer type.
6685 @item -freg-struct-return
6686 Use the convention that @code{struct} and @code{union} values are
6687 returned in registers when possible. This is more efficient for small
6688 structures than @samp{-fpcc-struct-return}.
6690 If you specify neither @samp{-fpcc-struct-return} nor its contrary
6691 @samp{-freg-struct-return}, GCC defaults to whichever convention is
6692 standard for the target. If there is no standard convention, GCC
6693 defaults to @samp{-fpcc-struct-return}, except on targets where GCC
6694 is the principal compiler. In those cases, we can choose the standard,
6695 and we chose the more efficient register return alternative.
6698 Allocate to an @code{enum} type only as many bytes as it needs for the
6699 declared range of possible values. Specifically, the @code{enum} type
6700 will be equivalent to the smallest integer type which has enough room.
6702 @item -fshort-double
6703 Use the same size for @code{double} as for @code{float}.
6706 Requests that the data and non-@code{const} variables of this
6707 compilation be shared data rather than private data. The distinction
6708 makes sense only on certain operating systems, where shared data is
6709 shared between processes running the same program, while private data
6710 exists in one copy per process.
6713 Allocate even uninitialized global variables in the data section of the
6714 object file, rather than generating them as common blocks. This has the
6715 effect that if the same variable is declared (without @code{extern}) in
6716 two different compilations, you will get an error when you link them.
6717 The only reason this might be useful is if you wish to verify that the
6718 program will work on other systems which always work this way.
6721 Ignore the @samp{#ident} directive.
6723 @item -fno-gnu-linker
6724 Do not output global initializations (such as C++ constructors and
6725 destructors) in the form used by the GNU linker (on systems where the GNU
6726 linker is the standard method of handling them). Use this option when
6727 you want to use a non-GNU linker, which also requires using the
6728 @code{collect2} program to make sure the system linker includes
6729 constructors and destructors. (@code{collect2} is included in the GCC
6730 distribution.) For systems which @emph{must} use @code{collect2}, the
6731 compiler driver @code{gcc} is configured to do this automatically.
6733 @item -finhibit-size-directive
6734 Don't output a @code{.size} assembler directive, or anything else that
6735 would cause trouble if the function is split in the middle, and the
6736 two halves are placed at locations far apart in memory. This option is
6737 used when compiling @file{crtstuff.c}; you should not need to use it
6741 Put extra commentary information in the generated assembly code to
6742 make it more readable. This option is generally only of use to those
6743 who actually need to read the generated assembly code (perhaps while
6744 debugging the compiler itself).
6746 @samp{-fno-verbose-asm}, the default, causes the
6747 extra information to be omitted and is useful when comparing two assembler
6751 Consider all memory references through pointers to be volatile.
6753 @item -fvolatile-global
6754 Consider all memory references to extern and global data items to
6755 be volatile. GCC does not consider static data items to be volatile
6756 because of this switch.
6758 @item -fvolatile-static
6759 Consider all memory references to static data to be volatile.
6762 @cindex global offset table
6764 Generate position-independent code (PIC) suitable for use in a shared
6765 library, if supported for the target machine. Such code accesses all
6766 constant addresses through a global offset table (GOT). The dynamic
6767 loader resolves the GOT entries when the program starts (the dynamic
6768 loader is not part of GCC; it is part of the operating system). If
6769 the GOT size for the linked executable exceeds a machine-specific
6770 maximum size, you get an error message from the linker indicating that
6771 @samp{-fpic} does not work; in that case, recompile with @samp{-fPIC}
6772 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
6773 on the m68k and RS/6000. The 386 has no such limit.)
6775 Position-independent code requires special support, and therefore works
6776 only on certain machines. For the 386, GCC supports PIC for System V
6777 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
6778 position-independent.
6781 If supported for the target machine, emit position-independent code,
6782 suitable for dynamic linking and avoiding any limit on the size of the
6783 global offset table. This option makes a difference on the m68k, m88k,
6786 Position-independent code requires special support, and therefore works
6787 only on certain machines.
6789 @item -ffixed-@var{reg}
6790 Treat the register named @var{reg} as a fixed register; generated code
6791 should never refer to it (except perhaps as a stack pointer, frame
6792 pointer or in some other fixed role).
6794 @var{reg} must be the name of a register. The register names accepted
6795 are machine-specific and are defined in the @code{REGISTER_NAMES}
6796 macro in the machine description macro file.
6798 This flag does not have a negative form, because it specifies a
6801 @item -fcall-used-@var{reg}
6802 Treat the register named @var{reg} as an allocable register that is
6803 clobbered by function calls. It may be allocated for temporaries or
6804 variables that do not live across a call. Functions compiled this way
6805 will not save and restore the register @var{reg}.
6807 It is an error to used this flag with the frame pointer or stack pointer.
6808 Use of this flag for other registers that have fixed pervasive roles in
6809 the machine's execution model will produce disastrous results.
6811 This flag does not have a negative form, because it specifies a
6814 @item -fcall-saved-@var{reg}
6815 Treat the register named @var{reg} as an allocable register saved by
6816 functions. It may be allocated even for temporaries or variables that
6817 live across a call. Functions compiled this way will save and restore
6818 the register @var{reg} if they use it.
6820 It is an error to used this flag with the frame pointer or stack pointer.
6821 Use of this flag for other registers that have fixed pervasive roles in
6822 the machine's execution model will produce disastrous results.
6824 A different sort of disaster will result from the use of this flag for
6825 a register in which function values may be returned.
6827 This flag does not have a negative form, because it specifies a
6831 Pack all structure members together without holes. Usually you would
6832 not want to use this option, since it makes the code suboptimal, and
6833 the offsets of structure members won't agree with system libraries.
6835 @item -fcheck-memory-usage
6836 Generate extra code to check each memory access. GCC will generate
6837 code that is suitable for a detector of bad memory accesses such as
6840 Normally, you should compile all, or none, of your code with this option.
6842 If you do mix code compiled with and without this option,
6843 you must ensure that all code that has side effects
6844 and that is called by code compiled with this option
6845 is, itself, compiled with this option.
6846 If you do not, you might get erroneous messages from the detector.
6848 If you use functions from a library that have side-effects (such as
6849 @code{read}), you might not be able to recompile the library and
6850 specify this option. In that case, you can enable the
6851 @samp{-fprefix-function-name} option, which requests GCC to encapsulate
6852 your code and make other functions look as if they were compiled with
6853 @samp{-fcheck-memory-usage}. This is done by calling ``stubs'',
6854 which are provided by the detector. If you cannot find or build
6855 stubs for every function you call, you might have to specify
6856 @samp{-fcheck-memory-usage} without @samp{-fprefix-function-name}.
6858 If you specify this option, you can not use the @code{asm} or
6859 @code{__asm__} keywords in functions with memory checking enabled. GNU
6860 CC cannot understand what the @code{asm} statement may do, and therefore
6861 cannot generate the appropriate code, so it will reject it. However, if
6862 you specify the function attribute @code{no_check_memory_usage} (see
6863 @pxref{Function Attributes}, GNU CC will disable memory checking within a
6864 function; you may use @code{asm} statements inside such functions. You
6865 may have an inline expansion of a non-checked function within a checked
6866 function; in that case GNU CC will not generate checks for the inlined
6867 function's memory accesses.
6869 If you move your @code{asm} statements to non-checked inline functions
6870 and they do access memory, you can add calls to the support code in your
6871 inline function, to indicate any reads, writes, or copies being done.
6872 These calls would be similar to those done in the stubs described above.
6874 @item -fprefix-function-name
6875 Request GCC to add a prefix to the symbols generated for function names.
6876 GCC adds a prefix to the names of functions defined as well as
6877 functions called. Code compiled with this option and code compiled
6878 without the option can't be linked together, unless stubs are used.
6880 If you compile the following code with @samp{-fprefix-function-name}
6882 extern void bar (int);
6891 GCC will compile the code as if it was written:
6893 extern void prefix_bar (int);
6897 return prefix_bar (a + 5);
6900 This option is designed to be used with @samp{-fcheck-memory-usage}.
6902 @item -finstrument-functions
6903 Generate instrumentation calls for entry and exit to functions. Just
6904 after function entry and just before function exit, the following
6905 profiling functions will be called with the address of the current
6906 function and its call site. (On some platforms,
6907 @code{__builtin_return_address} does not work beyond the current
6908 function, so the call site information may not be available to the
6909 profiling functions otherwise.)
6912 void __cyg_profile_func_enter (void *this_fn, void *call_site);
6913 void __cyg_profile_func_exit (void *this_fn, void *call_site);
6916 The first argument is the address of the start of the current function,
6917 which may be looked up exactly in the symbol table.
6919 This instrumentation is also done for functions expanded inline in other
6920 functions. The profiling calls will indicate where, conceptually, the
6921 inline function is entered and exited. This means that addressable
6922 versions of such functions must be available. If all your uses of a
6923 function are expanded inline, this may mean an additional expansion of
6924 code size. If you use @samp{extern inline} in your C code, an
6925 addressable version of such functions must be provided. (This is
6926 normally the case anyways, but if you get lucky and the optimizer always
6927 expands the functions inline, you might have gotten away without
6928 providing static copies.)
6930 A function may be given the attribute @code{no_instrument_function}, in
6931 which case this instrumentation will not be done. This can be used, for
6932 example, for the profiling functions listed above, high-priority
6933 interrupt routines, and any functions from which the profiling functions
6934 cannot safely be called (perhaps signal handlers, if the profiling
6935 routines generate output or allocate memory).
6938 Generate code to verify that you do not go beyond the boundary of the
6939 stack. You should specify this flag if you are running in an
6940 environment with multiple threads, but only rarely need to specify it in
6941 a single-threaded environment since stack overflow is automatically
6942 detected on nearly all systems if there is only one stack.
6944 @cindex aliasing of parameters
6945 @cindex parameters, aliased
6946 @item -fargument-alias
6947 @itemx -fargument-noalias
6948 @itemx -fargument-noalias-global
6949 Specify the possible relationships among parameters and between
6950 parameters and global data.
6952 @samp{-fargument-alias} specifies that arguments (parameters) may
6953 alias each other and may alias global storage.
6954 @samp{-fargument-noalias} specifies that arguments do not alias
6955 each other, but may alias global storage.
6956 @samp{-fargument-noalias-global} specifies that arguments do not
6957 alias each other and do not alias global storage.
6959 Each language will automatically use whatever option is required by
6960 the language standard. You should not need to use these options yourself.
6962 @item -fleading-underscore
6963 This option and its counterpart, -fno-leading-underscore, forcibly
6964 change the way C symbols are represented in the object file. One use
6965 is to help link with legacy assembly code.
6967 Be warned that you should know what you are doing when invoking this
6968 option, and that not all targets provide complete support for it.
6971 @node Environment Variables
6972 @section Environment Variables Affecting GCC
6973 @cindex environment variables
6975 This section describes several environment variables that affect how GCC
6976 operates. Some of them work by specifying directories or prefixes to use
6977 when searching for various kinds of files. Some are used to specify other
6978 aspects of the compilation environment.
6981 Note that you can also specify places to search using options such as
6982 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
6983 take precedence over places specified using environment variables, which
6984 in turn take precedence over those specified by the configuration of GCC.
6988 Note that you can also specify places to search using options such as
6989 @samp{-B}, @samp{-I} and @samp{-L} (@pxref{Directory Options}). These
6990 take precedence over places specified using environment variables, which
6991 in turn take precedence over those specified by the configuration of GCC.
6998 @c @itemx LC_COLLATE
7000 @c @itemx LC_MONETARY
7001 @c @itemx LC_NUMERIC
7006 @c @findex LC_COLLATE
7008 @c @findex LC_MONETARY
7009 @c @findex LC_NUMERIC
7013 These environment variables control the way that GCC uses
7014 localization information that allow GCC to work with different
7015 national conventions. GCC inspects the locale categories
7016 @code{LC_CTYPE} and @code{LC_MESSAGES} if it has been configured to do
7017 so. These locale categories can be set to any value supported by your
7018 installation. A typical value is @samp{en_UK} for English in the United
7021 The @code{LC_CTYPE} environment variable specifies character
7022 classification. GCC uses it to determine the character boundaries in
7023 a string; this is needed for some multibyte encodings that contain quote
7024 and escape characters that would otherwise be interpreted as a string
7027 The @code{LC_MESSAGES} environment variable specifies the language to
7028 use in diagnostic messages.
7030 If the @code{LC_ALL} environment variable is set, it overrides the value
7031 of @code{LC_CTYPE} and @code{LC_MESSAGES}; otherwise, @code{LC_CTYPE}
7032 and @code{LC_MESSAGES} default to the value of the @code{LANG}
7033 environment variable. If none of these variables are set, GCC
7034 defaults to traditional C English behavior.
7038 If @code{TMPDIR} is set, it specifies the directory to use for temporary
7039 files. GCC uses temporary files to hold the output of one stage of
7040 compilation which is to be used as input to the next stage: for example,
7041 the output of the preprocessor, which is the input to the compiler
7044 @item GCC_EXEC_PREFIX
7045 @findex GCC_EXEC_PREFIX
7046 If @code{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
7047 names of the subprograms executed by the compiler. No slash is added
7048 when this prefix is combined with the name of a subprogram, but you can
7049 specify a prefix that ends with a slash if you wish.
7051 If GCC cannot find the subprogram using the specified prefix, it
7052 tries looking in the usual places for the subprogram.
7054 The default value of @code{GCC_EXEC_PREFIX} is
7055 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
7056 of @code{prefix} when you ran the @file{configure} script.
7058 Other prefixes specified with @samp{-B} take precedence over this prefix.
7060 This prefix is also used for finding files such as @file{crt0.o} that are
7063 In addition, the prefix is used in an unusual way in finding the
7064 directories to search for header files. For each of the standard
7065 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
7066 (more precisely, with the value of @code{GCC_INCLUDE_DIR}), GCC tries
7067 replacing that beginning with the specified prefix to produce an
7068 alternate directory name. Thus, with @samp{-Bfoo/}, GCC will search
7069 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
7070 These alternate directories are searched first; the standard directories
7074 @findex COMPILER_PATH
7075 The value of @code{COMPILER_PATH} is a colon-separated list of
7076 directories, much like @code{PATH}. GCC tries the directories thus
7077 specified when searching for subprograms, if it can't find the
7078 subprograms using @code{GCC_EXEC_PREFIX}.
7081 @findex LIBRARY_PATH
7082 The value of @code{LIBRARY_PATH} is a colon-separated list of
7083 directories, much like @code{PATH}. When configured as a native compiler,
7084 GCC tries the directories thus specified when searching for special
7085 linker files, if it can't find them using @code{GCC_EXEC_PREFIX}. Linking
7086 using GCC also uses these directories when searching for ordinary
7087 libraries for the @samp{-l} option (but directories specified with
7088 @samp{-L} come first).
7090 @item C_INCLUDE_PATH
7091 @itemx CPLUS_INCLUDE_PATH
7092 @itemx OBJC_INCLUDE_PATH
7093 @findex C_INCLUDE_PATH
7094 @findex CPLUS_INCLUDE_PATH
7095 @findex OBJC_INCLUDE_PATH
7096 @c @itemx OBJCPLUS_INCLUDE_PATH
7097 These environment variables pertain to particular languages. Each
7098 variable's value is a colon-separated list of directories, much like
7099 @code{PATH}. When GCC searches for header files, it tries the
7100 directories listed in the variable for the language you are using, after
7101 the directories specified with @samp{-I} but before the standard header
7104 @item DEPENDENCIES_OUTPUT
7105 @findex DEPENDENCIES_OUTPUT
7106 @cindex dependencies for make as output
7107 If this variable is set, its value specifies how to output dependencies
7108 for Make based on the header files processed by the compiler. This
7109 output looks much like the output from the @samp{-M} option
7110 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
7111 in addition to the usual results of compilation.
7113 The value of @code{DEPENDENCIES_OUTPUT} can be just a file name, in
7114 which case the Make rules are written to that file, guessing the target
7115 name from the source file name. Or the value can have the form
7116 @samp{@var{file} @var{target}}, in which case the rules are written to
7117 file @var{file} using @var{target} as the target name.
7121 @cindex locale definition
7122 This variable is used to pass locale information to the compiler. One way in
7123 which this information is used is to determine the character set to be used
7124 when character literals, string literals and comments are parsed in C and C++.
7125 When the compiler is configured to allow multibyte characters,
7126 the following values for @code{LANG} are recognized:
7130 Recognize JIS characters.
7132 Recognize SJIS characters.
7134 Recognize EUCJP characters.
7137 If @code{LANG} is not defined, or if it has some other value, then the
7138 compiler will use mblen and mbtowc as defined by the default locale to
7139 recognize and translate multibyte characters.
7142 @node Running Protoize
7143 @section Running Protoize
7145 The program @code{protoize} is an optional part of GNU C. You can use
7146 it to add prototypes to a program, thus converting the program to ANSI
7147 C in one respect. The companion program @code{unprotoize} does the
7148 reverse: it removes argument types from any prototypes that are found.
7150 When you run these programs, you must specify a set of source files as
7151 command line arguments. The conversion programs start out by compiling
7152 these files to see what functions they define. The information gathered
7153 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
7155 After scanning comes actual conversion. The specified files are all
7156 eligible to be converted; any files they include (whether sources or
7157 just headers) are eligible as well.
7159 But not all the eligible files are converted. By default,
7160 @code{protoize} and @code{unprotoize} convert only source and header
7161 files in the current directory. You can specify additional directories
7162 whose files should be converted with the @samp{-d @var{directory}}
7163 option. You can also specify particular files to exclude with the
7164 @samp{-x @var{file}} option. A file is converted if it is eligible, its
7165 directory name matches one of the specified directory names, and its
7166 name within the directory has not been excluded.
7168 Basic conversion with @code{protoize} consists of rewriting most
7169 function definitions and function declarations to specify the types of
7170 the arguments. The only ones not rewritten are those for varargs
7173 @code{protoize} optionally inserts prototype declarations at the
7174 beginning of the source file, to make them available for any calls that
7175 precede the function's definition. Or it can insert prototype
7176 declarations with block scope in the blocks where undeclared functions
7179 Basic conversion with @code{unprotoize} consists of rewriting most
7180 function declarations to remove any argument types, and rewriting
7181 function definitions to the old-style pre-ANSI form.
7183 Both conversion programs print a warning for any function declaration or
7184 definition that they can't convert. You can suppress these warnings
7187 The output from @code{protoize} or @code{unprotoize} replaces the
7188 original source file. The original file is renamed to a name ending
7189 with @samp{.save}. If the @samp{.save} file already exists, then
7190 the source file is simply discarded.
7192 @code{protoize} and @code{unprotoize} both depend on GCC itself to
7193 scan the program and collect information about the functions it uses.
7194 So neither of these programs will work until GCC is installed.
7196 Here is a table of the options you can use with @code{protoize} and
7197 @code{unprotoize}. Each option works with both programs unless
7201 @item -B @var{directory}
7202 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
7203 usual directory (normally @file{/usr/local/lib}). This file contains
7204 prototype information about standard system functions. This option
7205 applies only to @code{protoize}.
7207 @item -c @var{compilation-options}
7208 Use @var{compilation-options} as the options when running @code{gcc} to
7209 produce the @samp{.X} files. The special option @samp{-aux-info} is
7210 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
7212 Note that the compilation options must be given as a single argument to
7213 @code{protoize} or @code{unprotoize}. If you want to specify several
7214 @code{gcc} options, you must quote the entire set of compilation options
7215 to make them a single word in the shell.
7217 There are certain @code{gcc} arguments that you cannot use, because they
7218 would produce the wrong kind of output. These include @samp{-g},
7219 @samp{-O}, @samp{-c}, @samp{-S}, and @samp{-o} If you include these in
7220 the @var{compilation-options}, they are ignored.
7223 Rename files to end in @samp{.C} instead of @samp{.c}.
7224 This is convenient if you are converting a C program to C++.
7225 This option applies only to @code{protoize}.
7228 Add explicit global declarations. This means inserting explicit
7229 declarations at the beginning of each source file for each function
7230 that is called in the file and was not declared. These declarations
7231 precede the first function definition that contains a call to an
7232 undeclared function. This option applies only to @code{protoize}.
7234 @item -i @var{string}
7235 Indent old-style parameter declarations with the string @var{string}.
7236 This option applies only to @code{protoize}.
7238 @code{unprotoize} converts prototyped function definitions to old-style
7239 function definitions, where the arguments are declared between the
7240 argument list and the initial @samp{@{}. By default, @code{unprotoize}
7241 uses five spaces as the indentation. If you want to indent with just
7242 one space instead, use @samp{-i " "}.
7245 Keep the @samp{.X} files. Normally, they are deleted after conversion
7249 Add explicit local declarations. @code{protoize} with @samp{-l} inserts
7250 a prototype declaration for each function in each block which calls the
7251 function without any declaration. This option applies only to
7255 Make no real changes. This mode just prints information about the conversions
7256 that would have been done without @samp{-n}.
7259 Make no @samp{.save} files. The original files are simply deleted.
7260 Use this option with caution.
7262 @item -p @var{program}
7263 Use the program @var{program} as the compiler. Normally, the name
7267 Work quietly. Most warnings are suppressed.
7270 Print the version number, just like @samp{-v} for @code{gcc}.
7273 If you need special compiler options to compile one of your program's
7274 source files, then you should generate that file's @samp{.X} file
7275 specially, by running @code{gcc} on that source file with the
7276 appropriate options and the option @samp{-aux-info}. Then run
7277 @code{protoize} on the entire set of files. @code{protoize} will use
7278 the existing @samp{.X} file because it is newer than the source file.
7282 gcc -Dfoo=bar file1.c -aux-info
7287 You need to include the special files along with the rest in the
7288 @code{protoize} command, even though their @samp{.X} files already
7289 exist, because otherwise they won't get converted.
7291 @xref{Protoize Caveats}, for more information on how to use
7292 @code{protoize} successfully.