1 @c Copyright (C) 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
2 @c 2000, 2001 Free Software Foundation, Inc.
3 @c This is part of the GCC manual.
4 @c For copying conditions, see the file gcc.texi.
8 Copyright @copyright{} 1988, 1989, 1992, 1993, 1994, 1995, 1996, 1997,
9 1998, 1999, 2000, 2001 Free Software Foundation, Inc.
11 Permission is granted to copy, distribute and/or modify this document
12 under the terms of the GNU Free Documentation License, Version 1.1 or
13 any later version published by the Free Software Foundation; with the
14 Invariant Sections being ``GNU General Public License'' and ``Funding
15 Free Software'', the Front-Cover texts being (a) (see below), and with
16 the Back-Cover Texts being (b) (see below). A copy of the license is
17 included in the gfdl(7) man page.
19 (a) The FSF's Front-Cover Text is:
23 (b) The FSF's Back-Cover Text is:
25 You have freedom to copy and modify this GNU Manual, like GNU
26 software. Copies published by the Free Software Foundation raise
27 funds for GNU development.
29 @c Set file name and title for the man page.
31 @settitle GNU project C and C++ compiler
33 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
34 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
35 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
36 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
37 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
38 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
39 [@option{-o} @var{outfile}] @var{infile}@dots{}
41 Only the most useful options are listed here; see below for the
42 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
45 gpl(7), gfdl(7), fsf-funding(7),
46 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
47 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
48 @file{ld}, @file{binutils} and @file{gdb}.
51 For instructions on reporting bugs, see
52 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
53 script to report bugs is recommended.
56 See the Info entry for @file{gcc}, or
57 @w{@uref{http://gcc.gnu.org/thanks.html}}, for contributors to GCC@.
62 @chapter GCC Command Options
63 @cindex GCC command options
64 @cindex command options
65 @cindex options, GCC command
67 @c man begin DESCRIPTION
69 When you invoke GCC, it normally does preprocessing, compilation,
70 assembly and linking. The ``overall options'' allow you to stop this
71 process at an intermediate stage. For example, the @option{-c} option
72 says not to run the linker. Then the output consists of object files
73 output by the assembler.
75 Other options are passed on to one stage of processing. Some options
76 control the preprocessor and others the compiler itself. Yet other
77 options control the assembler and linker; most of these are not
78 documented here, since you rarely need to use any of them.
80 @cindex C compilation options
81 Most of the command line options that you can use with GCC are useful
82 for C programs; when an option is only useful with another language
83 (usually C++), the explanation says so explicitly. If the description
84 for a particular option does not mention a source language, you can use
85 that option with all supported languages.
87 @cindex C++ compilation options
88 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
89 options for compiling C++ programs.
91 @cindex grouping options
92 @cindex options, grouping
93 The @command{gcc} program accepts options and file names as operands. Many
94 options have multi-letter names; therefore multiple single-letter options
95 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
98 @cindex order of options
99 @cindex options, order
100 You can mix options and other arguments. For the most part, the order
101 you use doesn't matter. Order does matter when you use several options
102 of the same kind; for example, if you specify @option{-L} more than once,
103 the directories are searched in the order specified.
105 Many options have long names starting with @samp{-f} or with
106 @samp{-W}---for example, @option{-fforce-mem},
107 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
108 these have both positive and negative forms; the negative form of
109 @option{-ffoo} would be @option{-fno-foo}. This manual documents
110 only one of these two forms, whichever one is not the default.
114 @xref{Option Index}, for an index to GCC's options.
117 * Option Summary:: Brief list of all options, without explanations.
118 * Overall Options:: Controlling the kind of output:
119 an executable, object files, assembler files,
120 or preprocessed source.
121 * Invoking G++:: Compiling C++ programs.
122 * C Dialect Options:: Controlling the variant of C language compiled.
123 * C++ Dialect Options:: Variations on C++.
124 * Objective-C Dialect Options:: Variations on Objective-C.
125 * Language Independent Options:: Controlling how diagnostics should be
127 * Warning Options:: How picky should the compiler be?
128 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
129 * Optimize Options:: How much optimization?
130 * Preprocessor Options:: Controlling header files and macro definitions.
131 Also, getting dependency information for Make.
132 * Assembler Options:: Passing options to the assembler.
133 * Link Options:: Specifying libraries and so on.
134 * Directory Options:: Where to find header files and libraries.
135 Where to find the compiler executable files.
136 * Spec Files:: How to pass switches to sub-processes.
137 * Target Options:: Running a cross-compiler, or an old version of GCC.
138 * Submodel Options:: Specifying minor hardware or convention variations,
139 such as 68010 vs 68020.
140 * Code Gen Options:: Specifying conventions for function calls, data layout
142 * Environment Variables:: Env vars that affect GCC.
143 * Running Protoize:: Automatically adding or removing function prototypes.
149 @section Option Summary
151 Here is a summary of all the options, grouped by type. Explanations are
152 in the following sections.
155 @item Overall Options
156 @xref{Overall Options,,Options Controlling the Kind of Output}.
158 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language} @gol
159 -v --target-help --help}
161 @item C Language Options
162 @xref{C Dialect Options,,Options Controlling C Dialect}.
164 -ansi -std=@var{standard} -aux-info @var{filename} @gol
165 -fno-asm -fno-builtin @gol
166 -fhosted -ffreestanding @gol
167 -trigraphs -traditional -traditional-cpp @gol
168 -fallow-single-precision -fcond-mismatch @gol
169 -fsigned-bitfields -fsigned-char @gol
170 -funsigned-bitfields -funsigned-char @gol
171 -fwritable-strings -fshort-wchar}
173 @item C++ Language Options
174 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
176 -fno-access-control -fcheck-new -fconserve-space @gol
177 -fno-const-strings -fdollars-in-identifiers @gol
178 -fno-elide-constructors @gol
179 -fno-enforce-eh-specs -fexternal-templates @gol
180 -falt-external-templates @gol
181 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
182 -fno-implicit-templates @gol
183 -fno-implicit-inline-templates @gol
184 -fno-implement-inlines -fms-extensions @gol
185 -fno-nonansi-builtins -fno-operator-names @gol
186 -fno-optional-diags -fpermissive @gol
187 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
188 -fuse-cxa-atexit -fvtable-gc -fno-weak -nostdinc++ @gol
189 -fno-default-inline -Wctor-dtor-privacy @gol
190 -Wnon-virtual-dtor -Wreorder @gol
191 -Weffc++ -Wno-deprecated @gol
192 -Wno-non-template-friend -Wold-style-cast @gol
193 -Woverloaded-virtual -Wno-pmf-conversions @gol
194 -Wsign-promo -Wsynth}
196 @item Objective-C Language Options
197 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
199 -fconstant-string-class=@var{class-name} @gol
200 -fgnu-runtime -fnext-runtime -gen-decls @gol
201 -Wno-protocol -Wselector}
203 @item Language Independent Options
204 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
206 -fmessage-length=@var{n} @gol
207 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
209 @item Warning Options
210 @xref{Warning Options,,Options to Request or Suppress Warnings}.
212 -fsyntax-only -pedantic -pedantic-errors @gol
213 -w -W -Wall -Waggregate-return @gol
214 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
215 -Wconversion -Wdisabled-optimization -Wdiv-by-zero -Werror @gol
216 -Wfloat-equal -Wformat -Wformat=2 @gol
217 -Wformat-nonliteral -Wformat-security @gol
218 -Wimplicit -Wimplicit-int @gol
219 -Wimplicit-function-declaration @gol
220 -Werror-implicit-function-declaration @gol
221 -Wimport -Winline @gol
222 -Wlarger-than-@var{len} -Wlong-long @gol
223 -Wmain -Wmissing-braces -Wmissing-declarations @gol
224 -Wmissing-format-attribute -Wmissing-noreturn @gol
225 -Wmultichar -Wno-format-extra-args -Wno-format-y2k @gol
226 -Wno-import -Wpacked -Wpadded @gol
227 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
228 -Wreturn-type -Wsequence-point -Wshadow @gol
229 -Wsign-compare -Wswitch -Wsystem-headers @gol
230 -Wtrigraphs -Wundef -Wuninitialized @gol
231 -Wunknown-pragmas -Wunreachable-code @gol
232 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
233 -Wunused-value -Wunused-variable -Wwrite-strings}
235 @item C-only Warning Options
237 -Wbad-function-cast -Wmissing-prototypes -Wnested-externs @gol
238 -Wstrict-prototypes -Wtraditional}
240 @item Debugging Options
241 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
243 -a -ax -d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
245 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
246 -fdump-tree-original@r{[}-@var{n}@r{]} -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
247 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
248 -fmem-report -fpretend-float @gol
249 -fprofile-arcs -ftest-coverage -ftime-report @gol
250 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
251 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+ @gol
252 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
253 -print-multi-directory -print-multi-lib @gol
254 -print-prog-name=@var{program} -print-search-dirs -Q @gol
257 @item Optimization Options
258 @xref{Optimize Options,,Options that Control Optimization}.
260 -falign-functions=@var{n} -falign-jumps=@var{n} @gol
261 -falign-labels=@var{n} -falign-loops=@var{n} @gol
262 -fbranch-probabilities -fcaller-saves @gol
263 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
264 -fdelayed-branch -fdelete-null-pointer-checks @gol
265 -fexpensive-optimizations -ffast-math -ffloat-store @gol
266 -fforce-addr -fforce-mem -ffunction-sections @gol
267 -fgcse -fgcse-lm -fgcse-sm @gol
268 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
269 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
270 -fmove-all-movables -fno-default-inline -fno-defer-pop @gol
271 -fno-function-cse -fno-guess-branch-probability @gol
272 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
273 -funsafe-math-optimizations -fno-trapping-math @gol
274 -fomit-frame-pointer -foptimize-register-move @gol
275 -foptimize-sibling-calls -freduce-all-givs @gol
276 -fregmove -frename-registers @gol
277 -frerun-cse-after-loop -frerun-loop-opt @gol
278 -fschedule-insns -fschedule-insns2 @gol
279 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
280 -fstrength-reduce -fstrict-aliasing -fthread-jumps -ftrapv @gol
281 -funroll-all-loops -funroll-loops @gol
282 --param @var{name}=@var{value}
283 -O -O0 -O1 -O2 -O3 -Os}
285 @item Preprocessor Options
286 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
288 -$ -A@var{question}=@var{answer} -A-@var{question}@r{[}=@var{answer}@r{]} @gol
289 -C -dD -dI -dM -dN @gol
290 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
291 -idirafter @var{dir} @gol
292 -include @var{file} -imacros @var{file} @gol
293 -iprefix @var{file} -iwithprefix @var{dir} @gol
294 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
295 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
296 -trigraphs -undef -U@var{macro} -Wp,@var{option}}
298 @item Assembler Option
299 @xref{Assembler Options,,Passing Options to the Assembler}.
304 @xref{Link Options,,Options for Linking}.
306 @var{object-file-name} -l@var{library} @gol
307 -nostartfiles -nodefaultlibs -nostdlib @gol
308 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
309 -Wl,@var{option} -Xlinker @var{option} @gol
312 @item Directory Options
313 @xref{Directory Options,,Options for Directory Search}.
315 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
318 @c I wrote this xref this way to avoid overfull hbox. -- rms
319 @xref{Target Options}.
321 -b @var{machine} -V @var{version}}
323 @item Machine Dependent Options
324 @xref{Submodel Options,,Hardware Models and Configurations}.
325 @emph{M680x0 Options}
327 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
328 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
329 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
330 -malign-int -mstrict-align}
332 @emph{M68hc1x Options}
334 -m6811 -m6812 -m68hc11 -m68hc12 @gol
335 -mauto-incdec -mshort -msoft-reg-count=@var{count}}
343 -mcpu=@var{cpu-type} @gol
344 -mtune=@var{cpu-type} @gol
345 -mcmodel=@var{code-model} @gol
347 -mapp-regs -mbroken-saverestore -mcypress @gol
348 -mepilogue -mfaster-structs -mflat @gol
349 -mfpu -mhard-float -mhard-quad-float @gol
350 -mimpure-text -mlive-g0 -mno-app-regs @gol
351 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu @gol
352 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
353 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
354 -msupersparc -munaligned-doubles -mv8}
356 @emph{Convex Options}
358 -mc1 -mc2 -mc32 -mc34 -mc38 @gol
359 -margcount -mnoargcount @gol
360 -mlong32 -mlong64 @gol
361 -mvolatile-cache -mvolatile-nocache}
363 @emph{AMD29K Options}
365 -m29000 -m29050 -mbw -mnbw -mdw -mndw @gol
366 -mlarge -mnormal -msmall @gol
367 -mkernel-registers -mno-reuse-arg-regs @gol
368 -mno-stack-check -mno-storem-bug @gol
369 -mreuse-arg-regs -msoft-float -mstack-check @gol
370 -mstorem-bug -muser-registers}
374 -mapcs-frame -mno-apcs-frame @gol
375 -mapcs-26 -mapcs-32 @gol
376 -mapcs-stack-check -mno-apcs-stack-check @gol
377 -mapcs-float -mno-apcs-float @gol
378 -mapcs-reentrant -mno-apcs-reentrant @gol
379 -msched-prolog -mno-sched-prolog @gol
380 -mlittle-endian -mbig-endian -mwords-little-endian @gol
381 -malignment-traps -mno-alignment-traps @gol
382 -msoft-float -mhard-float -mfpe @gol
383 -mthumb-interwork -mno-thumb-interwork @gol
384 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
385 -mstructure-size-boundary=@var{n} @gol
386 -mbsd -mxopen -mno-symrename @gol
387 -mabort-on-noreturn @gol
388 -mlong-calls -mno-long-calls @gol
389 -msingle-pic-base -mno-single-pic-base @gol
390 -mpic-register=@var{reg} @gol
391 -mnop-fun-dllimport @gol
392 -mpoke-function-name @gol
394 -mtpcs-frame -mtpcs-leaf-frame @gol
395 -mcaller-super-interworking -mcallee-super-interworking }
397 @emph{MN10200 Options}
401 @emph{MN10300 Options}
403 -mmult-bug -mno-mult-bug @gol
404 -mam33 -mno-am33 @gol
407 @emph{M32R/D Options}
409 -mm32rx -mcode-model=@var{model-type} -msdata=@var{sdata-type} @gol
414 -m88000 -m88100 -m88110 -mbig-pic @gol
415 -mcheck-zero-division -mhandle-large-shift @gol
416 -midentify-revision -mno-check-zero-division @gol
417 -mno-ocs-debug-info -mno-ocs-frame-position @gol
418 -mno-optimize-arg-area -mno-serialize-volatile @gol
419 -mno-underscores -mocs-debug-info @gol
420 -mocs-frame-position -moptimize-arg-area @gol
421 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
422 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
423 -mversion-03.00 -mwarn-passed-structs}
425 @emph{RS/6000 and PowerPC Options}
427 -mcpu=@var{cpu-type} @gol
428 -mtune=@var{cpu-type} @gol
429 -mpower -mno-power -mpower2 -mno-power2 @gol
430 -mpowerpc -mpowerpc64 -mno-powerpc @gol
431 -maltivec -mno-altivec @gol
432 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
433 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
434 -mnew-mnemonics -mold-mnemonics @gol
435 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc @gol
436 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe @gol
437 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
438 -mstring -mno-string -mupdate -mno-update @gol
439 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
440 -mstrict-align -mno-strict-align -mrelocatable @gol
441 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
442 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
443 -mcall-aix -mcall-sysv -mcall-netbsd @gol
445 -mprototype -mno-prototype @gol
446 -msim -mmvme -mads -myellowknife -memb -msdata @gol
447 -msdata=@var{opt} -mvxworks -G @var{num}}
451 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
452 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
453 -mminimum-fp-blocks -mnohc-struct-return}
457 -mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
458 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
459 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
460 -mgas -mgp32 -mgp64 @gol
461 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
462 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
463 -mmips-as -mmips-tfile -mno-abicalls @gol
464 -mno-embedded-data -mno-uninit-const-in-rodata @gol
465 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
466 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
467 -mrnames -msoft-float @gol
468 -m4650 -msingle-float -mmad @gol
469 -mstats -EL -EB -G @var{num} -nocpp @gol
470 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
471 -mfix7000 -mno-crt0 -mflush-func=@var{func} -mno-flush-func}
473 @emph{i386 and x86-64 Options}
475 -mcpu=@var{cpu-type} -march=@var{cpu-type} @gol
476 -mintel-syntax -mieee-fp -mno-fancy-math-387 @gol
477 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
478 -mno-wide-multiply -mrtd -malign-double @gol
479 -mpreferred-stack-boundary=@var{num} @gol
480 -mmmx -msse -m3dnow @gol
481 -mthreads -mno-align-stringops -minline-all-stringops @gol
482 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
483 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
489 -march=@var{architecture-type} @gol
490 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
491 -mfast-indirect-calls -mgas -mjump-in-delay @gol
492 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
493 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
494 -mno-jump-in-delay -mno-long-load-store @gol
495 -mno-portable-runtime -mno-soft-float @gol
496 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
497 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
498 -mschedule=@var{cpu-type} -mspace-regs}
500 @emph{Intel 960 Options}
502 -m@var{cpu-type} -masm-compat -mclean-linkage @gol
503 -mcode-align -mcomplex-addr -mleaf-procedures @gol
504 -mic-compat -mic2.0-compat -mic3.0-compat @gol
505 -mintel-asm -mno-clean-linkage -mno-code-align @gol
506 -mno-complex-addr -mno-leaf-procedures @gol
507 -mno-old-align -mno-strict-align -mno-tail-call @gol
508 -mnumerics -mold-align -msoft-float -mstrict-align @gol
511 @emph{DEC Alpha Options}
513 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float @gol
514 -malpha-as -mgas @gol
515 -mieee -mieee-with-inexact -mieee-conformant @gol
516 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
517 -mtrap-precision=@var{mode} -mbuild-constants @gol
518 -mcpu=@var{cpu-type} @gol
519 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max @gol
520 -mmemory-latency=@var{time}}
522 @emph{Clipper Options}
526 @emph{H8/300 Options}
528 -mrelax -mh -ms -mint32 -malign-300}
532 -m1 -m2 -m3 -m3e @gol
533 -m4-nofpu -m4-single-only -m4-single -m4 @gol
534 -mb -ml -mdalign -mrelax @gol
535 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
536 -mieee -misize -mpadstruct -mspace @gol
537 -mprefergot -musermode}
539 @emph{System V Options}
541 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
546 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
547 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
549 @emph{TMS320C3x/C4x Options}
551 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
552 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
553 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
554 -mparallel-insns -mparallel-mpy -mpreserve-float}
558 -mlong-calls -mno-long-calls -mep -mno-ep @gol
559 -mprolog-function -mno-prolog-function -mspace @gol
560 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
565 -m32032 -m32332 -m32532 -m32081 -m32381 @gol
566 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
567 -mregparam -mnoregparam -msb -mnosb @gol
568 -mbitfield -mnobitfield -mhimem -mnohimem}
572 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
573 -mcall-prologues -mno-tablejump -mtiny-stack}
577 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
578 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
579 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
580 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
581 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
585 -mlibfuncs -mno-libfuncs -mepsilon -mno-epsilon -mabi=gnu @gol
586 -mabi=mmixware -mzero-extend -mknuthdiv -mtoplevel-symbols @gol
591 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
592 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
593 -mconstant-gp -mauto-pic -minline-divide-min-latency @gol
594 -minline-divide-max-throughput -mno-dwarf2-asm @gol
595 -mfixed-range=@var{register-range}}
597 @emph{S/390 and zSeries Options}
599 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
600 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
601 -m64 -m31 -mdebug -mno-debug}
605 -mcpu=@var{cpu} -march=@var{cpu} -mtune=@var{cpu} @gol
606 -mmax-stack-frame=@var{n} -melinux-stacksize=@var{n} @gol
607 -metrax4 -metrax100 -mpdebug -mcc-init -mno-side-effects @gol
608 -mstack-align -mdata-align -mconst-align @gol
609 -m32-bit -m16-bit -m8-bit -mno-prologue-epilogue -mno-gotplt @gol
610 -melf -maout -melinux -mlinux -sim -sim2}
612 @item Code Generation Options
613 @xref{Code Gen Options,,Options for Code Generation Conventions}.
615 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
616 -ffixed-@var{reg} -fexceptions @gol
617 -fnon-call-exceptions -funwind-tables -fasynchronous-unwind-tables @gol
618 -finhibit-size-directive -finstrument-functions @gol
619 -fcheck-memory-usage -fprefix-function-name @gol
620 -fno-common -fno-ident -fno-gnu-linker @gol
621 -fpcc-struct-return -fpic -fPIC @gol
622 -freg-struct-return -fshared-data -fshort-enums @gol
623 -fshort-double -fvolatile @gol
624 -fvolatile-global -fvolatile-static @gol
625 -fverbose-asm -fpack-struct -fstack-check @gol
626 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
627 -fargument-alias -fargument-noalias @gol
628 -fargument-noalias-global -fleading-underscore}
632 * Overall Options:: Controlling the kind of output:
633 an executable, object files, assembler files,
634 or preprocessed source.
635 * C Dialect Options:: Controlling the variant of C language compiled.
636 * C++ Dialect Options:: Variations on C++.
637 * Objective-C Dialect Options:: Variations on Objective-C.
638 * Language Independent Options:: Controlling how diagnostics should be
640 * Warning Options:: How picky should the compiler be?
641 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
642 * Optimize Options:: How much optimization?
643 * Preprocessor Options:: Controlling header files and macro definitions.
644 Also, getting dependency information for Make.
645 * Assembler Options:: Passing options to the assembler.
646 * Link Options:: Specifying libraries and so on.
647 * Directory Options:: Where to find header files and libraries.
648 Where to find the compiler executable files.
649 * Spec Files:: How to pass switches to sub-processes.
650 * Target Options:: Running a cross-compiler, or an old version of GCC.
653 @node Overall Options
654 @section Options Controlling the Kind of Output
656 Compilation can involve up to four stages: preprocessing, compilation
657 proper, assembly and linking, always in that order. The first three
658 stages apply to an individual source file, and end by producing an
659 object file; linking combines all the object files (those newly
660 compiled, and those specified as input) into an executable file.
662 @cindex file name suffix
663 For any given input file, the file name suffix determines what kind of
668 C source code which must be preprocessed.
671 C source code which should not be preprocessed.
674 C++ source code which should not be preprocessed.
677 Objective-C source code. Note that you must link with the library
678 @file{libobjc.a} to make an Objective-C program work.
681 Objective-C source code which should not be preprocessed.
684 C header file (not to be compiled or linked).
688 @itemx @var{file}.cxx
689 @itemx @var{file}.cpp
690 @itemx @var{file}.c++
692 C++ source code which must be preprocessed. Note that in @samp{.cxx},
693 the last two letters must both be literally @samp{x}. Likewise,
694 @samp{.C} refers to a literal capital C@.
697 @itemx @var{file}.for
698 @itemx @var{file}.FOR
699 Fortran source code which should not be preprocessed.
702 @itemx @var{file}.fpp
703 @itemx @var{file}.FPP
704 Fortran source code which must be preprocessed (with the traditional
708 Fortran source code which must be preprocessed with a RATFOR
709 preprocessor (not included with GCC)@.
711 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
712 Using and Porting GNU Fortran}, for more details of the handling of
715 @c FIXME: Descriptions of Java file types.
722 Ada source code file which contains a library unit declaration (a
723 declaration of a package, subprogram, or generic, or a generic
724 instantiation), or a library unit renaming declaration (a package,
725 generic, or subprogram renaming declaration). Such files are also
728 @itemx @var{file}.adb
729 Ada source code file containing a library unit body (a subprogram or
730 package body). Such files are also called @dfn{bodies}.
732 @c GCC also knows about some suffixes for languages not yet included:
738 @itemx @var{file}.chi
739 CHILL source code (preprocessed with the traditional preprocessor).
745 Assembler code which must be preprocessed.
748 An object file to be fed straight into linking.
749 Any file name with no recognized suffix is treated this way.
753 You can specify the input language explicitly with the @option{-x} option:
756 @item -x @var{language}
757 Specify explicitly the @var{language} for the following input files
758 (rather than letting the compiler choose a default based on the file
759 name suffix). This option applies to all following input files until
760 the next @option{-x} option. Possible values for @var{language} are:
762 c c-header cpp-output
764 objective-c objc-cpp-output
765 assembler assembler-with-cpp
768 f77 f77-cpp-input ratfor
773 Turn off any specification of a language, so that subsequent files are
774 handled according to their file name suffixes (as they are if @option{-x}
775 has not been used at all).
777 @item -pass-exit-codes
778 @opindex pass-exit-codes
779 Normally the @command{gcc} program will exit with the code of 1 if any
780 phase of the compiler returns a non-success return code. If you specify
781 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
782 numerically highest error produced by any phase that returned an error
786 If you only want some of the stages of compilation, you can use
787 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
788 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
789 @command{gcc} is to stop. Note that some combinations (for example,
790 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
795 Compile or assemble the source files, but do not link. The linking
796 stage simply is not done. The ultimate output is in the form of an
797 object file for each source file.
799 By default, the object file name for a source file is made by replacing
800 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
802 Unrecognized input files, not requiring compilation or assembly, are
807 Stop after the stage of compilation proper; do not assemble. The output
808 is in the form of an assembler code file for each non-assembler input
811 By default, the assembler file name for a source file is made by
812 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
814 Input files that don't require compilation are ignored.
818 Stop after the preprocessing stage; do not run the compiler proper. The
819 output is in the form of preprocessed source code, which is sent to the
822 Input files which don't require preprocessing are ignored.
824 @cindex output file option
827 Place output in file @var{file}. This applies regardless to whatever
828 sort of output is being produced, whether it be an executable file,
829 an object file, an assembler file or preprocessed C code.
831 Since only one output file can be specified, it does not make sense to
832 use @option{-o} when compiling more than one input file, unless you are
833 producing an executable file as output.
835 If @option{-o} is not specified, the default is to put an executable file
836 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
837 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
838 all preprocessed C source on standard output.
842 Print (on standard error output) the commands executed to run the stages
843 of compilation. Also print the version number of the compiler driver
844 program and of the preprocessor and the compiler proper.
848 Use pipes rather than temporary files for communication between the
849 various stages of compilation. This fails to work on some systems where
850 the assembler is unable to read from a pipe; but the GNU assembler has
855 Print (on the standard output) a description of the command line options
856 understood by @command{gcc}. If the @option{-v} option is also specified
857 then @option{--help} will also be passed on to the various processes
858 invoked by @command{gcc}, so that they can display the command line options
859 they accept. If the @option{-W} option is also specified then command
860 line options which have no documentation associated with them will also
865 Print (on the standard output) a description of target specific command
866 line options for each tool.
870 @section Compiling C++ Programs
872 @cindex suffixes for C++ source
873 @cindex C++ source file suffixes
874 C++ source files conventionally use one of the suffixes @samp{.C},
875 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
876 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
877 files with these names and compiles them as C++ programs even if you
878 call the compiler the same way as for compiling C programs (usually with
879 the name @command{gcc}).
883 However, C++ programs often require class libraries as well as a
884 compiler that understands the C++ language---and under some
885 circumstances, you might want to compile programs from standard input,
886 or otherwise without a suffix that flags them as C++ programs.
887 @command{g++} is a program that calls GCC with the default language
888 set to C++, and automatically specifies linking against the C++
889 library. On many systems, @command{g++} is also
890 installed with the name @command{c++}.
892 @cindex invoking @command{g++}
893 When you compile C++ programs, you may specify many of the same
894 command-line options that you use for compiling programs in any
895 language; or command-line options meaningful for C and related
896 languages; or options that are meaningful only for C++ programs.
897 @xref{C Dialect Options,,Options Controlling C Dialect}, for
898 explanations of options for languages related to C@.
899 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
900 explanations of options that are meaningful only for C++ programs.
902 @node C Dialect Options
903 @section Options Controlling C Dialect
904 @cindex dialect options
905 @cindex language dialect options
906 @cindex options, dialect
908 The following options control the dialect of C (or languages derived
909 from C, such as C++ and Objective-C) that the compiler accepts:
916 In C mode, support all ISO C89 programs. In C++ mode,
917 remove GNU extensions that conflict with ISO C++.
919 This turns off certain features of GCC that are incompatible with ISO
920 C89 (when compiling C code), or of standard C++ (when compiling C++ code),
921 such as the @code{asm} and @code{typeof} keywords, and
922 predefined macros such as @code{unix} and @code{vax} that identify the
923 type of system you are using. It also enables the undesirable and
924 rarely used ISO trigraph feature. For the C compiler,
925 it disables recognition of C++ style @samp{//} comments as well as
926 the @code{inline} keyword.
928 The alternate keywords @code{__asm__}, @code{__extension__},
929 @code{__inline__} and @code{__typeof__} continue to work despite
930 @option{-ansi}. You would not want to use them in an ISO C program, of
931 course, but it is useful to put them in header files that might be included
932 in compilations done with @option{-ansi}. Alternate predefined macros
933 such as @code{__unix__} and @code{__vax__} are also available, with or
934 without @option{-ansi}.
936 The @option{-ansi} option does not cause non-ISO programs to be
937 rejected gratuitously. For that, @option{-pedantic} is required in
938 addition to @option{-ansi}. @xref{Warning Options}.
940 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
941 option is used. Some header files may notice this macro and refrain
942 from declaring certain functions or defining certain macros that the
943 ISO standard doesn't call for; this is to avoid interfering with any
944 programs that might use these names for other things.
946 Functions which would normally be built in but do not have semantics
947 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
948 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
949 built-in functions provided by GCC}, for details of the functions
954 Determine the language standard. This option is currently only
955 supported when compiling C@. A value for this option must be provided;
961 ISO C89 (same as @option{-ansi}).
964 ISO C89 as modified in amendment 1.
970 ISO C99. Note that this standard is not yet fully supported; see
971 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
972 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
975 Default, ISO C89 plus GNU extensions (including some C99 features).
979 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
980 this will become the default. The name @samp{gnu9x} is deprecated.
984 Even when this option is not specified, you can still use some of the
985 features of newer standards in so far as they do not conflict with
986 previous C standards. For example, you may use @code{__restrict__} even
987 when @option{-std=c99} is not specified.
989 The @option{-std} options specifying some version of ISO C have the same
990 effects as @option{-ansi}, except that features that were not in ISO C89
991 but are in the specified version (for example, @samp{//} comments and
992 the @code{inline} keyword in ISO C99) are not disabled.
994 @xref{Standards,,Language Standards Supported by GCC}, for details of
995 these standard versions.
997 @item -aux-info @var{filename}
999 Output to the given filename prototyped declarations for all functions
1000 declared and/or defined in a translation unit, including those in header
1001 files. This option is silently ignored in any language other than C@.
1003 Besides declarations, the file indicates, in comments, the origin of
1004 each declaration (source file and line), whether the declaration was
1005 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
1006 @samp{O} for old, respectively, in the first character after the line
1007 number and the colon), and whether it came from a declaration or a
1008 definition (@samp{C} or @samp{F}, respectively, in the following
1009 character). In the case of function definitions, a K&R-style list of
1010 arguments followed by their declarations is also provided, inside
1011 comments, after the declaration.
1015 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
1016 keyword, so that code can use these words as identifiers. You can use
1017 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
1018 instead. @option{-ansi} implies @option{-fno-asm}.
1020 In C++, this switch only affects the @code{typeof} keyword, since
1021 @code{asm} and @code{inline} are standard keywords. You may want to
1022 use the @option{-fno-gnu-keywords} flag instead, which has the same
1023 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
1024 switch only affects the @code{asm} and @code{typeof} keywords, since
1025 @code{inline} is a standard keyword in ISO C99.
1028 @opindex fno-builtin
1029 @cindex built-in functions
1030 Don't recognize built-in functions that do not begin with
1031 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1032 functions provided by GCC}, for details of the functions affected,
1033 including those which are not built-in functions when @option{-ansi} or
1034 @option{-std} options for strict ISO C conformance are used because they
1035 do not have an ISO standard meaning.
1037 GCC normally generates special code to handle certain built-in functions
1038 more efficiently; for instance, calls to @code{alloca} may become single
1039 instructions that adjust the stack directly, and calls to @code{memcpy}
1040 may become inline copy loops. The resulting code is often both smaller
1041 and faster, but since the function calls no longer appear as such, you
1042 cannot set a breakpoint on those calls, nor can you change the behavior
1043 of the functions by linking with a different library.
1045 In C++, @option{-fno-builtin} is always in effect. The @option{-fbuiltin}
1046 option has no effect. Therefore, in C++, the only way to get the
1047 optimization benefits of built-in functions is to call the function
1048 using the @samp{__builtin_} prefix. The GNU C++ Standard Library uses
1049 built-in functions to implement many functions (like
1050 @code{std::strchr}), so that you automatically get efficient code.
1054 @cindex hosted environment
1056 Assert that compilation takes place in a hosted environment. This implies
1057 @option{-fbuiltin}. A hosted environment is one in which the
1058 entire standard library is available, and in which @code{main} has a return
1059 type of @code{int}. Examples are nearly everything except a kernel.
1060 This is equivalent to @option{-fno-freestanding}.
1062 @item -ffreestanding
1063 @opindex ffreestanding
1064 @cindex hosted environment
1066 Assert that compilation takes place in a freestanding environment. This
1067 implies @option{-fno-builtin}. A freestanding environment
1068 is one in which the standard library may not exist, and program startup may
1069 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1070 This is equivalent to @option{-fno-hosted}.
1072 @xref{Standards,,Language Standards Supported by GCC}, for details of
1073 freestanding and hosted environments.
1077 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1078 options for strict ISO C conformance) implies @option{-trigraphs}.
1080 @cindex traditional C language
1081 @cindex C language, traditional
1083 @opindex traditional
1084 Attempt to support some aspects of traditional C compilers.
1089 All @code{extern} declarations take effect globally even if they
1090 are written inside of a function definition. This includes implicit
1091 declarations of functions.
1094 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
1095 and @code{volatile} are not recognized. (You can still use the
1096 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
1100 Comparisons between pointers and integers are always allowed.
1103 Integer types @code{unsigned short} and @code{unsigned char} promote
1104 to @code{unsigned int}.
1107 Out-of-range floating point literals are not an error.
1110 Certain constructs which ISO regards as a single invalid preprocessing
1111 number, such as @samp{0xe-0xd}, are treated as expressions instead.
1114 String ``constants'' are not necessarily constant; they are stored in
1115 writable space, and identical looking constants are allocated
1116 separately. (This is the same as the effect of
1117 @option{-fwritable-strings}.)
1119 @cindex @code{longjmp} and automatic variables
1121 All automatic variables not declared @code{register} are preserved by
1122 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1123 not declared @code{volatile} may be clobbered.
1128 @cindex escape sequences, traditional
1129 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1130 literal characters @samp{x} and @samp{a} respectively. Without
1131 @w{@option{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1132 representation of a character, and @samp{\a} produces a bell.
1135 You may wish to use @option{-fno-builtin} as well as @option{-traditional}
1136 if your program uses names that are normally GNU C built-in functions for
1137 other purposes of its own.
1139 You cannot use @option{-traditional} if you include any header files that
1140 rely on ISO C features. Some vendors are starting to ship systems with
1141 ISO C header files and you cannot use @option{-traditional} on such
1142 systems to compile files that include any system headers.
1144 The @option{-traditional} option also enables @option{-traditional-cpp}.
1146 @item -traditional-cpp
1147 @opindex traditional-cpp
1148 Attempt to support some aspects of traditional C preprocessors.
1149 See the GNU CPP manual for details.
1151 @item -fcond-mismatch
1152 @opindex fcond-mismatch
1153 Allow conditional expressions with mismatched types in the second and
1154 third arguments. The value of such an expression is void. This option
1155 is not supported for C++.
1157 @item -funsigned-char
1158 @opindex funsigned-char
1159 Let the type @code{char} be unsigned, like @code{unsigned char}.
1161 Each kind of machine has a default for what @code{char} should
1162 be. It is either like @code{unsigned char} by default or like
1163 @code{signed char} by default.
1165 Ideally, a portable program should always use @code{signed char} or
1166 @code{unsigned char} when it depends on the signedness of an object.
1167 But many programs have been written to use plain @code{char} and
1168 expect it to be signed, or expect it to be unsigned, depending on the
1169 machines they were written for. This option, and its inverse, let you
1170 make such a program work with the opposite default.
1172 The type @code{char} is always a distinct type from each of
1173 @code{signed char} or @code{unsigned char}, even though its behavior
1174 is always just like one of those two.
1177 @opindex fsigned-char
1178 Let the type @code{char} be signed, like @code{signed char}.
1180 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1181 the negative form of @option{-funsigned-char}. Likewise, the option
1182 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1184 @item -fsigned-bitfields
1185 @itemx -funsigned-bitfields
1186 @itemx -fno-signed-bitfields
1187 @itemx -fno-unsigned-bitfields
1188 @opindex fsigned-bitfields
1189 @opindex funsigned-bitfields
1190 @opindex fno-signed-bitfields
1191 @opindex fno-unsigned-bitfields
1192 These options control whether a bit-field is signed or unsigned, when the
1193 declaration does not use either @code{signed} or @code{unsigned}. By
1194 default, such a bit-field is signed, because this is consistent: the
1195 basic integer types such as @code{int} are signed types.
1197 However, when @option{-traditional} is used, bit-fields are all unsigned
1200 @item -fwritable-strings
1201 @opindex fwritable-strings
1202 Store string constants in the writable data segment and don't uniquize
1203 them. This is for compatibility with old programs which assume they can
1204 write into string constants. The option @option{-traditional} also has
1207 Writing into string constants is a very bad idea; ``constants'' should
1210 @item -fallow-single-precision
1211 @opindex fallow-single-precision
1212 Do not promote single precision math operations to double precision,
1213 even when compiling with @option{-traditional}.
1215 Traditional K&R C promotes all floating point operations to double
1216 precision, regardless of the sizes of the operands. On the
1217 architecture for which you are compiling, single precision may be faster
1218 than double precision. If you must use @option{-traditional}, but want
1219 to use single precision operations when the operands are single
1220 precision, use this option. This option has no effect when compiling
1221 with ISO or GNU C conventions (the default).
1224 @opindex fshort-wchar
1225 Override the underlying type for @samp{wchar_t} to be @samp{short
1226 unsigned int} instead of the default for the target. This option is
1227 useful for building programs to run under WINE@.
1230 @node C++ Dialect Options
1231 @section Options Controlling C++ Dialect
1233 @cindex compiler options, C++
1234 @cindex C++ options, command line
1235 @cindex options, C++
1236 This section describes the command-line options that are only meaningful
1237 for C++ programs; but you can also use most of the GNU compiler options
1238 regardless of what language your program is in. For example, you
1239 might compile a file @code{firstClass.C} like this:
1242 g++ -g -frepo -O -c firstClass.C
1246 In this example, only @option{-frepo} is an option meant
1247 only for C++ programs; you can use the other options with any
1248 language supported by GCC@.
1250 Here is a list of options that are @emph{only} for compiling C++ programs:
1253 @item -fno-access-control
1254 @opindex fno-access-control
1255 Turn off all access checking. This switch is mainly useful for working
1256 around bugs in the access control code.
1260 Check that the pointer returned by @code{operator new} is non-null
1261 before attempting to modify the storage allocated. The current Working
1262 Paper requires that @code{operator new} never return a null pointer, so
1263 this check is normally unnecessary.
1265 An alternative to using this option is to specify that your
1266 @code{operator new} does not throw any exceptions; if you declare it
1267 @samp{throw()}, G++ will check the return value. See also @samp{new
1270 @item -fconserve-space
1271 @opindex fconserve-space
1272 Put uninitialized or runtime-initialized global variables into the
1273 common segment, as C does. This saves space in the executable at the
1274 cost of not diagnosing duplicate definitions. If you compile with this
1275 flag and your program mysteriously crashes after @code{main()} has
1276 completed, you may have an object that is being destroyed twice because
1277 two definitions were merged.
1279 This option is no longer useful on most targets, now that support has
1280 been added for putting variables into BSS without making them common.
1282 @item -fno-const-strings
1283 @opindex fno-const-strings
1284 Give string constants type @code{char *} instead of type @code{const
1285 char *}. By default, G++ uses type @code{const char *} as required by
1286 the standard. Even if you use @option{-fno-const-strings}, you cannot
1287 actually modify the value of a string constant, unless you also use
1288 @option{-fwritable-strings}.
1290 This option might be removed in a future release of G++. For maximum
1291 portability, you should structure your code so that it works with
1292 string constants that have type @code{const char *}.
1294 @item -fdollars-in-identifiers
1295 @opindex fdollars-in-identifiers
1296 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1297 @samp{$} with the option @option{-fno-dollars-in-identifiers}. (GNU C allows
1298 @samp{$} by default on most target systems, but there are a few exceptions.)
1299 Traditional C allowed the character @samp{$} to form part of
1300 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1302 @item -fno-elide-constructors
1303 @opindex fno-elide-constructors
1304 The C++ standard allows an implementation to omit creating a temporary
1305 which is only used to initialize another object of the same type.
1306 Specifying this option disables that optimization, and forces G++ to
1307 call the copy constructor in all cases.
1309 @item -fno-enforce-eh-specs
1310 @opindex fno-enforce-eh-specs
1311 Don't check for violation of exception specifications at runtime. This
1312 option violates the C++ standard, but may be useful for reducing code
1313 size in production builds, much like defining @samp{NDEBUG}. The compiler
1314 will still optimize based on the exception specifications.
1316 @item -fexternal-templates
1317 @opindex fexternal-templates
1319 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1320 template instantiation; template instances are emitted or not according
1321 to the location of the template definition. @xref{Template
1322 Instantiation}, for more information.
1324 This option is deprecated.
1326 @item -falt-external-templates
1327 @opindex falt-external-templates
1328 Similar to @option{-fexternal-templates}, but template instances are
1329 emitted or not according to the place where they are first instantiated.
1330 @xref{Template Instantiation}, for more information.
1332 This option is deprecated.
1335 @itemx -fno-for-scope
1337 @opindex fno-for-scope
1338 If @option{-ffor-scope} is specified, the scope of variables declared in
1339 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1340 as specified by the C++ standard.
1341 If @option{-fno-for-scope} is specified, the scope of variables declared in
1342 a @i{for-init-statement} extends to the end of the enclosing scope,
1343 as was the case in old versions of G++, and other (traditional)
1344 implementations of C++.
1346 The default if neither flag is given to follow the standard,
1347 but to allow and give a warning for old-style code that would
1348 otherwise be invalid, or have different behavior.
1350 @item -fno-gnu-keywords
1351 @opindex fno-gnu-keywords
1352 Do not recognize @code{typeof} as a keyword, so that code can use this
1353 word as an identifier. You can use the keyword @code{__typeof__} instead.
1354 @option{-ansi} implies @option{-fno-gnu-keywords}.
1356 @item -fno-implicit-templates
1357 @opindex fno-implicit-templates
1358 Never emit code for non-inline templates which are instantiated
1359 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1360 @xref{Template Instantiation}, for more information.
1362 @item -fno-implicit-inline-templates
1363 @opindex fno-implicit-inline-templates
1364 Don't emit code for implicit instantiations of inline templates, either.
1365 The default is to handle inlines differently so that compiles with and
1366 without optimization will need the same set of explicit instantiations.
1368 @item -fno-implement-inlines
1369 @opindex fno-implement-inlines
1370 To save space, do not emit out-of-line copies of inline functions
1371 controlled by @samp{#pragma implementation}. This will cause linker
1372 errors if these functions are not inlined everywhere they are called.
1374 @item -fms-extensions
1375 @opindex fms-extensions
1376 Disable pedantic warnings about constructs used in MFC, such as implicit
1377 int and getting a pointer to member function via non-standard syntax.
1379 @item -fno-nonansi-builtins
1380 @opindex fno-nonansi-builtins
1381 Disable built-in declarations of functions that are not mandated by
1382 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1383 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1385 @item -fno-operator-names
1386 @opindex fno-operator-names
1387 Do not treat the operator name keywords @code{and}, @code{bitand},
1388 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1389 synonyms as keywords.
1391 @item -fno-optional-diags
1392 @opindex fno-optional-diags
1393 Disable diagnostics that the standard says a compiler does not need to
1394 issue. Currently, the only such diagnostic issued by G++ is the one for
1395 a name having multiple meanings within a class.
1398 @opindex fpermissive
1399 Downgrade messages about nonconformant code from errors to warnings. By
1400 default, G++ effectively sets @option{-pedantic-errors} without
1401 @option{-pedantic}; this option reverses that. This behavior and this
1402 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1406 Enable automatic template instantiation at link time. This option also
1407 implies @option{-fno-implicit-templates}. @xref{Template
1408 Instantiation}, for more information.
1412 Disable generation of information about every class with virtual
1413 functions for use by the C++ runtime type identification features
1414 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1415 of the language, you can save some space by using this flag. Note that
1416 exception handling uses the same information, but it will generate it as
1421 Emit statistics about front-end processing at the end of the compilation.
1422 This information is generally only useful to the G++ development team.
1424 @item -ftemplate-depth-@var{n}
1425 @opindex ftemplate-depth
1426 Set the maximum instantiation depth for template classes to @var{n}.
1427 A limit on the template instantiation depth is needed to detect
1428 endless recursions during template class instantiation. ANSI/ISO C++
1429 conforming programs must not rely on a maximum depth greater than 17.
1431 @item -fuse-cxa-atexit
1432 @opindex fuse-cxa-atexit
1433 Register destructors for objects with static storage duration with the
1434 @code{__cxa_atexit} function rather than the @code{atexit} function.
1435 This option is required for fully standards-compliant handling of static
1436 destructors, but will only work if your C library supports
1437 @code{__cxa_atexit}.
1441 Emit special relocations for vtables and virtual function references
1442 so that the linker can identify unused virtual functions and zero out
1443 vtable slots that refer to them. This is most useful with
1444 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1445 also discard the functions themselves.
1447 This optimization requires GNU as and GNU ld. Not all systems support
1448 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1452 Do not use weak symbol support, even if it is provided by the linker.
1453 By default, G++ will use weak symbols if they are available. This
1454 option exists only for testing, and should not be used by end-users;
1455 it will result in inferior code and has no benefits. This option may
1456 be removed in a future release of G++.
1460 Do not search for header files in the standard directories specific to
1461 C++, but do still search the other standard directories. (This option
1462 is used when building the C++ library.)
1465 In addition, these optimization, warning, and code generation options
1466 have meanings only for C++ programs:
1469 @item -fno-default-inline
1470 @opindex fno-default-inline
1471 Do not assume @samp{inline} for functions defined inside a class scope.
1472 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1473 functions will have linkage like inline functions; they just won't be
1476 @item -Wctor-dtor-privacy @r{(C++ only)}
1477 @opindex Wctor-dtor-privacy
1478 Warn when a class seems unusable, because all the constructors or
1479 destructors in a class are private and the class has no friends or
1480 public static member functions.
1482 @item -Wnon-virtual-dtor @r{(C++ only)}
1483 @opindex Wnon-virtual-dtor
1484 Warn when a class declares a non-virtual destructor that should probably
1485 be virtual, because it looks like the class will be used polymorphically.
1487 @item -Wreorder @r{(C++ only)}
1489 @cindex reordering, warning
1490 @cindex warning for reordering of member initializers
1491 Warn when the order of member initializers given in the code does not
1492 match the order in which they must be executed. For instance:
1498 A(): j (0), i (1) @{ @}
1502 Here the compiler will warn that the member initializers for @samp{i}
1503 and @samp{j} will be rearranged to match the declaration order of the
1507 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1510 @item -Weffc++ @r{(C++ only)}
1512 Warn about violations of various style guidelines from Scott Meyers'
1513 @cite{Effective C++} books. If you use this option, you should be aware
1514 that the standard library headers do not obey all of these guidelines;
1515 you can use @samp{grep -v} to filter out those warnings.
1517 @item -Wno-deprecated @r{(C++ only)}
1518 @opindex Wno-deprecated
1519 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1521 @item -Wno-non-template-friend @r{(C++ only)}
1522 @opindex Wno-non-template-friend
1523 Disable warnings when non-templatized friend functions are declared
1524 within a template. With the advent of explicit template specification
1525 support in G++, if the name of the friend is an unqualified-id (i.e.,
1526 @samp{friend foo(int)}), the C++ language specification demands that the
1527 friend declare or define an ordinary, nontemplate function. (Section
1528 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1529 could be interpreted as a particular specialization of a templatized
1530 function. Because this non-conforming behavior is no longer the default
1531 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1532 check existing code for potential trouble spots, and is on by default.
1533 This new compiler behavior can be turned off with
1534 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1535 but disables the helpful warning.
1537 @item -Wold-style-cast @r{(C++ only)}
1538 @opindex Wold-style-cast
1539 Warn if an old-style (C-style) cast is used within a C++ program. The
1540 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1541 @samp{const_cast}) are less vulnerable to unintended effects, and much
1544 @item -Woverloaded-virtual @r{(C++ only)}
1545 @opindex Woverloaded-virtual
1546 @cindex overloaded virtual fn, warning
1547 @cindex warning for overloaded virtual fn
1548 Warn when a function declaration hides virtual functions from a
1549 base class. For example, in:
1556 struct B: public A @{
1561 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1569 will fail to compile.
1571 @item -Wno-pmf-conversions @r{(C++ only)}
1572 @opindex Wno-pmf-conversions
1573 Disable the diagnostic for converting a bound pointer to member function
1576 @item -Wsign-promo @r{(C++ only)}
1577 @opindex Wsign-promo
1578 Warn when overload resolution chooses a promotion from unsigned or
1579 enumeral type to a signed type over a conversion to an unsigned type of
1580 the same size. Previous versions of G++ would try to preserve
1581 unsignedness, but the standard mandates the current behavior.
1583 @item -Wsynth @r{(C++ only)}
1585 @cindex warning for synthesized methods
1586 @cindex synthesized methods, warning
1587 Warn when G++'s synthesis behavior does not match that of cfront. For
1593 A& operator = (int);
1603 In this example, G++ will synthesize a default @samp{A& operator =
1604 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1607 @node Objective-C Dialect Options
1608 @section Options Controlling Objective-C Dialect
1610 @cindex compiler options, Objective-C
1611 @cindex Objective-C options, command line
1612 @cindex options, Objective-C
1613 This section describes the command-line options that are only meaningful
1614 for Objective-C programs; but you can also use most of the GNU compiler
1615 options regardless of what language your program is in. For example,
1616 you might compile a file @code{some_class.m} like this:
1619 gcc -g -fgnu-runtime -O -c some_class.m
1623 In this example, only @option{-fgnu-runtime} is an option meant only for
1624 Objective-C programs; you can use the other options with any language
1627 Here is a list of options that are @emph{only} for compiling Objective-C
1631 @item -fconstant-string-class=@var{class-name}
1632 @opindex fconstant-string-class
1633 Use @var{class-name} as the name of the class to instantiate for each
1634 literal string specified with the syntax @code{@@"@dots{}"}. The default
1635 class name is @code{NXConstantString}.
1638 @opindex fgnu-runtime
1639 Generate object code compatible with the standard GNU Objective-C
1640 runtime. This is the default for most types of systems.
1642 @item -fnext-runtime
1643 @opindex fnext-runtime
1644 Generate output compatible with the NeXT runtime. This is the default
1645 for NeXT-based systems, including Darwin and Mac OS X@.
1649 Dump interface declarations for all classes seen in the source file to a
1650 file named @file{@var{sourcename}.decl}.
1653 @opindex Wno-protocol
1654 Do not warn if methods required by a protocol are not implemented
1655 in the class adopting it.
1659 Warn if a selector has multiple methods of different types defined.
1661 @c not documented because only avail via -Wp
1662 @c @item -print-objc-runtime-info
1666 @node Language Independent Options
1667 @section Options to Control Diagnostic Messages Formatting
1668 @cindex options to control diagnostics formatting
1669 @cindex diagnostic messages
1670 @cindex message formatting
1672 Traditionally, diagnostic messages have been formatted irrespective of
1673 the output device's aspect (e.g.@: its width, @dots{}). The options described
1674 below can be used to control the diagnostic messages formatting
1675 algorithm, e.g.@: how many characters per line, how often source location
1676 information should be reported. Right now, only the C++ front end can
1677 honor these options. However it is expected, in the near future, that
1678 the remaining front ends would be able to digest them correctly.
1681 @item -fmessage-length=@var{n}
1682 @opindex fmessage-length
1683 Try to format error messages so that they fit on lines of about @var{n}
1684 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1685 the front ends supported by GCC@. If @var{n} is zero, then no
1686 line-wrapping will be done; each error message will appear on a single
1689 @opindex fdiagnostics-show-location
1690 @item -fdiagnostics-show-location=once
1691 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1692 reporter to emit @emph{once} source location information; that is, in
1693 case the message is too long to fit on a single physical line and has to
1694 be wrapped, the source location won't be emitted (as prefix) again,
1695 over and over, in subsequent continuation lines. This is the default
1698 @item -fdiagnostics-show-location=every-line
1699 Only meaningful in line-wrapping mode. Instructs the diagnostic
1700 messages reporter to emit the same source location information (as
1701 prefix) for physical lines that result from the process of breaking
1702 a message which is too long to fit on a single line.
1706 @node Warning Options
1707 @section Options to Request or Suppress Warnings
1708 @cindex options to control warnings
1709 @cindex warning messages
1710 @cindex messages, warning
1711 @cindex suppressing warnings
1713 Warnings are diagnostic messages that report constructions which
1714 are not inherently erroneous but which are risky or suggest there
1715 may have been an error.
1717 You can request many specific warnings with options beginning @samp{-W},
1718 for example @option{-Wimplicit} to request warnings on implicit
1719 declarations. Each of these specific warning options also has a
1720 negative form beginning @samp{-Wno-} to turn off warnings;
1721 for example, @option{-Wno-implicit}. This manual lists only one of the
1722 two forms, whichever is not the default.
1724 These options control the amount and kinds of warnings produced by GCC:
1727 @cindex syntax checking
1729 @opindex fsyntax-only
1730 Check the code for syntax errors, but don't do anything beyond that.
1734 Issue all the warnings demanded by strict ISO C and ISO C++;
1735 reject all programs that use forbidden extensions, and some other
1736 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1737 version of the ISO C standard specified by any @option{-std} option used.
1739 Valid ISO C and ISO C++ programs should compile properly with or without
1740 this option (though a rare few will require @option{-ansi} or a
1741 @option{-std} option specifying the required version of ISO C)@. However,
1742 without this option, certain GNU extensions and traditional C and C++
1743 features are supported as well. With this option, they are rejected.
1745 @option{-pedantic} does not cause warning messages for use of the
1746 alternate keywords whose names begin and end with @samp{__}. Pedantic
1747 warnings are also disabled in the expression that follows
1748 @code{__extension__}. However, only system header files should use
1749 these escape routes; application programs should avoid them.
1750 @xref{Alternate Keywords}.
1752 Some users try to use @option{-pedantic} to check programs for strict ISO
1753 C conformance. They soon find that it does not do quite what they want:
1754 it finds some non-ISO practices, but not all---only those for which
1755 ISO C @emph{requires} a diagnostic, and some others for which
1756 diagnostics have been added.
1758 A feature to report any failure to conform to ISO C might be useful in
1759 some instances, but would require considerable additional work and would
1760 be quite different from @option{-pedantic}. We don't have plans to
1761 support such a feature in the near future.
1763 Where the standard specified with @option{-std} represents a GNU
1764 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1765 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1766 extended dialect is based. Warnings from @option{-pedantic} are given
1767 where they are required by the base standard. (It would not make sense
1768 for such warnings to be given only for features not in the specified GNU
1769 C dialect, since by definition the GNU dialects of C include all
1770 features the compiler supports with the given option, and there would be
1771 nothing to warn about.)
1773 @item -pedantic-errors
1774 @opindex pedantic-errors
1775 Like @option{-pedantic}, except that errors are produced rather than
1780 Inhibit all warning messages.
1784 Inhibit warning messages about the use of @samp{#import}.
1786 @item -Wchar-subscripts
1787 @opindex Wchar-subscripts
1788 Warn if an array subscript has type @code{char}. This is a common cause
1789 of error, as programmers often forget that this type is signed on some
1794 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1795 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1799 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1800 the arguments supplied have types appropriate to the format string
1801 specified, and that the conversions specified in the format string make
1802 sense. This includes standard functions, and others specified by format
1803 attributes (@pxref{Function Attributes}), in the @code{printf},
1804 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1805 not in the C standard) families.
1807 The formats are checked against the format features supported by GNU
1808 libc version 2.2. These include all ISO C89 and C99 features, as well
1809 as features from the Single Unix Specification and some BSD and GNU
1810 extensions. Other library implementations may not support all these
1811 features; GCC does not support warning about features that go beyond a
1812 particular library's limitations. However, if @option{-pedantic} is used
1813 with @option{-Wformat}, warnings will be given about format features not
1814 in the selected standard version (but not for @code{strfmon} formats,
1815 since those are not in any version of the C standard). @xref{C Dialect
1816 Options,,Options Controlling C Dialect}.
1818 @option{-Wformat} is included in @option{-Wall}. For more control over some
1819 aspects of format checking, the options @option{-Wno-format-y2k},
1820 @option{-Wno-format-extra-args}, @option{-Wformat-nonliteral},
1821 @option{-Wformat-security} and @option{-Wformat=2} are available, but are
1822 not included in @option{-Wall}.
1824 @item -Wno-format-y2k
1825 @opindex Wno-format-y2k
1826 If @option{-Wformat} is specified, do not warn about @code{strftime}
1827 formats which may yield only a two-digit year.
1829 @item -Wno-format-extra-args
1830 @opindex Wno-format-extra-args
1831 If @option{-Wformat} is specified, do not warn about excess arguments to a
1832 @code{printf} or @code{scanf} format function. The C standard specifies
1833 that such arguments are ignored.
1835 @item -Wformat-nonliteral
1836 @opindex Wformat-nonliteral
1837 If @option{-Wformat} is specified, also warn if the format string is not a
1838 string literal and so cannot be checked, unless the format function
1839 takes its format arguments as a @code{va_list}.
1841 @item -Wformat-security
1842 @opindex Wformat-security
1843 If @option{-Wformat} is specified, also warn about uses of format
1844 functions that represent possible security problems. At present, this
1845 warns about calls to @code{printf} and @code{scanf} functions where the
1846 format string is not a string literal and there are no format arguments,
1847 as in @code{printf (foo);}. This may be a security hole if the format
1848 string came from untrusted input and contains @samp{%n}. (This is
1849 currently a subset of what @option{-Wformat-nonliteral} warns about, but
1850 in future warnings may be added to @option{-Wformat-security} that are not
1851 included in @option{-Wformat-nonliteral}.)
1855 Enable @option{-Wformat} plus format checks not included in
1856 @option{-Wformat}. Currently equivalent to @samp{-Wformat
1857 -Wformat-nonliteral -Wformat-security}.
1859 @item -Wimplicit-int
1860 @opindex Wimplicit-int
1861 Warn when a declaration does not specify a type.
1863 @item -Wimplicit-function-declaration
1864 @itemx -Werror-implicit-function-declaration
1865 @opindex Wimplicit-function-declaration
1866 @opindex Werror-implicit-function-declaration
1867 Give a warning (or error) whenever a function is used before being
1872 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
1876 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1877 function with external linkage, returning int, taking either zero
1878 arguments, two, or three arguments of appropriate types.
1880 @item -Wmissing-braces
1881 @opindex Wmissing-braces
1882 Warn if an aggregate or union initializer is not fully bracketed. In
1883 the following example, the initializer for @samp{a} is not fully
1884 bracketed, but that for @samp{b} is fully bracketed.
1887 int a[2][2] = @{ 0, 1, 2, 3 @};
1888 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
1892 @opindex Wparentheses
1893 Warn if parentheses are omitted in certain contexts, such
1894 as when there is an assignment in a context where a truth value
1895 is expected, or when operators are nested whose precedence people
1896 often get confused about.
1898 Also warn about constructions where there may be confusion to which
1899 @code{if} statement an @code{else} branch belongs. Here is an example of
1914 In C, every @code{else} branch belongs to the innermost possible @code{if}
1915 statement, which in this example is @code{if (b)}. This is often not
1916 what the programmer expected, as illustrated in the above example by
1917 indentation the programmer chose. When there is the potential for this
1918 confusion, GCC will issue a warning when this flag is specified.
1919 To eliminate the warning, add explicit braces around the innermost
1920 @code{if} statement so there is no way the @code{else} could belong to
1921 the enclosing @code{if}. The resulting code would look like this:
1937 @item -Wsequence-point
1938 @opindex Wsequence-point
1939 Warn about code that may have undefined semantics because of violations
1940 of sequence point rules in the C standard.
1942 The C standard defines the order in which expressions in a C program are
1943 evaluated in terms of @dfn{sequence points}, which represent a partial
1944 ordering between the execution of parts of the program: those executed
1945 before the sequence point, and those executed after it. These occur
1946 after the evaluation of a full expression (one which is not part of a
1947 larger expression), after the evaluation of the first operand of a
1948 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1949 function is called (but after the evaluation of its arguments and the
1950 expression denoting the called function), and in certain other places.
1951 Other than as expressed by the sequence point rules, the order of
1952 evaluation of subexpressions of an expression is not specified. All
1953 these rules describe only a partial order rather than a total order,
1954 since, for example, if two functions are called within one expression
1955 with no sequence point between them, the order in which the functions
1956 are called is not specified. However, the standards committee have
1957 ruled that function calls do not overlap.
1959 It is not specified when between sequence points modifications to the
1960 values of objects take effect. Programs whose behavior depends on this
1961 have undefined behavior; the C standard specifies that ``Between the
1962 previous and next sequence point an object shall have its stored value
1963 modified at most once by the evaluation of an expression. Furthermore,
1964 the prior value shall be read only to determine the value to be
1965 stored.''. If a program breaks these rules, the results on any
1966 particular implementation are entirely unpredictable.
1968 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1969 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1970 diagnosed by this option, and it may give an occasional false positive
1971 result, but in general it has been found fairly effective at detecting
1972 this sort of problem in programs.
1974 The present implementation of this option only works for C programs. A
1975 future implementation may also work for C++ programs.
1977 The C standard is worded confusingly, therefore there is some debate
1978 over the precise meaning of the sequence point rules in subtle cases.
1979 Links to discussions of the problem, including proposed formal
1980 definitions, may be found on our readings page, at
1981 @w{@uref{http://gcc.gnu.org/readings.html}}.
1984 @opindex Wreturn-type
1985 Warn whenever a function is defined with a return-type that defaults to
1986 @code{int}. Also warn about any @code{return} statement with no
1987 return-value in a function whose return-type is not @code{void}.
1989 For C++, a function without return type always produces a diagnostic
1990 message, even when @option{-Wno-return-type} is specified. The only
1991 exceptions are @samp{main} and functions defined in system headers.
1995 Warn whenever a @code{switch} statement has an index of enumeral type
1996 and lacks a @code{case} for one or more of the named codes of that
1997 enumeration. (The presence of a @code{default} label prevents this
1998 warning.) @code{case} labels outside the enumeration range also
1999 provoke warnings when this option is used.
2003 Warn if any trigraphs are encountered that might change the meaning of
2004 the program (trigraphs within comments are not warned about).
2006 @item -Wunused-function
2007 @opindex Wunused-function
2008 Warn whenever a static function is declared but not defined or a
2009 non\-inline static function is unused.
2011 @item -Wunused-label
2012 @opindex Wunused-label
2013 Warn whenever a label is declared but not used.
2015 To suppress this warning use the @samp{unused} attribute
2016 (@pxref{Variable Attributes}).
2018 @item -Wunused-parameter
2019 @opindex Wunused-parameter
2020 Warn whenever a function parameter is unused aside from its declaration.
2022 To suppress this warning use the @samp{unused} attribute
2023 (@pxref{Variable Attributes}).
2025 @item -Wunused-variable
2026 @opindex Wunused-variable
2027 Warn whenever a local variable or non-constant static variable is unused
2028 aside from its declaration
2030 To suppress this warning use the @samp{unused} attribute
2031 (@pxref{Variable Attributes}).
2033 @item -Wunused-value
2034 @opindex Wunused-value
2035 Warn whenever a statement computes a result that is explicitly not used.
2037 To suppress this warning cast the expression to @samp{void}.
2041 All all the above @option{-Wunused} options combined.
2043 In order to get a warning about an unused function parameter, you must
2044 either specify @samp{-W -Wunused} or separately specify
2045 @option{-Wunused-parameter}.
2047 @item -Wuninitialized
2048 @opindex Wuninitialized
2049 Warn if an automatic variable is used without first being initialized or
2050 if a variable may be clobbered by a @code{setjmp} call.
2052 These warnings are possible only in optimizing compilation,
2053 because they require data flow information that is computed only
2054 when optimizing. If you don't specify @option{-O}, you simply won't
2057 These warnings occur only for variables that are candidates for
2058 register allocation. Therefore, they do not occur for a variable that
2059 is declared @code{volatile}, or whose address is taken, or whose size
2060 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2061 structures, unions or arrays, even when they are in registers.
2063 Note that there may be no warning about a variable that is used only
2064 to compute a value that itself is never used, because such
2065 computations may be deleted by data flow analysis before the warnings
2068 These warnings are made optional because GCC is not smart
2069 enough to see all the reasons why the code might be correct
2070 despite appearing to have an error. Here is one example of how
2091 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2092 always initialized, but GCC doesn't know this. Here is
2093 another common case:
2098 if (change_y) save_y = y, y = new_y;
2100 if (change_y) y = save_y;
2105 This has no bug because @code{save_y} is used only if it is set.
2107 @cindex @code{longjmp} warnings
2108 This option also warns when a non-volatile automatic variable might be
2109 changed by a call to @code{longjmp}. These warnings as well are possible
2110 only in optimizing compilation.
2112 The compiler sees only the calls to @code{setjmp}. It cannot know
2113 where @code{longjmp} will be called; in fact, a signal handler could
2114 call it at any point in the code. As a result, you may get a warning
2115 even when there is in fact no problem because @code{longjmp} cannot
2116 in fact be called at the place which would cause a problem.
2118 Some spurious warnings can be avoided if you declare all the functions
2119 you use that never return as @code{noreturn}. @xref{Function
2122 @item -Wreorder @r{(C++ only)}
2124 @cindex reordering, warning
2125 @cindex warning for reordering of member initializers
2126 Warn when the order of member initializers given in the code does not
2127 match the order in which they must be executed. For instance:
2129 @item -Wunknown-pragmas
2130 @opindex Wunknown-pragmas
2131 @cindex warning for unknown pragmas
2132 @cindex unknown pragmas, warning
2133 @cindex pragmas, warning of unknown
2134 Warn when a #pragma directive is encountered which is not understood by
2135 GCC@. If this command line option is used, warnings will even be issued
2136 for unknown pragmas in system header files. This is not the case if
2137 the warnings were only enabled by the @option{-Wall} command line option.
2141 All of the above @samp{-W} options combined. This enables all the
2142 warnings about constructions that some users consider questionable, and
2143 that are easy to avoid (or modify to prevent the warning), even in
2144 conjunction with macros.
2147 @opindex Wno-div-by-zero
2148 @opindex Wdiv-by-zero
2149 Warn about compile-time integer division by zero. This is default. To
2150 inhibit the warning messages, use @option{-Wno-div-by-zero}. Floating
2151 point division by zero is not warned about, as it can be a legitimate
2152 way of obtaining infinities and NaNs.
2155 @opindex Wno-multichar
2157 Warn if a multicharacter constant (@samp{'FOOF'}) is used. This is
2158 default. To inhibit the warning messages, use @option{-Wno-multichar}.
2159 Usually they indicate a typo in the user's code, as they have
2160 implementation-defined values, and should not be used in portable code.
2162 @item -Wsystem-headers
2163 @opindex Wsystem-headers
2164 @cindex warnings from system headers
2165 @cindex system headers, warnings from
2166 Print warning messages for constructs found in system header files.
2167 Warnings from system headers are normally suppressed, on the assumption
2168 that they usually do not indicate real problems and would only make the
2169 compiler output harder to read. Using this command line option tells
2170 GCC to emit warnings from system headers as if they occurred in user
2171 code. However, note that using @option{-Wall} in conjunction with this
2172 option will @emph{not} warn about unknown pragmas in system
2173 headers---for that, @option{-Wunknown-pragmas} must also be used.
2176 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2177 Some of them warn about constructions that users generally do not
2178 consider questionable, but which occasionally you might wish to check
2179 for; others warn about constructions that are necessary or hard to avoid
2180 in some cases, and there is no simple way to modify the code to suppress
2186 Print extra warning messages for these events:
2190 A function can return either with or without a value. (Falling
2191 off the end of the function body is considered returning without
2192 a value.) For example, this function would evoke such a
2206 An expression-statement or the left-hand side of a comma expression
2207 contains no side effects.
2208 To suppress the warning, cast the unused expression to void.
2209 For example, an expression such as @samp{x[i,j]} will cause a warning,
2210 but @samp{x[(void)i,j]} will not.
2213 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2216 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2217 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2218 that of ordinary mathematical notation.
2221 Storage-class specifiers like @code{static} are not the first things in
2222 a declaration. According to the C Standard, this usage is obsolescent.
2225 The return type of a function has a type qualifier such as @code{const}.
2226 Such a type qualifier has no effect, since the value returned by a
2227 function is not an lvalue. (But don't warn about the GNU extension of
2228 @code{volatile void} return types. That extension will be warned about
2229 if @option{-pedantic} is specified.)
2232 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2236 A comparison between signed and unsigned values could produce an
2237 incorrect result when the signed value is converted to unsigned.
2238 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2241 An aggregate has a partly bracketed initializer.
2242 For example, the following code would evoke such a warning,
2243 because braces are missing around the initializer for @code{x.h}:
2246 struct s @{ int f, g; @};
2247 struct t @{ struct s h; int i; @};
2248 struct t x = @{ 1, 2, 3 @};
2252 An aggregate has an initializer which does not initialize all members.
2253 For example, the following code would cause such a warning, because
2254 @code{x.h} would be implicitly initialized to zero:
2257 struct s @{ int f, g, h; @};
2258 struct s x = @{ 3, 4 @};
2263 @opindex Wfloat-equal
2264 Warn if floating point values are used in equality comparisons.
2266 The idea behind this is that sometimes it is convenient (for the
2267 programmer) to consider floating-point values as approximations to
2268 infinitely precise real numbers. If you are doing this, then you need
2269 to compute (by analysing the code, or in some other way) the maximum or
2270 likely maximum error that the computation introduces, and allow for it
2271 when performing comparisons (and when producing output, but that's a
2272 different problem). In particular, instead of testing for equality, you
2273 would check to see whether the two values have ranges that overlap; and
2274 this is done with the relational operators, so equality comparisons are
2277 @item -Wtraditional @r{(C only)}
2278 @opindex Wtraditional
2279 Warn about certain constructs that behave differently in traditional and
2280 ISO C@. Also warn about ISO C constructs that have no traditional C
2281 equivalent, and/or problematic constructs which should be avoided.
2285 Macro parameters that appear within string literals in the macro body.
2286 In traditional C macro replacement takes place within string literals,
2287 but does not in ISO C@.
2290 In traditional C, some preprocessor directives did not exist.
2291 Traditional preprocessors would only consider a line to be a directive
2292 if the @samp{#} appeared in column 1 on the line. Therefore
2293 @option{-Wtraditional} warns about directives that traditional C
2294 understands but would ignore because the @samp{#} does not appear as the
2295 first character on the line. It also suggests you hide directives like
2296 @samp{#pragma} not understood by traditional C by indenting them. Some
2297 traditional implementations would not recognise @samp{#elif}, so it
2298 suggests avoiding it altogether.
2301 A function-like macro that appears without arguments.
2304 The unary plus operator.
2307 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2308 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2309 constants.) Note, these suffixes appear in macros defined in the system
2310 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2311 Use of these macros in user code might normally lead to spurious
2312 warnings, however gcc's integrated preprocessor has enough context to
2313 avoid warning in these cases.
2316 A function declared external in one block and then used after the end of
2320 A @code{switch} statement has an operand of type @code{long}.
2323 A non-@code{static} function declaration follows a @code{static} one.
2324 This construct is not accepted by some traditional C compilers.
2327 The ISO type of an integer constant has a different width or
2328 signedness from its traditional type. This warning is only issued if
2329 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2330 typically represent bit patterns, are not warned about.
2333 Usage of ISO string concatenation is detected.
2336 Initialization of automatic aggregates.
2339 Identifier conflicts with labels. Traditional C lacks a separate
2340 namespace for labels.
2343 Initialization of unions. If the initializer is zero, the warning is
2344 omitted. This is done under the assumption that the zero initializer in
2345 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2346 initializer warnings and relies on default initialization to zero in the
2350 Conversions by prototypes between fixed/floating point values and vice
2351 versa. The absence of these prototypes when compiling with traditional
2352 C would cause serious problems. This is a subset of the possible
2353 conversion warnings, for the full set use @option{-Wconversion}.
2358 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2362 Warn whenever a local variable shadows another local variable, parameter or
2363 global variable or whenever a built-in function is shadowed.
2365 @item -Wlarger-than-@var{len}
2366 @opindex Wlarger-than
2367 Warn whenever an object of larger than @var{len} bytes is defined.
2369 @item -Wpointer-arith
2370 @opindex Wpointer-arith
2371 Warn about anything that depends on the ``size of'' a function type or
2372 of @code{void}. GNU C assigns these types a size of 1, for
2373 convenience in calculations with @code{void *} pointers and pointers
2376 @item -Wbad-function-cast @r{(C only)}
2377 @opindex Wbad-function-cast
2378 Warn whenever a function call is cast to a non-matching type.
2379 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2383 Warn whenever a pointer is cast so as to remove a type qualifier from
2384 the target type. For example, warn if a @code{const char *} is cast
2385 to an ordinary @code{char *}.
2388 @opindex Wcast-align
2389 Warn whenever a pointer is cast such that the required alignment of the
2390 target is increased. For example, warn if a @code{char *} is cast to
2391 an @code{int *} on machines where integers can only be accessed at
2392 two- or four-byte boundaries.
2394 @item -Wwrite-strings
2395 @opindex Wwrite-strings
2396 When compiling C, give string constants the type @code{const
2397 char[@var{length}]} so that
2398 copying the address of one into a non-@code{const} @code{char *}
2399 pointer will get a warning; when compiling C++, warn about the
2400 deprecated conversion from string constants to @code{char *}.
2401 These warnings will help you find at
2402 compile time code that can try to write into a string constant, but
2403 only if you have been very careful about using @code{const} in
2404 declarations and prototypes. Otherwise, it will just be a nuisance;
2405 this is why we did not make @option{-Wall} request these warnings.
2408 @opindex Wconversion
2409 Warn if a prototype causes a type conversion that is different from what
2410 would happen to the same argument in the absence of a prototype. This
2411 includes conversions of fixed point to floating and vice versa, and
2412 conversions changing the width or signedness of a fixed point argument
2413 except when the same as the default promotion.
2415 Also, warn if a negative integer constant expression is implicitly
2416 converted to an unsigned type. For example, warn about the assignment
2417 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2418 casts like @code{(unsigned) -1}.
2420 @item -Wsign-compare
2421 @opindex Wsign-compare
2422 @cindex warning for comparison of signed and unsigned values
2423 @cindex comparison of signed and unsigned values, warning
2424 @cindex signed and unsigned values, comparison warning
2425 Warn when a comparison between signed and unsigned values could produce
2426 an incorrect result when the signed value is converted to unsigned.
2427 This warning is also enabled by @option{-W}; to get the other warnings
2428 of @option{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2430 @item -Waggregate-return
2431 @opindex Waggregate-return
2432 Warn if any functions that return structures or unions are defined or
2433 called. (In languages where you can return an array, this also elicits
2436 @item -Wstrict-prototypes @r{(C only)}
2437 @opindex Wstrict-prototypes
2438 Warn if a function is declared or defined without specifying the
2439 argument types. (An old-style function definition is permitted without
2440 a warning if preceded by a declaration which specifies the argument
2443 @item -Wmissing-prototypes @r{(C only)}
2444 @opindex Wmissing-prototypes
2445 Warn if a global function is defined without a previous prototype
2446 declaration. This warning is issued even if the definition itself
2447 provides a prototype. The aim is to detect global functions that fail
2448 to be declared in header files.
2450 @item -Wmissing-declarations
2451 @opindex Wmissing-declarations
2452 Warn if a global function is defined without a previous declaration.
2453 Do so even if the definition itself provides a prototype.
2454 Use this option to detect global functions that are not declared in
2457 @item -Wmissing-noreturn
2458 @opindex Wmissing-noreturn
2459 Warn about functions which might be candidates for attribute @code{noreturn}.
2460 Note these are only possible candidates, not absolute ones. Care should
2461 be taken to manually verify functions actually do not ever return before
2462 adding the @code{noreturn} attribute, otherwise subtle code generation
2463 bugs could be introduced. You will not get a warning for @code{main} in
2464 hosted C environments.
2466 @item -Wmissing-format-attribute
2467 @opindex Wmissing-format-attribute
2469 If @option{-Wformat} is enabled, also warn about functions which might be
2470 candidates for @code{format} attributes. Note these are only possible
2471 candidates, not absolute ones. GCC will guess that @code{format}
2472 attributes might be appropriate for any function that calls a function
2473 like @code{vprintf} or @code{vscanf}, but this might not always be the
2474 case, and some functions for which @code{format} attributes are
2475 appropriate may not be detected. This option has no effect unless
2476 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2480 Warn if a structure is given the packed attribute, but the packed
2481 attribute has no effect on the layout or size of the structure.
2482 Such structures may be mis-aligned for little benefit. For
2483 instance, in this code, the variable @code{f.x} in @code{struct bar}
2484 will be misaligned even though @code{struct bar} does not itself
2485 have the packed attribute:
2492 @} __attribute__((packed));
2502 Warn if padding is included in a structure, either to align an element
2503 of the structure or to align the whole structure. Sometimes when this
2504 happens it is possible to rearrange the fields of the structure to
2505 reduce the padding and so make the structure smaller.
2507 @item -Wredundant-decls
2508 @opindex Wredundant-decls
2509 Warn if anything is declared more than once in the same scope, even in
2510 cases where multiple declaration is valid and changes nothing.
2512 @item -Wnested-externs @r{(C only)}
2513 @opindex Wnested-externs
2514 Warn if an @code{extern} declaration is encountered within a function.
2516 @item -Wunreachable-code
2517 @opindex Wunreachable-code
2518 Warn if the compiler detects that code will never be executed.
2520 This option is intended to warn when the compiler detects that at
2521 least a whole line of source code will never be executed, because
2522 some condition is never satisfied or because it is after a
2523 procedure that never returns.
2525 It is possible for this option to produce a warning even though there
2526 are circumstances under which part of the affected line can be executed,
2527 so care should be taken when removing apparently-unreachable code.
2529 For instance, when a function is inlined, a warning may mean that the
2530 line is unreachable in only one inlined copy of the function.
2532 This option is not made part of @option{-Wall} because in a debugging
2533 version of a program there is often substantial code which checks
2534 correct functioning of the program and is, hopefully, unreachable
2535 because the program does work. Another common use of unreachable
2536 code is to provide behaviour which is selectable at compile-time.
2540 Warn if a function can not be inlined and it was declared as inline.
2544 @opindex Wno-long-long
2545 Warn if @samp{long long} type is used. This is default. To inhibit
2546 the warning messages, use @option{-Wno-long-long}. Flags
2547 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2548 only when @option{-pedantic} flag is used.
2550 @item -Wdisabled-optimization
2551 @opindex Wdisabled-optimization
2552 Warn if a requested optimization pass is disabled. This warning does
2553 not generally indicate that there is anything wrong with your code; it
2554 merely indicates that GCC's optimizers were unable to handle the code
2555 effectively. Often, the problem is that your code is too big or too
2556 complex; GCC will refuse to optimize programs when the optimization
2557 itself is likely to take inordinate amounts of time.
2561 Make all warnings into errors.
2564 @node Debugging Options
2565 @section Options for Debugging Your Program or GCC
2566 @cindex options, debugging
2567 @cindex debugging information options
2569 GCC has various special options that are used for debugging
2570 either your program or GCC:
2575 Produce debugging information in the operating system's native format
2576 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2579 On most systems that use stabs format, @option{-g} enables use of extra
2580 debugging information that only GDB can use; this extra information
2581 makes debugging work better in GDB but will probably make other debuggers
2583 refuse to read the program. If you want to control for certain whether
2584 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2585 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, or @option{-gdwarf-1}
2588 Unlike most other C compilers, GCC allows you to use @option{-g} with
2589 @option{-O}. The shortcuts taken by optimized code may occasionally
2590 produce surprising results: some variables you declared may not exist
2591 at all; flow of control may briefly move where you did not expect it;
2592 some statements may not be executed because they compute constant
2593 results or their values were already at hand; some statements may
2594 execute in different places because they were moved out of loops.
2596 Nevertheless it proves possible to debug optimized output. This makes
2597 it reasonable to use the optimizer for programs that might have bugs.
2599 The following options are useful when GCC is generated with the
2600 capability for more than one debugging format.
2604 Produce debugging information for use by GDB@. This means to use the
2605 most expressive format available (DWARF 2, stabs, or the native format
2606 if neither of those are supported), including GDB extensions if at all
2611 Produce debugging information in stabs format (if that is supported),
2612 without GDB extensions. This is the format used by DBX on most BSD
2613 systems. On MIPS, Alpha and System V Release 4 systems this option
2614 produces stabs debugging output which is not understood by DBX or SDB@.
2615 On System V Release 4 systems this option requires the GNU assembler.
2619 Produce debugging information in stabs format (if that is supported),
2620 using GNU extensions understood only by the GNU debugger (GDB)@. The
2621 use of these extensions is likely to make other debuggers crash or
2622 refuse to read the program.
2626 Produce debugging information in COFF format (if that is supported).
2627 This is the format used by SDB on most System V systems prior to
2632 Produce debugging information in XCOFF format (if that is supported).
2633 This is the format used by the DBX debugger on IBM RS/6000 systems.
2637 Produce debugging information in XCOFF format (if that is supported),
2638 using GNU extensions understood only by the GNU debugger (GDB)@. The
2639 use of these extensions is likely to make other debuggers crash or
2640 refuse to read the program, and may cause assemblers other than the GNU
2641 assembler (GAS) to fail with an error.
2645 Produce debugging information in DWARF version 1 format (if that is
2646 supported). This is the format used by SDB on most System V Release 4
2651 Produce debugging information in DWARF version 1 format (if that is
2652 supported), using GNU extensions understood only by the GNU debugger
2653 (GDB)@. The use of these extensions is likely to make other debuggers
2654 crash or refuse to read the program.
2658 Produce debugging information in DWARF version 2 format (if that is
2659 supported). This is the format used by DBX on IRIX 6.
2662 @itemx -ggdb@var{level}
2663 @itemx -gstabs@var{level}
2664 @itemx -gcoff@var{level}
2665 @itemx -gxcoff@var{level}
2666 @itemx -gdwarf@var{level}
2667 @itemx -gdwarf-2@var{level}
2668 Request debugging information and also use @var{level} to specify how
2669 much information. The default level is 2.
2671 Level 1 produces minimal information, enough for making backtraces in
2672 parts of the program that you don't plan to debug. This includes
2673 descriptions of functions and external variables, but no information
2674 about local variables and no line numbers.
2676 Level 3 includes extra information, such as all the macro definitions
2677 present in the program. Some debuggers support macro expansion when
2678 you use @option{-g3}.
2683 Generate extra code to write profile information suitable for the
2684 analysis program @code{prof}. You must use this option when compiling
2685 the source files you want data about, and you must also use it when
2688 @cindex @code{gprof}
2691 Generate extra code to write profile information suitable for the
2692 analysis program @code{gprof}. You must use this option when compiling
2693 the source files you want data about, and you must also use it when
2699 Generate extra code to write profile information for basic blocks, which will
2700 record the number of times each basic block is executed, the basic block start
2701 address, and the function name containing the basic block. If @option{-g} is
2702 used, the line number and filename of the start of the basic block will also be
2703 recorded. If not overridden by the machine description, the default action is
2704 to append to the text file @file{bb.out}.
2706 This data could be analyzed by a program like @code{tcov}. Note,
2707 however, that the format of the data is not what @code{tcov} expects.
2708 Eventually GNU @code{gprof} should be extended to process this data.
2712 Makes the compiler print out each function name as it is compiled, and
2713 print some statistics about each pass when it finishes.
2716 @opindex ftime-report
2717 Makes the compiler print some statistics about the time consumed by each
2718 pass when it finishes.
2721 @opindex fmem-report
2722 Makes the compiler print some statistics about permanent memory
2723 allocation when it finishes.
2727 Generate extra code to profile basic blocks. Your executable will
2728 produce output that is a superset of that produced when @option{-a} is
2729 used. Additional output is the source and target address of the basic
2730 blocks where a jump takes place, the number of times a jump is executed,
2731 and (optionally) the complete sequence of basic blocks being executed.
2732 The output is appended to file @file{bb.out}.
2734 You can examine different profiling aspects without recompilation. Your
2735 executable will read a list of function names from file @file{bb.in}.
2736 Profiling starts when a function on the list is entered and stops when
2737 that invocation is exited. To exclude a function from profiling, prefix
2738 its name with @samp{-}. If a function name is not unique, you can
2739 disambiguate it by writing it in the form
2740 @samp{/path/filename.d:functionname}. Your executable will write the
2741 available paths and filenames in file @file{bb.out}.
2743 Several function names have a special meaning:
2746 Write source, target and frequency of jumps to file @file{bb.out}.
2747 @item __bb_hidecall__
2748 Exclude function calls from frequency count.
2749 @item __bb_showret__
2750 Include function returns in frequency count.
2752 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2753 The file will be compressed using the program @samp{gzip}, which must
2754 exist in your @env{PATH}. On systems without the @samp{popen}
2755 function, the file will be named @file{bbtrace} and will not be
2756 compressed. @strong{Profiling for even a few seconds on these systems
2757 will produce a very large file.} Note: @code{__bb_hidecall__} and
2758 @code{__bb_showret__} will not affect the sequence written to
2762 Here's a short example using different profiling parameters
2763 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2764 1 and 2 and is called twice from block 3 of function @code{main}. After
2765 the calls, block 3 transfers control to block 4 of @code{main}.
2767 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2768 the following sequence of blocks is written to file @file{bbtrace.gz}:
2769 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2770 the return is to a point inside the block and not to the top. The
2771 block address 0 always indicates, that control is transferred
2772 to the trace from somewhere outside the observed functions. With
2773 @samp{-foo} added to @file{bb.in}, the blocks of function
2774 @code{foo} are removed from the trace, so only 0 3 4 remains.
2776 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2777 jump frequencies will be written to file @file{bb.out}. The
2778 frequencies are obtained by constructing a trace of blocks
2779 and incrementing a counter for every neighbouring pair of blocks
2780 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2784 Jump from block 0x0 to block 0x3 executed 1 time(s)
2785 Jump from block 0x3 to block 0x1 executed 1 time(s)
2786 Jump from block 0x1 to block 0x2 executed 2 time(s)
2787 Jump from block 0x2 to block 0x1 executed 1 time(s)
2788 Jump from block 0x2 to block 0x4 executed 1 time(s)
2791 With @code{__bb_hidecall__}, control transfer due to call instructions
2792 is removed from the trace, that is the trace is cut into three parts: 0
2793 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2794 to return instructions is added to the trace. The trace becomes: 0 3 1
2795 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2796 written to @file{bbtrace.gz}. It is solely used for counting jump
2799 @item -fprofile-arcs
2800 @opindex fprofile-arcs
2801 Instrument @dfn{arcs} during compilation to generate coverage data
2802 or for profile-directed block ordering. During execution the program
2803 records how many times each branch is executed and how many times it is
2804 taken. When the compiled program exits it saves this data to a file
2805 called @file{@var{sourcename}.da} for each source file.
2807 For profile-directed block ordering, compile the program with
2808 @option{-fprofile-arcs} plus optimization and code generation options,
2809 generate the arc profile information by running the program on a
2810 selected workload, and then compile the program again with the same
2811 optimization and code generation options plus
2812 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2813 Control Optimization}).
2815 The other use of @option{-fprofile-arcs} is for use with @code{gcov},
2816 when it is used with the @option{-ftest-coverage} option. GCC
2817 supports two methods of determining code coverage: the options that
2818 support @code{gcov}, and options @option{-a} and @option{-ax}, which
2819 write information to text files. The options that support @code{gcov}
2820 do not need to instrument every arc in the program, so a program compiled
2821 with them runs faster than a program compiled with @option{-a}, which
2822 adds instrumentation code to every basic block in the program. The
2823 tradeoff: since @code{gcov} does not have execution counts for all
2824 branches, it must start with the execution counts for the instrumented
2825 branches, and then iterate over the program flow graph until the entire
2826 graph has been solved. Hence, @code{gcov} runs a little more slowly than
2827 a program which uses information from @option{-a} and @option{-ax}.
2829 With @option{-fprofile-arcs}, for each function of your program GCC
2830 creates a program flow graph, then finds a spanning tree for the graph.
2831 Only arcs that are not on the spanning tree have to be instrumented: the
2832 compiler adds code to count the number of times that these arcs are
2833 executed. When an arc is the only exit or only entrance to a block, the
2834 instrumentation code can be added to the block; otherwise, a new basic
2835 block must be created to hold the instrumentation code.
2837 This option makes it possible to estimate branch probabilities and to
2838 calculate basic block execution counts. In general, basic block
2839 execution counts as provided by @option{-a} do not give enough
2840 information to estimate all branch probabilities.
2843 @item -ftest-coverage
2844 @opindex ftest-coverage
2845 Create data files for the @code{gcov} code-coverage utility
2846 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2847 The data file names begin with the name of your source file:
2850 @item @var{sourcename}.bb
2851 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2852 associate basic block execution counts with line numbers.
2854 @item @var{sourcename}.bbg
2855 A list of all arcs in the program flow graph. This allows @code{gcov}
2856 to reconstruct the program flow graph, so that it can compute all basic
2857 block and arc execution counts from the information in the
2858 @code{@var{sourcename}.da} file.
2861 Use @option{-ftest-coverage} with @option{-fprofile-arcs}; the latter
2862 option adds instrumentation to the program, which then writes
2863 execution counts to another data file:
2866 @item @var{sourcename}.da
2867 Runtime arc execution counts, used in conjunction with the arc
2868 information in the file @code{@var{sourcename}.bbg}.
2871 Coverage data will map better to the source files if
2872 @option{-ftest-coverage} is used without optimization.
2874 @item -d@var{letters}
2876 Says to make debugging dumps during compilation at times specified by
2877 @var{letters}. This is used for debugging the compiler. The file names
2878 for most of the dumps are made by appending a pass number and a word to
2879 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2880 Here are the possible letters for use in @var{letters}, and their meanings:
2885 Annotate the assembler output with miscellaneous debugging information.
2888 Dump after computing branch probabilities, to @file{@var{file}.14.bp}.
2891 Dump after block reordering, to @file{@var{file}.28.bbro}.
2894 Dump after instruction combination, to the file @file{@var{file}.16.combine}.
2897 Dump after the first if conversion, to the file @file{@var{file}.17.ce}.
2900 Dump after delayed branch scheduling, to @file{@var{file}.31.dbr}.
2903 Dump all macro definitions, at the end of preprocessing, in addition to
2907 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
2908 @file{@var{file}.07.ussa}.
2911 Dump after the second if conversion, to @file{@var{file}.26.ce2}.
2914 Dump after life analysis, to @file{@var{file}.15.life}.
2917 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.09.addressof}.
2920 Dump after global register allocation, to @file{@var{file}.21.greg}.
2923 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
2926 Dump after post-reload optimizations, to @file{@var{file}.22.postreload}.
2929 Dump after GCSE, to @file{@var{file}.10.gcse}.
2932 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2935 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
2938 Dump after conversion from registers to stack, to @file{@var{file}.32.stack}.
2941 Dump after local register allocation, to @file{@var{file}.20.lreg}.
2944 Dump after loop optimization, to @file{@var{file}.11.loop}.
2947 Dump after performing the machine dependent reorganisation pass, to
2948 @file{@var{file}.30.mach}.
2951 Dump after register renumbering, to @file{@var{file}.25.rnreg}.
2954 Dump after the register move pass, to @file{@var{file}.18.regmove}.
2957 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2960 Dump after the second scheduling pass, to @file{@var{file}.27.sched2}.
2963 Dump after CSE (including the jump optimization that sometimes follows
2964 CSE), to @file{@var{file}.08.cse}.
2967 Dump after the first scheduling pass, to @file{@var{file}.19.sched}.
2970 Dump after the second CSE pass (including the jump optimization that
2971 sometimes follows CSE), to @file{@var{file}.12.cse2}.
2974 Dump after the second flow pass, to @file{@var{file}.23.flow2}.
2977 Dump after SSA dead code elimination, to @file{@var{file}.06.ssadce}.
2980 Dump after the peephole pass, to @file{@var{file}.24.peephole2}.
2983 Produce all the dumps listed above.
2986 Print statistics on memory usage, at the end of the run, to
2990 Annotate the assembler output with a comment indicating which
2991 pattern and alternative was used. The length of each instruction is
2995 Dump the RTL in the assembler output as a comment before each instruction.
2996 Also turns on @option{-dp} annotation.
2999 For each of the other indicated dump files (except for
3000 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
3001 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
3004 Just generate RTL for a function instead of compiling it. Usually used
3008 Dump debugging information during parsing, to standard error.
3011 @item -fdump-unnumbered
3012 @opindex fdump-unnumbered
3013 When doing debugging dumps (see @option{-d} option above), suppress instruction
3014 numbers and line number note output. This makes it more feasible to
3015 use diff on debugging dumps for compiler invocations with different
3016 options, in particular with and without @option{-g}.
3018 @item -fdump-class-hierarchy @r{(C++ only)}
3019 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
3020 @opindex fdump-class-hierarchy
3021 Dump a representation of each class's hierarchy and virtual function
3022 table layout to a file. The file name is made by appending @file{.class}
3023 to the source file name. If the @samp{-@var{options}} form is used,
3024 @var{options} controls the details of the dump as described for the
3025 @option{-fdump-tree} options.
3027 @item -fdump-tree-@var{switch} @r{(C++ only)}
3028 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3030 Control the dumping at various stages of processing the intermediate
3031 language tree to a file. The file name is generated by appending a switch
3032 specific suffix to the source file name. If the @samp{-@var{options}}
3033 form is used, @var{options} is a list of @samp{-} separated options that
3034 control the details of the dump. Not all options are applicable to all
3035 dumps, those which are not meaningful will be ignored. The following
3036 options are available
3040 Print the address of each node. Usually this is not meaningful as it
3041 changes according to the environment and source file. Its primary use
3042 is for tying up a dump file with a debug environment.
3044 Inhibit dumping of members of a scope or body of a function merely
3045 because that scope has been reached. Only dump such items when they
3046 are directly reachable by some other path.
3048 Turn on all options.
3051 The following tree dumps are possible:
3054 Dump before any tree based optimization, to @file{@var{file}.original}.
3056 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3058 Dump after function inlining, to @file{@var{file}.inlined}.
3061 @item -fpretend-float
3062 @opindex fpretend-float
3063 When running a cross-compiler, pretend that the target machine uses the
3064 same floating point format as the host machine. This causes incorrect
3065 output of the actual floating constants, but the actual instruction
3066 sequence will probably be the same as GCC would make when running on
3071 Store the usual ``temporary'' intermediate files permanently; place them
3072 in the current directory and name them based on the source file. Thus,
3073 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3074 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3075 preprocessed @file{foo.i} output file even though the compiler now
3076 normally uses an integrated preprocessor.
3080 Report the CPU time taken by each subprocess in the compilation
3081 sequence. For C source files, this is the compiler proper and assembler
3082 (plus the linker if linking is done). The output looks like this:
3089 The first number on each line is the ``user time,'' that is time spent
3090 executing the program itself. The second number is ``system time,''
3091 time spent executing operating system routines on behalf of the program.
3092 Both numbers are in seconds.
3094 @item -print-file-name=@var{library}
3095 @opindex print-file-name
3096 Print the full absolute name of the library file @var{library} that
3097 would be used when linking---and don't do anything else. With this
3098 option, GCC does not compile or link anything; it just prints the
3101 @item -print-multi-directory
3102 @opindex print-multi-directory
3103 Print the directory name corresponding to the multilib selected by any
3104 other switches present in the command line. This directory is supposed
3105 to exist in @env{GCC_EXEC_PREFIX}.
3107 @item -print-multi-lib
3108 @opindex print-multi-lib
3109 Print the mapping from multilib directory names to compiler switches
3110 that enable them. The directory name is separated from the switches by
3111 @samp{;}, and each switch starts with an @samp{@@} instead of the
3112 @samp{-}, without spaces between multiple switches. This is supposed to
3113 ease shell-processing.
3115 @item -print-prog-name=@var{program}
3116 @opindex print-prog-name
3117 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3119 @item -print-libgcc-file-name
3120 @opindex print-libgcc-file-name
3121 Same as @option{-print-file-name=libgcc.a}.
3123 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3124 but you do want to link with @file{libgcc.a}. You can do
3127 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3130 @item -print-search-dirs
3131 @opindex print-search-dirs
3132 Print the name of the configured installation directory and a list of
3133 program and library directories gcc will search---and don't do anything else.
3135 This is useful when gcc prints the error message
3136 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3137 To resolve this you either need to put @file{cpp0} and the other compiler
3138 components where gcc expects to find them, or you can set the environment
3139 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3140 Don't forget the trailing '/'.
3141 @xref{Environment Variables}.
3144 @opindex dumpmachine
3145 Print the compiler's target machine (for example,
3146 @samp{i686-pc-linux-gnu})---and don't do anything else.
3149 @opindex dumpversion
3150 Print the compiler version (for example, @samp{3.0})---and don't do
3155 Print the compiler's built-in specs---and don't do anything else. (This
3156 is used when GCC itself is being built.) @xref{Spec Files}.
3159 @node Optimize Options
3160 @section Options That Control Optimization
3161 @cindex optimize options
3162 @cindex options, optimization
3164 These options control various sorts of optimizations:
3171 Optimize. Optimizing compilation takes somewhat more time, and a lot
3172 more memory for a large function.
3174 Without @option{-O}, the compiler's goal is to reduce the cost of
3175 compilation and to make debugging produce the expected results.
3176 Statements are independent: if you stop the program with a breakpoint
3177 between statements, you can then assign a new value to any variable or
3178 change the program counter to any other statement in the function and
3179 get exactly the results you would expect from the source code.
3181 With @option{-O}, the compiler tries to reduce code size and execution
3182 time, without performing any optimizations that take a great deal of
3187 Optimize even more. GCC performs nearly all supported optimizations
3188 that do not involve a space-speed tradeoff. The compiler does not
3189 perform loop unrolling or function inlining when you specify @option{-O2}.
3190 As compared to @option{-O}, this option increases both compilation time
3191 and the performance of the generated code.
3193 @option{-O2} turns on all optional optimizations except for loop unrolling,
3194 function inlining, and register renaming. It also turns on the
3195 @option{-fforce-mem} option on all machines and frame pointer elimination
3196 on machines where doing so does not interfere with debugging.
3198 Please note the warning under @option{-fgcse} about
3199 invoking @option{-O2} on programs that use computed gotos.
3203 Optimize yet more. @option{-O3} turns on all optimizations specified by
3204 @option{-O2} and also turns on the @option{-finline-functions} and
3205 @option{-frename-registers} options.
3213 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3214 do not typically increase code size. It also performs further
3215 optimizations designed to reduce code size.
3217 If you use multiple @option{-O} options, with or without level numbers,
3218 the last such option is the one that is effective.
3221 Options of the form @option{-f@var{flag}} specify machine-independent
3222 flags. Most flags have both positive and negative forms; the negative
3223 form of @option{-ffoo} would be @option{-fno-foo}. In the table below,
3224 only one of the forms is listed---the one which is not the default.
3225 You can figure out the other form by either removing @samp{no-} or
3230 @opindex ffloat-store
3231 Do not store floating point variables in registers, and inhibit other
3232 options that might change whether a floating point value is taken from a
3235 @cindex floating point precision
3236 This option prevents undesirable excess precision on machines such as
3237 the 68000 where the floating registers (of the 68881) keep more
3238 precision than a @code{double} is supposed to have. Similarly for the
3239 x86 architecture. For most programs, the excess precision does only
3240 good, but a few programs rely on the precise definition of IEEE floating
3241 point. Use @option{-ffloat-store} for such programs, after modifying
3242 them to store all pertinent intermediate computations into variables.
3244 @item -fno-default-inline
3245 @opindex fno-default-inline
3246 Do not make member functions inline by default merely because they are
3247 defined inside the class scope (C++ only). Otherwise, when you specify
3248 @w{@option{-O}}, member functions defined inside class scope are compiled
3249 inline by default; i.e., you don't need to add @samp{inline} in front of
3250 the member function name.
3252 @item -fno-defer-pop
3253 @opindex fno-defer-pop
3254 Always pop the arguments to each function call as soon as that function
3255 returns. For machines which must pop arguments after a function call,
3256 the compiler normally lets arguments accumulate on the stack for several
3257 function calls and pops them all at once.
3261 Force memory operands to be copied into registers before doing
3262 arithmetic on them. This produces better code by making all memory
3263 references potential common subexpressions. When they are not common
3264 subexpressions, instruction combination should eliminate the separate
3265 register-load. The @option{-O2} option turns on this option.
3268 @opindex fforce-addr
3269 Force memory address constants to be copied into registers before
3270 doing arithmetic on them. This may produce better code just as
3271 @option{-fforce-mem} may.
3273 @item -fomit-frame-pointer
3274 @opindex fomit-frame-pointer
3275 Don't keep the frame pointer in a register for functions that
3276 don't need one. This avoids the instructions to save, set up and
3277 restore frame pointers; it also makes an extra register available
3278 in many functions. @strong{It also makes debugging impossible on
3282 On some machines, such as the VAX, this flag has no effect, because
3283 the standard calling sequence automatically handles the frame pointer
3284 and nothing is saved by pretending it doesn't exist. The
3285 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3286 whether a target machine supports this flag. @xref{Registers}.
3289 On some machines, such as the VAX, this flag has no effect, because
3290 the standard calling sequence automatically handles the frame pointer
3291 and nothing is saved by pretending it doesn't exist. The
3292 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3293 whether a target machine supports this flag. @xref{Registers,,Register
3294 Usage, gcc.info, Using and Porting GCC}.
3297 @item -foptimize-sibling-calls
3298 @opindex foptimize-sibling-calls
3299 Optimize sibling and tail recursive calls.
3303 This option generates traps for signed overflow on addition, subtraction,
3304 multiplication operations.
3308 Don't pay attention to the @code{inline} keyword. Normally this option
3309 is used to keep the compiler from expanding any functions inline.
3310 Note that if you are not optimizing, no functions can be expanded inline.
3312 @item -finline-functions
3313 @opindex finline-functions
3314 Integrate all simple functions into their callers. The compiler
3315 heuristically decides which functions are simple enough to be worth
3316 integrating in this way.
3318 If all calls to a given function are integrated, and the function is
3319 declared @code{static}, then the function is normally not output as
3320 assembler code in its own right.
3322 @item -finline-limit=@var{n}
3323 @opindex finline-limit
3324 By default, gcc limits the size of functions that can be inlined. This flag
3325 allows the control of this limit for functions that are explicitly marked as
3326 inline (ie marked with the inline keyword or defined within the class
3327 definition in c++). @var{n} is the size of functions that can be inlined in
3328 number of pseudo instructions (not counting parameter handling). The default
3329 value of @var{n} is 600.
3330 Increasing this value can result in more inlined code at
3331 the cost of compilation time and memory consumption. Decreasing usually makes
3332 the compilation faster and less code will be inlined (which presumably
3333 means slower programs). This option is particularly useful for programs that
3334 use inlining heavily such as those based on recursive templates with C++.
3336 @emph{Note:} pseudo instruction represents, in this particular context, an
3337 abstract measurement of function's size. In no way, it represents a count
3338 of assembly instructions and as such its exact meaning might change from one
3339 release to an another.
3341 @item -fkeep-inline-functions
3342 @opindex fkeep-inline-functions
3343 Even if all calls to a given function are integrated, and the function
3344 is declared @code{static}, nevertheless output a separate run-time
3345 callable version of the function. This switch does not affect
3346 @code{extern inline} functions.
3348 @item -fkeep-static-consts
3349 @opindex fkeep-static-consts
3350 Emit variables declared @code{static const} when optimization isn't turned
3351 on, even if the variables aren't referenced.
3353 GCC enables this option by default. If you want to force the compiler to
3354 check if the variable was referenced, regardless of whether or not
3355 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3357 @item -fmerge-constants
3358 Attempt to merge identical constants (string constants and floating point
3359 constants) accross compilation units.
3361 This option is default for optimized compilation if assembler and linker
3362 support it. Use @option{-fno-merge-constants} to inhibit this behaviour.
3364 @item -fmerge-all-constants
3365 Attempt to merge identical constants and identical variables.
3367 This option implies @option{-fmerge-constants}. In addition to
3368 @option{-fmerge-constants} this considers e.g. even constant initialized
3369 arrays or initialized constant variables with integral or floating point
3370 types. Languages like C or C++ require each non-automatic variable to
3371 have distinct location, so using this option will result in non-conforming
3374 @item -fno-function-cse
3375 @opindex fno-function-cse
3376 Do not put function addresses in registers; make each instruction that
3377 calls a constant function contain the function's address explicitly.
3379 This option results in less efficient code, but some strange hacks
3380 that alter the assembler output may be confused by the optimizations
3381 performed when this option is not used.
3385 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, and @*
3386 @option{-fno-trapping-math}.
3388 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
3390 This option should never be turned on by any @option{-O} option since
3391 it can result in incorrect output for programs which depend on
3392 an exact implementation of IEEE or ISO rules/specifications for
3395 @item -fno-math-errno
3396 @opindex fno-math-errno
3397 Do not set ERRNO after calling math functions that are executed
3398 with a single instruction, e.g., sqrt. A program that relies on
3399 IEEE exceptions for math error handling may want to use this flag
3400 for speed while maintaining IEEE arithmetic compatibility.
3402 This option should never be turned on by any @option{-O} option since
3403 it can result in incorrect output for programs which depend on
3404 an exact implementation of IEEE or ISO rules/specifications for
3407 The default is @option{-fmath-errno}.
3409 @item -funsafe-math-optimizations
3410 @opindex funsafe-math-optimizations
3411 Allow optimizations for floating-point arithmetic that (a) assume
3412 that arguments and results are valid and (b) may violate IEEE or
3413 ANSI standards. When used at link-time, it may include libraries
3414 or startup files that change the default FPU control word or other
3415 similar optimizations.
3417 This option should never be turned on by any @option{-O} option since
3418 it can result in incorrect output for programs which depend on
3419 an exact implementation of IEEE or ISO rules/specifications for
3422 The default is @option{-fno-unsafe-math-optimizations}.
3424 @item -fno-trapping-math
3425 @opindex fno-trapping-math
3426 Compile code assuming that floating-point operations cannot generate
3427 user-visible traps. Setting this option may allow faster code
3428 if one relies on ``non-stop'' IEEE arithmetic, for example.
3430 This option should never be turned on by any @option{-O} option since
3431 it can result in incorrect output for programs which depend on
3432 an exact implementation of IEEE or ISO rules/specifications for
3435 The default is @option{-ftrapping-math}.
3438 The following options control specific optimizations. The @option{-O2}
3439 option turns on all of these optimizations except @option{-funroll-loops}
3440 and @option{-funroll-all-loops}. On most machines, the @option{-O} option
3441 turns on the @option{-fthread-jumps} and @option{-fdelayed-branch} options,
3442 but specific machines may handle it differently.
3444 You can use the following flags in the rare cases when ``fine-tuning''
3445 of optimizations to be performed is desired.
3447 Not all of the optimizations performed by GCC have @option{-f} options
3451 @item -fstrength-reduce
3452 @opindex fstrength-reduce
3453 Perform the optimizations of loop strength reduction and
3454 elimination of iteration variables.
3456 @item -fthread-jumps
3457 @opindex fthread-jumps
3458 Perform optimizations where we check to see if a jump branches to a
3459 location where another comparison subsumed by the first is found. If
3460 so, the first branch is redirected to either the destination of the
3461 second branch or a point immediately following it, depending on whether
3462 the condition is known to be true or false.
3464 @item -fcse-follow-jumps
3465 @opindex fcse-follow-jumps
3466 In common subexpression elimination, scan through jump instructions
3467 when the target of the jump is not reached by any other path. For
3468 example, when CSE encounters an @code{if} statement with an
3469 @code{else} clause, CSE will follow the jump when the condition
3472 @item -fcse-skip-blocks
3473 @opindex fcse-skip-blocks
3474 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3475 follow jumps which conditionally skip over blocks. When CSE
3476 encounters a simple @code{if} statement with no else clause,
3477 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3478 body of the @code{if}.
3480 @item -frerun-cse-after-loop
3481 @opindex frerun-cse-after-loop
3482 Re-run common subexpression elimination after loop optimizations has been
3485 @item -frerun-loop-opt
3486 @opindex frerun-loop-opt
3487 Run the loop optimizer twice.
3491 Perform a global common subexpression elimination pass.
3492 This pass also performs global constant and copy propagation.
3494 @emph{Note:} When compiling a program using computed gotos, a GCC
3495 extension, you may get better runtime performance if you disable
3496 the global common subexpression elmination pass by adding
3497 @option{-fno-gcse} to the command line.
3501 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3502 attempt to move loads which are only killed by stores into themselves. This
3503 allows a loop containing a load/store sequence to be changed to a load outside
3504 the loop, and a copy/store within the loop.
3508 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3509 subexpression elimination. This pass will attempt to move stores out of loops.
3510 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3511 can be changed to a load before the loop and a store after the loop.
3513 @item -fdelete-null-pointer-checks
3514 @opindex fdelete-null-pointer-checks
3515 Use global dataflow analysis to identify and eliminate useless checks
3516 for null pointers. The compiler assumes that dereferencing a null
3517 pointer would have halted the program. If a pointer is checked after
3518 it has already been dereferenced, it cannot be null.
3520 In some environments, this assumption is not true, and programs can
3521 safely dereference null pointers. Use
3522 @option{-fno-delete-null-pointer-checks} to disable this optimization
3523 for programs which depend on that behavior.
3525 @item -fexpensive-optimizations
3526 @opindex fexpensive-optimizations
3527 Perform a number of minor optimizations that are relatively expensive.
3529 @item -foptimize-register-move
3531 @opindex foptimize-register-move
3533 Attempt to reassign register numbers in move instructions and as
3534 operands of other simple instructions in order to maximize the amount of
3535 register tying. This is especially helpful on machines with two-operand
3536 instructions. GCC enables this optimization by default with @option{-O2}
3539 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3542 @item -fdelayed-branch
3543 @opindex fdelayed-branch
3544 If supported for the target machine, attempt to reorder instructions
3545 to exploit instruction slots available after delayed branch
3548 @item -fschedule-insns
3549 @opindex fschedule-insns
3550 If supported for the target machine, attempt to reorder instructions to
3551 eliminate execution stalls due to required data being unavailable. This
3552 helps machines that have slow floating point or memory load instructions
3553 by allowing other instructions to be issued until the result of the load
3554 or floating point instruction is required.
3556 @item -fschedule-insns2
3557 @opindex fschedule-insns2
3558 Similar to @option{-fschedule-insns}, but requests an additional pass of
3559 instruction scheduling after register allocation has been done. This is
3560 especially useful on machines with a relatively small number of
3561 registers and where memory load instructions take more than one cycle.
3563 @item -ffunction-sections
3564 @itemx -fdata-sections
3565 @opindex ffunction-sections
3566 @opindex fdata-sections
3567 Place each function or data item into its own section in the output
3568 file if the target supports arbitrary sections. The name of the
3569 function or the name of the data item determines the section's name
3572 Use these options on systems where the linker can perform optimizations
3573 to improve locality of reference in the instruction space. HPPA
3574 processors running HP-UX and Sparc processors running Solaris 2 have
3575 linkers with such optimizations. Other systems using the ELF object format
3576 as well as AIX may have these optimizations in the future.
3578 Only use these options when there are significant benefits from doing
3579 so. When you specify these options, the assembler and linker will
3580 create larger object and executable files and will also be slower.
3581 You will not be able to use @code{gprof} on all systems if you
3582 specify this option and you may have problems with debugging if
3583 you specify both this option and @option{-g}.
3585 @item -fcaller-saves
3586 @opindex fcaller-saves
3587 Enable values to be allocated in registers that will be clobbered by
3588 function calls, by emitting extra instructions to save and restore the
3589 registers around such calls. Such allocation is done only when it
3590 seems to result in better code than would otherwise be produced.
3592 This option is always enabled by default on certain machines, usually
3593 those which have no call-preserved registers to use instead.
3595 For all machines, optimization level 2 and higher enables this flag by
3598 @item -funroll-loops
3599 @opindex funroll-loops
3600 Unroll loops whose number of iterations can be determined at compile
3601 time or upon entry to the loop. @option{-funroll-loops} implies both
3602 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
3603 option makes code larger, and may or may not make it run faster.
3605 @item -funroll-all-loops
3606 @opindex funroll-all-loops
3607 Unroll all loops, even if their number of iterations is uncertain when
3608 the loop is entered. This usually makes programs run more slowly.
3609 @option{-funroll-all-loops} implies the same options as
3610 @option{-funroll-loops},
3613 @item -fmove-all-movables
3614 @opindex fmove-all-movables
3615 Forces all invariant computations in loops to be moved
3618 @item -freduce-all-givs
3619 @opindex freduce-all-givs
3620 Forces all general-induction variables in loops to be
3623 @emph{Note:} When compiling programs written in Fortran,
3624 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
3625 by default when you use the optimizer.
3627 These options may generate better or worse code; results are highly
3628 dependent on the structure of loops within the source code.
3630 These two options are intended to be removed someday, once
3631 they have helped determine the efficacy of various
3632 approaches to improving loop optimizations.
3634 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3635 know how use of these options affects
3636 the performance of your production code.
3637 We're very interested in code that runs @emph{slower}
3638 when these options are @emph{enabled}.
3641 @itemx -fno-peephole2
3642 @opindex fno-peephole
3643 @opindex fno-peephole2
3644 Disable any machine-specific peephole optimizations. The difference
3645 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
3646 are implemented in the compiler; some targets use one, some use the
3647 other, a few use both.
3649 @item -fbranch-probabilities
3650 @opindex fbranch-probabilities
3651 After running a program compiled with @option{-fprofile-arcs}
3652 (@pxref{Debugging Options,, Options for Debugging Your Program or
3653 @command{gcc}}), you can compile it a second time using
3654 @option{-fbranch-probabilities}, to improve optimizations based on
3655 the number of times each branch was taken. When the program
3656 compiled with @option{-fprofile-arcs} exits it saves arc execution
3657 counts to a file called @file{@var{sourcename}.da} for each source
3658 file The information in this data file is very dependent on the
3659 structure of the generated code, so you must use the same source code
3660 and the same optimization options for both compilations.
3663 With @option{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3664 note on the first instruction of each basic block, and a
3665 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3666 These can be used to improve optimization. Currently, they are only
3667 used in one place: in @file{reorg.c}, instead of guessing which path a
3668 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3669 exactly determine which path is taken more often.
3672 @item -fno-guess-branch-probability
3673 @opindex fno-guess-branch-probability
3674 Do not guess branch probabilities using a randomized model.
3676 Sometimes gcc will opt to use a randomized model to guess branch
3677 probabilities, when none are available from either profiling feedback
3678 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
3679 different runs of the compiler on the same program may produce different
3682 In a hard real-time system, people don't want different runs of the
3683 compiler to produce code that has different behavior; minimizing
3684 non-determinism is of paramount import. This switch allows users to
3685 reduce non-determinism, possibly at the expense of inferior
3688 @item -fstrict-aliasing
3689 @opindex fstrict-aliasing
3690 Allows the compiler to assume the strictest aliasing rules applicable to
3691 the language being compiled. For C (and C++), this activates
3692 optimizations based on the type of expressions. In particular, an
3693 object of one type is assumed never to reside at the same address as an
3694 object of a different type, unless the types are almost the same. For
3695 example, an @code{unsigned int} can alias an @code{int}, but not a
3696 @code{void*} or a @code{double}. A character type may alias any other
3699 Pay special attention to code like this:
3712 The practice of reading from a different union member than the one most
3713 recently written to (called ``type-punning'') is common. Even with
3714 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
3715 is accessed through the union type. So, the code above will work as
3716 expected. However, this code might not:
3728 Every language that wishes to perform language-specific alias analysis
3729 should define a function that computes, given an @code{tree}
3730 node, an alias set for the node. Nodes in different alias sets are not
3731 allowed to alias. For an example, see the C front-end function
3732 @code{c_get_alias_set}.
3735 @item -falign-functions
3736 @itemx -falign-functions=@var{n}
3737 @opindex falign-functions
3738 Align the start of functions to the next power-of-two greater than
3739 @var{n}, skipping up to @var{n} bytes. For instance,
3740 @option{-falign-functions=32} aligns functions to the next 32-byte
3741 boundary, but @option{-falign-functions=24} would align to the next
3742 32-byte boundary only if this can be done by skipping 23 bytes or less.
3744 @option{-fno-align-functions} and @option{-falign-functions=1} are
3745 equivalent and mean that functions will not be aligned.
3747 Some assemblers only support this flag when @var{n} is a power of two;
3748 in that case, it is rounded up.
3750 If @var{n} is not specified, use a machine-dependent default.
3752 @item -falign-labels
3753 @itemx -falign-labels=@var{n}
3754 @opindex falign-labels
3755 Align all branch targets to a power-of-two boundary, skipping up to
3756 @var{n} bytes like @option{-falign-functions}. This option can easily
3757 make code slower, because it must insert dummy operations for when the
3758 branch target is reached in the usual flow of the code.
3760 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
3761 are greater than this value, then their values are used instead.
3763 If @var{n} is not specified, use a machine-dependent default which is
3764 very likely to be @samp{1}, meaning no alignment.
3767 @itemx -falign-loops=@var{n}
3768 @opindex falign-loops
3769 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3770 like @option{-falign-functions}. The hope is that the loop will be
3771 executed many times, which will make up for any execution of the dummy
3774 If @var{n} is not specified, use a machine-dependent default.
3777 @itemx -falign-jumps=@var{n}
3778 @opindex falign-jumps
3779 Align branch targets to a power-of-two boundary, for branch targets
3780 where the targets can only be reached by jumping, skipping up to @var{n}
3781 bytes like @option{-falign-functions}. In this case, no dummy operations
3784 If @var{n} is not specified, use a machine-dependent default.
3788 Perform optimizations in static single assignment form. Each function's
3789 flow graph is translated into SSA form, optimizations are performed, and
3790 the flow graph is translated back from SSA form. Users should not
3791 specify this option, since it is not yet ready for production use.
3795 Perform Sparse Conditional Constant Propagation in SSA form. Requires
3796 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
3800 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
3801 Like @option{-fssa}, this is an experimental feature.
3803 @item -fsingle-precision-constant
3804 @opindex fsingle-precision-constant
3805 Treat floating point constant as single precision constant instead of
3806 implicitly converting it to double precision constant.
3808 @item -frename-registers
3809 @opindex frename-registers
3810 Attempt to avoid false dependencies in scheduled code by making use
3811 of registers left over after register allocation. This optimization
3812 will most benefit processors with lots of registers. It can, however,
3813 make debugging impossible, since variables will no longer stay in
3814 a ``home register''.
3816 @item --param @var{name}=@var{value}
3818 In some places, GCC uses various constants to control the amount of
3819 optimization that is done. For example, GCC will not inline functions
3820 that contain more that a certain number of instructions. You can
3821 control some of these constants on the command-line using the
3822 @option{--param} option.
3824 In each case, the @var{value} is an integer. The allowable choices for
3825 @var{name} are given in the following table:
3828 @item max-delay-slot-insn-search
3829 The maximum number of instructions to consider when looking for an
3830 instruction to fill a delay slot. If more than this arbitrary number of
3831 instructions is searched, the time savings from filling the delay slot
3832 will be minimal so stop searching. Increasing values mean more
3833 aggressive optimization, making the compile time increase with probably
3834 small improvement in executable run time.
3836 @item max-delay-slot-live-search
3837 When trying to fill delay slots, the maximum number of instructions to
3838 consider when searching for a block with valid live register
3839 information. Increasing this arbitrarily chosen value means more
3840 aggressive optimization, increasing the compile time. This parameter
3841 should be removed when the delay slot code is rewritten to maintain the
3844 @item max-gcse-memory
3845 The approximate maximum amount of memory that will be allocated in
3846 order to perform the global common subexpression elimination
3847 optimization. If more memory than specified is required, the
3848 optimization will not be done.
3850 @item max-gcse-passes
3851 The maximum number of passes of GCSE to run.
3853 @item max-pending-list-length
3854 The maximum number of pending dependencies scheduling will allow
3855 before flushing the current state and starting over. Large functions
3856 with few branches or calls can create excessively large lists which
3857 needlessly consume memory and resources.
3859 @item max-inline-insns
3860 If an function contains more than this many instructions, it
3861 will not be inlined. This option is precisely equivalent to
3862 @option{-finline-limit}.
3867 @node Preprocessor Options
3868 @section Options Controlling the Preprocessor
3869 @cindex preprocessor options
3870 @cindex options, preprocessor
3872 These options control the C preprocessor, which is run on each C source
3873 file before actual compilation.
3875 If you use the @option{-E} option, nothing is done except preprocessing.
3876 Some of these options make sense only together with @option{-E} because
3877 they cause the preprocessor output to be unsuitable for actual
3881 @item -include @var{file}
3883 Process @var{file} as input before processing the regular input file.
3884 In effect, the contents of @var{file} are compiled first. Any @option{-D}
3885 and @option{-U} options on the command line are always processed before
3886 @option{-include @var{file}}, regardless of the order in which they are
3887 written. All the @option{-include} and @option{-imacros} options are
3888 processed in the order in which they are written.
3890 @item -imacros @var{file}
3892 Process @var{file} as input, discarding the resulting output, before
3893 processing the regular input file. Because the output generated from
3894 @var{file} is discarded, the only effect of @option{-imacros @var{file}}
3895 is to make the macros defined in @var{file} available for use in the
3896 main input. All the @option{-include} and @option{-imacros} options are
3897 processed in the order in which they are written.
3899 @item -idirafter @var{dir}
3901 @cindex second include path
3902 Add the directory @var{dir} to the second include path. The directories
3903 on the second include path are searched when a header file is not found
3904 in any of the directories in the main include path (the one that
3905 @option{-I} adds to).
3907 @item -iprefix @var{prefix}
3909 Specify @var{prefix} as the prefix for subsequent @option{-iwithprefix}
3912 @item -iwithprefix @var{dir}
3913 @opindex iwithprefix
3914 Add a directory to the second include path. The directory's name is
3915 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3916 specified previously with @option{-iprefix}. If you have not specified a
3917 prefix yet, the directory containing the installed passes of the
3918 compiler is used as the default.
3920 @item -iwithprefixbefore @var{dir}
3921 @opindex iwithprefixbefore
3922 Add a directory to the main include path. The directory's name is made
3923 by concatenating @var{prefix} and @var{dir}, as in the case of
3924 @option{-iwithprefix}.
3926 @item -isystem @var{dir}
3928 Add a directory to the beginning of the second include path, marking it
3929 as a system directory, so that it gets the same special treatment as
3930 is applied to the standard system directories.
3934 Do not search the standard system directories for header files. Only
3935 the directories you have specified with @option{-I} options (and the
3936 current directory, if appropriate) are searched. @xref{Directory
3937 Options}, for information on @option{-I}.
3939 By using both @option{-nostdinc} and @option{-I-}, you can limit the include-file
3940 search path to only those directories you specify explicitly.
3944 When searching for a header file in a directory, remap file names if a
3945 file named @file{header.gcc} exists in that directory. This can be used
3946 to work around limitations of file systems with file name restrictions.
3947 The @file{header.gcc} file should contain a series of lines with two
3948 tokens on each line: the first token is the name to map, and the second
3949 token is the actual name to use.
3953 Do not predefine any nonstandard macros. (Including architecture flags).
3957 Run only the C preprocessor. Preprocess all the C source files
3958 specified and output the results to standard output or to the
3959 specified output file.
3963 Tell the preprocessor not to discard comments. Used with the
3968 Tell the preprocessor not to generate @samp{#line} directives.
3969 Used with the @option{-E} option.
3972 @cindex dependencies, make
3975 Instead of outputting the result of preprocessing, output a rule
3976 suitable for @code{make} describing the dependencies of the main source
3977 file. The preprocessor outputs one @code{make} rule containing the
3978 object file name for that source file, a colon, and the names of all the
3979 included files. Unless overridden explicitly, the object file name
3980 consists of the basename of the source file with any suffix replaced with
3981 object file suffix. If there are many included files then the
3982 rule is split into several lines using @samp{\}-newline.
3984 @option{-M} implies @option{-E}.
3988 Like @option{-M}, but mention only the files included with @samp{#include
3989 "@var{file}"}. System header files included with @samp{#include
3990 <@var{file}>} are omitted.
3994 Like @option{-M} but the dependency information is written to a file
3995 rather than stdout. @code{gcc} will use the same file name and
3996 directory as the object file, but with the suffix @file{.d} instead.
3998 This is in addition to compiling the main file as specified---@option{-MD}
3999 does not inhibit ordinary compilation the way @option{-M} does,
4000 unless you also specify @option{-MG}.
4002 With Mach, you can use the utility @code{md} to merge multiple
4003 dependency files into a single dependency file suitable for using with
4004 the @samp{make} command.
4008 Like @option{-MD} except mention only user header files, not system
4011 @item -MF @var{file}
4013 When used with @option{-M} or @option{-MM}, specifies a file to write the
4014 dependencies to. This allows the preprocessor to write the preprocessed
4015 file to stdout normally. If no @option{-MF} switch is given, CPP sends
4016 the rules to stdout and suppresses normal preprocessed output.
4018 Another way to specify output of a @code{make} rule is by setting
4019 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
4024 When used with @option{-M} or @option{-MM}, @option{-MG} says to treat missing
4025 header files as generated files and assume they live in the same
4026 directory as the source file. It suppresses preprocessed output, as a
4027 missing header file is ordinarily an error.
4029 This feature is used in automatic updating of makefiles.
4033 This option instructs CPP to add a phony target for each dependency
4034 other than the main file, causing each to depend on nothing. These
4035 dummy rules work around errors @code{make} gives if you remove header
4036 files without updating the @code{Makefile} to match.
4038 This is typical output:-
4041 /tmp/test.o: /tmp/test.c /tmp/test.h
4046 @item -MQ @var{target}
4047 @item -MT @var{target}
4050 By default CPP uses the main file name, including any path, and appends
4051 the object suffix, normally ``.o'', to it to obtain the name of the
4052 target for dependency generation. With @option{-MT} you can specify a
4053 target yourself, overriding the default one.
4055 If you want multiple targets, you can specify them as a single argument
4056 to @option{-MT}, or use multiple @option{-MT} options.
4058 The targets you specify are output in the order they appear on the
4059 command line. @option{-MQ} is identical to @option{-MT}, except that the
4060 target name is quoted for Make, but with @option{-MT} it isn't. For
4061 example, @option{-MT '$(objpfx)foo.o'} gives
4064 $(objpfx)foo.o: /tmp/foo.c
4067 but @option{-MQ '$(objpfx)foo.o'} gives
4070 $$(objpfx)foo.o: /tmp/foo.c
4073 The default target is automatically quoted, as if it were given with
4078 Print the name of each header file used, in addition to other normal
4081 @item -A@var{question}(@var{answer})
4083 Assert the answer @var{answer} for @var{question}, in case it is tested
4084 with a preprocessing conditional such as @samp{#if
4085 #@var{question}(@var{answer})}. @option{-A-} disables the standard
4086 assertions that normally describe the target machine.
4090 Define macro @var{macro} with the string @samp{1} as its definition.
4092 @item -D@var{macro}=@var{defn}
4093 Define macro @var{macro} as @var{defn}. All instances of @option{-D} on
4094 the command line are processed before any @option{-U} options.
4096 Any @option{-D} and @option{-U} options on the command line are processed in
4097 order, and always before @option{-imacros @var{file}}, regardless of the
4098 order in which they are written.
4102 Undefine macro @var{macro}. @option{-U} options are evaluated after all
4103 @option{-D} options, but before any @option{-include} and @option{-imacros}
4106 Any @option{-D} and @option{-U} options on the command line are processed in
4107 order, and always before @option{-imacros @var{file}}, regardless of the
4108 order in which they are written.
4112 Tell the preprocessor to output only a list of the macro definitions
4113 that are in effect at the end of preprocessing. Used with the @option{-E}
4118 Tell the preprocessing to pass all macro definitions into the output, in
4119 their proper sequence in the rest of the output.
4123 Like @option{-dD} except that the macro arguments and contents are omitted.
4124 Only @samp{#define @var{name}} is included in the output.
4128 Output @samp{#include} directives in addition to the result of
4131 @item -fpreprocessed
4132 @opindex fpreprocessed
4133 Indicate to the preprocessor that the input file has already been
4134 preprocessed. This suppresses things like macro expansion, trigraph
4135 conversion, escaped newline splicing, and processing of most directives.
4136 The preprocessor still recognizes and removes comments, so that you can
4137 pass a file preprocessed with @option{-C} to the compiler without
4138 problems. In this mode the integrated preprocessor is little more than
4139 a tokenizer for the front ends.
4141 @option{-fpreprocessed} is implicit if the input file has one of the
4142 extensions @samp{i}, @samp{ii} or @samp{mi}. These are the extensions
4143 that GCC uses for preprocessed files created by @option{-save-temps}.
4147 Process ISO standard trigraph sequences. These are three-character
4148 sequences, all starting with @samp{??}, that are defined by ISO C to
4149 stand for single characters. For example, @samp{??/} stands for
4150 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
4151 default, GCC ignores trigraphs, but in standard-conforming modes it
4152 converts them. See the @option{-std} and @option{-ansi} options.
4154 The nine trigraph sequences are
4157 @expansion{} @samp{[}
4160 @expansion{} @samp{]}
4163 @expansion{} @samp{@{}
4166 @expansion{} @samp{@}}
4169 @expansion{} @samp{#}
4172 @expansion{} @samp{\}
4175 @expansion{} @samp{^}
4178 @expansion{} @samp{|}
4181 @expansion{} @samp{~}
4185 Trigraph support is not popular, so many compilers do not implement it
4186 properly. Portable code should not rely on trigraphs being either
4187 converted or ignored.
4189 @item -Wp,@var{option}
4191 Pass @var{option} as an option to the preprocessor. If @var{option}
4192 contains commas, it is split into multiple options at the commas.
4195 @node Assembler Options
4196 @section Passing Options to the Assembler
4198 @c prevent bad page break with this line
4199 You can pass options to the assembler.
4202 @item -Wa,@var{option}
4204 Pass @var{option} as an option to the assembler. If @var{option}
4205 contains commas, it is split into multiple options at the commas.
4209 @section Options for Linking
4210 @cindex link options
4211 @cindex options, linking
4213 These options come into play when the compiler links object files into
4214 an executable output file. They are meaningless if the compiler is
4215 not doing a link step.
4219 @item @var{object-file-name}
4220 A file name that does not end in a special recognized suffix is
4221 considered to name an object file or library. (Object files are
4222 distinguished from libraries by the linker according to the file
4223 contents.) If linking is done, these object files are used as input
4232 If any of these options is used, then the linker is not run, and
4233 object file names should not be used as arguments. @xref{Overall
4237 @item -l@var{library}
4238 @itemx -l @var{library}
4240 Search the library named @var{library} when linking. (The second
4241 alternative with the library as a separate argument is only for
4242 POSIX compliance and is not recommended.)
4244 It makes a difference where in the command you write this option; the
4245 linker searches and processes libraries and object files in the order they
4246 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4247 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4248 to functions in @samp{z}, those functions may not be loaded.
4250 The linker searches a standard list of directories for the library,
4251 which is actually a file named @file{lib@var{library}.a}. The linker
4252 then uses this file as if it had been specified precisely by name.
4254 The directories searched include several standard system directories
4255 plus any that you specify with @option{-L}.
4257 Normally the files found this way are library files---archive files
4258 whose members are object files. The linker handles an archive file by
4259 scanning through it for members which define symbols that have so far
4260 been referenced but not defined. But if the file that is found is an
4261 ordinary object file, it is linked in the usual fashion. The only
4262 difference between using an @option{-l} option and specifying a file name
4263 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4264 and searches several directories.
4268 You need this special case of the @option{-l} option in order to
4269 link an Objective-C program.
4272 @opindex nostartfiles
4273 Do not use the standard system startup files when linking.
4274 The standard system libraries are used normally, unless @option{-nostdlib}
4275 or @option{-nodefaultlibs} is used.
4277 @item -nodefaultlibs
4278 @opindex nodefaultlibs
4279 Do not use the standard system libraries when linking.
4280 Only the libraries you specify will be passed to the linker.
4281 The standard startup files are used normally, unless @option{-nostartfiles}
4282 is used. The compiler may generate calls to memcmp, memset, and memcpy
4283 for System V (and ISO C) environments or to bcopy and bzero for
4284 BSD environments. These entries are usually resolved by entries in
4285 libc. These entry points should be supplied through some other
4286 mechanism when this option is specified.
4290 Do not use the standard system startup files or libraries when linking.
4291 No startup files and only the libraries you specify will be passed to
4292 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4293 for System V (and ISO C) environments or to bcopy and bzero for
4294 BSD environments. These entries are usually resolved by entries in
4295 libc. These entry points should be supplied through some other
4296 mechanism when this option is specified.
4298 @cindex @option{-lgcc}, use with @option{-nostdlib}
4299 @cindex @option{-nostdlib} and unresolved references
4300 @cindex unresolved references and @option{-nostdlib}
4301 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4302 @cindex @option{-nodefaultlibs} and unresolved references
4303 @cindex unresolved references and @option{-nodefaultlibs}
4304 One of the standard libraries bypassed by @option{-nostdlib} and
4305 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4306 that GCC uses to overcome shortcomings of particular machines, or special
4307 needs for some languages.
4309 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
4313 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
4314 for more discussion of @file{libgcc.a}.)
4316 In most cases, you need @file{libgcc.a} even when you want to avoid
4317 other standard libraries. In other words, when you specify @option{-nostdlib}
4318 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4319 This ensures that you have no unresolved references to internal GCC
4320 library subroutines. (For example, @samp{__main}, used to ensure C++
4321 constructors will be called; @pxref{Collect2,,@command{collect2}}.)
4325 Remove all symbol table and relocation information from the executable.
4329 On systems that support dynamic linking, this prevents linking with the shared
4330 libraries. On other systems, this option has no effect.
4334 Produce a shared object which can then be linked with other objects to
4335 form an executable. Not all systems support this option. For predictable
4336 results, you must also specify the same set of options that were used to
4337 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4338 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4339 needs to build supplementary stub code for constructors to work. On
4340 multi-libbed systems, @samp{gcc -shared} must select the correct support
4341 libraries to link against. Failing to supply the correct flags may lead
4342 to subtle defects. Supplying them in cases where they are not necessary
4345 @item -shared-libgcc
4346 @itemx -static-libgcc
4347 @opindex shared-libgcc
4348 @opindex static-libgcc
4349 On systems that provide @file{libgcc} as a shared library, these options
4350 force the use of either the shared or static version respectively.
4351 If no shared version of @file{libgcc} was built when the compiler was
4352 configured, these options have no effect.
4354 There are several situations in which an application should use the
4355 shared @file{libgcc} instead of the static version. The most common
4356 of these is when the application wishes to throw and catch exceptions
4357 across different shared libraries. In that case, each of the libraries
4358 as well as the application itself should use the shared @file{libgcc}.
4360 Therefore, whenever you specify the @option{-shared} option, the GCC
4361 driver automatically adds @option{-shared-libgcc}, unless you explicitly
4362 specify @option{-static-libgcc}. The G++ driver automatically adds
4363 @option{-shared-libgcc} when you build a main executable as well because
4364 for C++ programs that is typically the right thing to do.
4365 (Exception-handling will not work reliably otherwise.)
4367 However, when linking a main executable written in C, you must
4368 explicitly say @option{-shared-libgcc} if you want to use the shared
4373 Bind references to global symbols when building a shared object. Warn
4374 about any unresolved references (unless overridden by the link editor
4375 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4378 @item -Xlinker @var{option}
4380 Pass @var{option} as an option to the linker. You can use this to
4381 supply system-specific linker options which GCC does not know how to
4384 If you want to pass an option that takes an argument, you must use
4385 @option{-Xlinker} twice, once for the option and once for the argument.
4386 For example, to pass @option{-assert definitions}, you must write
4387 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4388 @option{-Xlinker "-assert definitions"}, because this passes the entire
4389 string as a single argument, which is not what the linker expects.
4391 @item -Wl,@var{option}
4393 Pass @var{option} as an option to the linker. If @var{option} contains
4394 commas, it is split into multiple options at the commas.
4396 @item -u @var{symbol}
4398 Pretend the symbol @var{symbol} is undefined, to force linking of
4399 library modules to define it. You can use @option{-u} multiple times with
4400 different symbols to force loading of additional library modules.
4403 @node Directory Options
4404 @section Options for Directory Search
4405 @cindex directory options
4406 @cindex options, directory search
4409 These options specify directories to search for header files, for
4410 libraries and for parts of the compiler:
4415 Add the directory @var{dir} to the head of the list of directories to be
4416 searched for header files. This can be used to override a system header
4417 file, substituting your own version, since these directories are
4418 searched before the system header file directories. However, you should
4419 not use this option to add directories that contain vendor-supplied
4420 system header files (use @option{-isystem} for that). If you use more than
4421 one @option{-I} option, the directories are scanned in left-to-right
4422 order; the standard system directories come after.
4424 If a standard system include directory, or a directory specified with
4425 @option{-isystem}, is also specified with @option{-I}, it will be
4426 searched only in the position requested by @option{-I}. Also, it will
4427 not be considered a system include directory. If that directory really
4428 does contain system headers, there is a good chance that they will
4429 break. For instance, if GCC's installation procedure edited the headers
4430 in @file{/usr/include} to fix bugs, @samp{-I/usr/include} will cause the
4431 original, buggy headers to be found instead of the corrected ones. GCC
4432 will issue a warning when a system include directory is hidden in this
4437 Any directories you specify with @option{-I} options before the @option{-I-}
4438 option are searched only for the case of @samp{#include "@var{file}"};
4439 they are not searched for @samp{#include <@var{file}>}.
4441 If additional directories are specified with @option{-I} options after
4442 the @option{-I-}, these directories are searched for all @samp{#include}
4443 directives. (Ordinarily @emph{all} @option{-I} directories are used
4446 In addition, the @option{-I-} option inhibits the use of the current
4447 directory (where the current input file came from) as the first search
4448 directory for @samp{#include "@var{file}"}. There is no way to
4449 override this effect of @option{-I-}. With @option{-I.} you can specify
4450 searching the directory which was current when the compiler was
4451 invoked. That is not exactly the same as what the preprocessor does
4452 by default, but it is often satisfactory.
4454 @option{-I-} does not inhibit the use of the standard system directories
4455 for header files. Thus, @option{-I-} and @option{-nostdinc} are
4460 Add directory @var{dir} to the list of directories to be searched
4463 @item -B@var{prefix}
4465 This option specifies where to find the executables, libraries,
4466 include files, and data files of the compiler itself.
4468 The compiler driver program runs one or more of the subprograms
4469 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4470 @var{prefix} as a prefix for each program it tries to run, both with and
4471 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4473 For each subprogram to be run, the compiler driver first tries the
4474 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
4475 was not specified, the driver tries two standard prefixes, which are
4476 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4477 those results in a file name that is found, the unmodified program
4478 name is searched for using the directories specified in your
4479 @env{PATH} environment variable.
4481 The compiler will check to see if the path provided by the @option{-B}
4482 refers to a directory, and if necessary it will add a directory
4483 separator character at the end of the path.
4485 @option{-B} prefixes that effectively specify directory names also apply
4486 to libraries in the linker, because the compiler translates these
4487 options into @option{-L} options for the linker. They also apply to
4488 includes files in the preprocessor, because the compiler translates these
4489 options into @option{-isystem} options for the preprocessor. In this case,
4490 the compiler appends @samp{include} to the prefix.
4492 The run-time support file @file{libgcc.a} can also be searched for using
4493 the @option{-B} prefix, if needed. If it is not found there, the two
4494 standard prefixes above are tried, and that is all. The file is left
4495 out of the link if it is not found by those means.
4497 Another way to specify a prefix much like the @option{-B} prefix is to use
4498 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4501 As a special kludge, if the path provided by @option{-B} is
4502 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
4503 9, then it will be replaced by @file{[dir/]include}. This is to help
4504 with boot-strapping the compiler.
4506 @item -specs=@var{file}
4508 Process @var{file} after the compiler reads in the standard @file{specs}
4509 file, in order to override the defaults that the @file{gcc} driver
4510 program uses when determining what switches to pass to @file{cc1},
4511 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
4512 @option{-specs=@var{file}} can be specified on the command line, and they
4513 are processed in order, from left to right.
4519 @section Specifying subprocesses and the switches to pass to them
4521 @command{gcc} is a driver program. It performs its job by invoking a
4522 sequence of other programs to do the work of compiling, assembling and
4523 linking. GCC interprets its command-line parameters and uses these to
4524 deduce which programs it should invoke, and which command-line options
4525 it ought to place on their command lines. This behaviour is controlled
4526 by @dfn{spec strings}. In most cases there is one spec string for each
4527 program that GCC can invoke, but a few programs have multiple spec
4528 strings to control their behaviour. The spec strings built into GCC can
4529 be overridden by using the @option{-specs=} command-line switch to specify
4532 @dfn{Spec files} are plaintext files that are used to construct spec
4533 strings. They consist of a sequence of directives separated by blank
4534 lines. The type of directive is determined by the first non-whitespace
4535 character on the line and it can be one of the following:
4538 @item %@var{command}
4539 Issues a @var{command} to the spec file processor. The commands that can
4543 @item %include <@var{file}>
4545 Search for @var{file} and insert its text at the current point in the
4548 @item %include_noerr <@var{file}>
4549 @cindex %include_noerr
4550 Just like @samp{%include}, but do not generate an error message if the include
4551 file cannot be found.
4553 @item %rename @var{old_name} @var{new_name}
4555 Rename the spec string @var{old_name} to @var{new_name}.
4559 @item *[@var{spec_name}]:
4560 This tells the compiler to create, override or delete the named spec
4561 string. All lines after this directive up to the next directive or
4562 blank line are considered to be the text for the spec string. If this
4563 results in an empty string then the spec will be deleted. (Or, if the
4564 spec did not exist, then nothing will happened.) Otherwise, if the spec
4565 does not currently exist a new spec will be created. If the spec does
4566 exist then its contents will be overridden by the text of this
4567 directive, unless the first character of that text is the @samp{+}
4568 character, in which case the text will be appended to the spec.
4570 @item [@var{suffix}]:
4571 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4572 and up to the next directive or blank line are considered to make up the
4573 spec string for the indicated suffix. When the compiler encounters an
4574 input file with the named suffix, it will processes the spec string in
4575 order to work out how to compile that file. For example:
4582 This says that any input file whose name ends in @samp{.ZZ} should be
4583 passed to the program @samp{z-compile}, which should be invoked with the
4584 command-line switch @option{-input} and with the result of performing the
4585 @samp{%i} substitution. (See below.)
4587 As an alternative to providing a spec string, the text that follows a
4588 suffix directive can be one of the following:
4591 @item @@@var{language}
4592 This says that the suffix is an alias for a known @var{language}. This is
4593 similar to using the @option{-x} command-line switch to GCC to specify a
4594 language explicitly. For example:
4601 Says that .ZZ files are, in fact, C++ source files.
4604 This causes an error messages saying:
4607 @var{name} compiler not installed on this system.
4611 GCC already has an extensive list of suffixes built into it.
4612 This directive will add an entry to the end of the list of suffixes, but
4613 since the list is searched from the end backwards, it is effectively
4614 possible to override earlier entries using this technique.
4618 GCC has the following spec strings built into it. Spec files can
4619 override these strings or create their own. Note that individual
4620 targets can also add their own spec strings to this list.
4623 asm Options to pass to the assembler
4624 asm_final Options to pass to the assembler post-processor
4625 cpp Options to pass to the C preprocessor
4626 cc1 Options to pass to the C compiler
4627 cc1plus Options to pass to the C++ compiler
4628 endfile Object files to include at the end of the link
4629 link Options to pass to the linker
4630 lib Libraries to include on the command line to the linker
4631 libgcc Decides which GCC support library to pass to the linker
4632 linker Sets the name of the linker
4633 predefines Defines to be passed to the C preprocessor
4634 signed_char Defines to pass to CPP to say whether @code{char} is signed
4636 startfile Object files to include at the start of the link
4639 Here is a small example of a spec file:
4645 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4648 This example renames the spec called @samp{lib} to @samp{old_lib} and
4649 then overrides the previous definition of @samp{lib} with a new one.
4650 The new definition adds in some extra command-line options before
4651 including the text of the old definition.
4653 @dfn{Spec strings} are a list of command-line options to be passed to their
4654 corresponding program. In addition, the spec strings can contain
4655 @samp{%}-prefixed sequences to substitute variable text or to
4656 conditionally insert text into the command line. Using these constructs
4657 it is possible to generate quite complex command lines.
4659 Here is a table of all defined @samp{%}-sequences for spec
4660 strings. Note that spaces are not generated automatically around the
4661 results of expanding these sequences. Therefore you can concatenate them
4662 together or combine them with constant text in a single argument.
4666 Substitute one @samp{%} into the program name or argument.
4669 Substitute the name of the input file being processed.
4672 Substitute the basename of the input file being processed.
4673 This is the substring up to (and not including) the last period
4674 and not including the directory.
4677 This is the same as @samp{%b}, but include the file suffix (text after
4681 Marks the argument containing or following the @samp{%d} as a
4682 temporary file name, so that that file will be deleted if GCC exits
4683 successfully. Unlike @samp{%g}, this contributes no text to the
4686 @item %g@var{suffix}
4687 Substitute a file name that has suffix @var{suffix} and is chosen
4688 once per compilation, and mark the argument in the same way as
4689 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4690 name is now chosen in a way that is hard to predict even when previously
4691 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
4692 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4693 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4694 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4695 was simply substituted with a file name chosen once per compilation,
4696 without regard to any appended suffix (which was therefore treated
4697 just like ordinary text), making such attacks more likely to succeed.
4699 @item %u@var{suffix}
4700 Like @samp{%g}, but generates a new temporary file name even if
4701 @samp{%u@var{suffix}} was already seen.
4703 @item %U@var{suffix}
4704 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4705 new one if there is no such last file name. In the absence of any
4706 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4707 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
4708 would involve the generation of two distinct file names, one
4709 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4710 simply substituted with a file name chosen for the previous @samp{%u},
4711 without regard to any appended suffix.
4713 @item %j@var{SUFFIX}
4714 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
4715 writable, and if save-temps is off; otherwise, substitute the name
4716 of a temporary file, just like @samp{%u}. This temporary file is not
4717 meant for communication between processes, but rather as a junk
4720 @item %.@var{SUFFIX}
4721 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4722 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4723 terminated by the next space or %.
4726 Marks the argument containing or following the @samp{%w} as the
4727 designated output file of this compilation. This puts the argument
4728 into the sequence of arguments that @samp{%o} will substitute later.
4731 Substitutes the names of all the output files, with spaces
4732 automatically placed around them. You should write spaces
4733 around the @samp{%o} as well or the results are undefined.
4734 @samp{%o} is for use in the specs for running the linker.
4735 Input files whose names have no recognized suffix are not compiled
4736 at all, but they are included among the output files, so they will
4740 Substitutes the suffix for object files. Note that this is
4741 handled specially when it immediately follows @samp{%g, %u, or %U},
4742 because of the need for those to form complete file names. The
4743 handling is such that @samp{%O} is treated exactly as if it had already
4744 been substituted, except that @samp{%g, %u, and %U} do not currently
4745 support additional @var{suffix} characters following @samp{%O} as they would
4746 following, for example, @samp{.o}.
4749 Substitutes the standard macro predefinitions for the
4750 current target machine. Use this when running @code{cpp}.
4753 Like @samp{%p}, but puts @samp{__} before and after the name of each
4754 predefined macro, except for macros that start with @samp{__} or with
4755 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4759 Substitute a @option{-iprefix} option made from @env{GCC_EXEC_PREFIX}.
4762 Current argument is the name of a library or startup file of some sort.
4763 Search for that file in a standard list of directories and substitute
4764 the full name found.
4767 Print @var{str} as an error message. @var{str} is terminated by a newline.
4768 Use this when inconsistent options are detected.
4771 Output @samp{-} if the input for the current command is coming from a pipe.
4774 Substitute the contents of spec string @var{name} at this point.
4777 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
4779 @item %x@{@var{option}@}
4780 Accumulate an option for @samp{%X}.
4783 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
4787 Output the accumulated assembler options specified by @option{-Wa}.
4790 Output the accumulated preprocessor options specified by @option{-Wp}.
4793 Substitute the major version number of GCC@.
4794 (For version 2.9.5, this is 2.)
4797 Substitute the minor version number of GCC@.
4798 (For version 2.9.5, this is 9.)
4801 Substitute the patch level number of GCC@.
4802 (For version 2.9.5, this is 5.)
4805 Process the @code{asm} spec. This is used to compute the
4806 switches to be passed to the assembler.
4809 Process the @code{asm_final} spec. This is a spec string for
4810 passing switches to an assembler post-processor, if such a program is
4814 Process the @code{link} spec. This is the spec for computing the
4815 command line passed to the linker. Typically it will make use of the
4816 @samp{%L %G %S %D and %E} sequences.
4819 Dump out a @option{-L} option for each directory that GCC believes might
4820 contain startup files. If the target supports multilibs then the
4821 current multilib directory will be prepended to each of these paths.
4824 Output the multilib directory with directory separators replaced with
4825 @samp{_}. If multilib directories are not set, or the multilib directory is
4826 @file{.} then this option emits nothing.
4829 Process the @code{lib} spec. This is a spec string for deciding which
4830 libraries should be included on the command line to the linker.
4833 Process the @code{libgcc} spec. This is a spec string for deciding
4834 which GCC support library should be included on the command line to the linker.
4837 Process the @code{startfile} spec. This is a spec for deciding which
4838 object files should be the first ones passed to the linker. Typically
4839 this might be a file named @file{crt0.o}.
4842 Process the @code{endfile} spec. This is a spec string that specifies
4843 the last object files that will be passed to the linker.
4846 Process the @code{cpp} spec. This is used to construct the arguments
4847 to be passed to the C preprocessor.
4850 Process the @code{signed_char} spec. This is intended to be used
4851 to tell cpp whether a char is signed. It typically has the definition:
4853 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4857 Process the @code{cc1} spec. This is used to construct the options to be
4858 passed to the actual C compiler (@samp{cc1}).
4861 Process the @code{cc1plus} spec. This is used to construct the options to be
4862 passed to the actual C++ compiler (@samp{cc1plus}).
4865 Substitute the variable part of a matched option. See below.
4866 Note that each comma in the substituted string is replaced by
4870 Substitutes the @code{-S} switch, if that switch was given to GCC@.
4871 If that switch was not specified, this substitutes nothing. Note that
4872 the leading dash is omitted when specifying this option, and it is
4873 automatically inserted if the substitution is performed. Thus the spec
4874 string @samp{%@{foo@}} would match the command-line option @option{-foo}
4875 and would output the command line option @option{-foo}.
4877 @item %W@{@code{S}@}
4878 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4881 @item %@{@code{S}*@}
4882 Substitutes all the switches specified to GCC whose names start
4883 with @code{-S}, but which also take an argument. This is used for
4884 switches like @option{-o}, @option{-D}, @option{-I}, etc.
4885 GCC considers @option{-o foo} as being
4886 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4887 text, including the space. Thus two arguments would be generated.
4889 @item %@{^@code{S}*@}
4890 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4891 argument. Thus %@{^o*@} would only generate one argument, not two.
4893 @item %@{@code{S}*&@code{T}*@}
4894 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
4895 (the order of @code{S} and @code{T} in the spec is not significant).
4896 There can be any number of ampersand-separated variables; for each the
4897 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
4899 @item %@{<@code{S}@}
4900 Remove all occurrences of @code{-S} from the command line. Note---this
4901 command is position dependent. @samp{%} commands in the spec string
4902 before this option will see @code{-S}, @samp{%} commands in the spec
4903 string after this option will not.
4905 @item %@{@code{S}*:@code{X}@}
4906 Substitutes @code{X} if one or more switches whose names start with
4907 @code{-S} are specified to GCC@. Note that the tail part of the
4908 @code{-S} option (i.e.@: the part matched by the @samp{*}) will be substituted
4909 for each occurrence of @samp{%*} within @code{X}.
4911 @item %@{@code{S}:@code{X}@}
4912 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC@.
4914 @item %@{!@code{S}:@code{X}@}
4915 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC@.
4917 @item %@{|@code{S}:@code{X}@}
4918 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4920 @item %@{|!@code{S}:@code{X}@}
4921 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4923 @item %@{.@code{S}:@code{X}@}
4924 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4926 @item %@{!.@code{S}:@code{X}@}
4927 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4929 @item %@{@code{S}|@code{P}:@code{X}@}
4930 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@. This may be
4931 combined with @samp{!} and @samp{.} sequences as well, although they
4932 have a stronger binding than the @samp{|}. For example a spec string
4936 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4939 will output the following command-line options from the following input
4940 command-line options:
4945 -d fred.c -foo -baz -boggle
4946 -d jim.d -bar -baz -boggle
4951 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4952 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4953 or spaces, or even newlines. They are processed as usual, as described
4956 The @option{-O}, @option{-f}, @option{-m}, and @option{-W}
4957 switches are handled specifically in these
4958 constructs. If another value of @option{-O} or the negated form of a @option{-f}, @option{-m}, or
4959 @option{-W} switch is found later in the command line, the earlier switch
4960 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4961 letter, which passes all matching options.
4963 The character @samp{|} at the beginning of the predicate text is used to indicate
4964 that a command should be piped to the following command, but only if @option{-pipe}
4967 It is built into GCC which switches take arguments and which do not.
4968 (You might think it would be useful to generalize this to allow each
4969 compiler's spec to say which switches take arguments. But this cannot
4970 be done in a consistent fashion. GCC cannot even decide which input
4971 files have been specified without knowing which switches take arguments,
4972 and it must know which input files to compile in order to tell which
4975 GCC also knows implicitly that arguments starting in @option{-l} are to be
4976 treated as compiler output files, and passed to the linker in their
4977 proper position among the other output files.
4979 @c man begin OPTIONS
4981 @node Target Options
4982 @section Specifying Target Machine and Compiler Version
4983 @cindex target options
4984 @cindex cross compiling
4985 @cindex specifying machine version
4986 @cindex specifying compiler version and target machine
4987 @cindex compiler version, specifying
4988 @cindex target machine, specifying
4990 By default, GCC compiles code for the same type of machine that you
4991 are using. However, it can also be installed as a cross-compiler, to
4992 compile for some other type of machine. In fact, several different
4993 configurations of GCC, for different target machines, can be
4994 installed side by side. Then you specify which one to use with the
4997 In addition, older and newer versions of GCC can be installed side
4998 by side. One of them (probably the newest) will be the default, but
4999 you may sometimes wish to use another.
5002 @item -b @var{machine}
5004 The argument @var{machine} specifies the target machine for compilation.
5005 This is useful when you have installed GCC as a cross-compiler.
5007 The value to use for @var{machine} is the same as was specified as the
5008 machine type when configuring GCC as a cross-compiler. For
5009 example, if a cross-compiler was configured with @samp{configure
5010 i386v}, meaning to compile for an 80386 running System V, then you
5011 would specify @option{-b i386v} to run that cross compiler.
5013 When you do not specify @option{-b}, it normally means to compile for
5014 the same type of machine that you are using.
5016 @item -V @var{version}
5018 The argument @var{version} specifies which version of GCC to run.
5019 This is useful when multiple versions are installed. For example,
5020 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
5022 The default version, when you do not specify @option{-V}, is the last
5023 version of GCC that you installed.
5026 The @option{-b} and @option{-V} options actually work by controlling part of
5027 the file name used for the executable files and libraries used for
5028 compilation. A given version of GCC, for a given target machine, is
5029 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.
5031 Thus, sites can customize the effect of @option{-b} or @option{-V} either by
5032 changing the names of these directories or adding alternate names (or
5033 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
5034 file @file{80386} is a link to the file @file{i386v}, then @option{-b
5035 80386} becomes an alias for @option{-b i386v}.
5037 In one respect, the @option{-b} or @option{-V} do not completely change
5038 to a different compiler: the top-level driver program @command{gcc}
5039 that you originally invoked continues to run and invoke the other
5040 executables (preprocessor, compiler per se, assembler and linker)
5041 that do the real work. However, since no real work is done in the
5042 driver program, it usually does not matter that the driver program
5043 in use is not the one for the specified target. It is common for the
5044 interface to the other executables to change incompatibly between
5045 compiler versions, so unless the version specified is very close to that
5046 of the driver (for example, @option{-V 3.0} with a driver program from GCC
5047 version 3.0.1), use of @option{-V} may not work; for example, using
5048 @option{-V 2.95.2} will not work with a driver program from GCC 3.0.
5050 The only way that the driver program depends on the target machine is
5051 in the parsing and handling of special machine-specific options.
5052 However, this is controlled by a file which is found, along with the
5053 other executables, in the directory for the specified version and
5054 target machine. As a result, a single installed driver program adapts
5055 to any specified target machine, and sufficiently similar compiler
5058 The driver program executable does control one significant thing,
5059 however: the default version and target machine. Therefore, you can
5060 install different instances of the driver program, compiled for
5061 different targets or versions, under different names.
5063 For example, if the driver for version 2.0 is installed as @command{ogcc}
5064 and that for version 2.1 is installed as @command{gcc}, then the command
5065 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
5066 2.0 by default. However, you can choose either version with either
5067 command with the @option{-V} option.
5069 @node Submodel Options
5070 @section Hardware Models and Configurations
5071 @cindex submodel options
5072 @cindex specifying hardware config
5073 @cindex hardware models and configurations, specifying
5074 @cindex machine dependent options
5076 Earlier we discussed the standard option @option{-b} which chooses among
5077 different installed compilers for completely different target
5078 machines, such as VAX vs.@: 68000 vs.@: 80386.
5080 In addition, each of these target machine types can have its own
5081 special options, starting with @samp{-m}, to choose among various
5082 hardware models or configurations---for example, 68010 vs 68020,
5083 floating coprocessor or none. A single installed version of the
5084 compiler can compile for any model or configuration, according to the
5087 Some configurations of the compiler also support additional special
5088 options, usually for compatibility with other compilers on the same
5092 These options are defined by the macro @code{TARGET_SWITCHES} in the
5093 machine description. The default for the options is also defined by
5094 that macro, which enables you to change the defaults.
5109 * RS/6000 and PowerPC Options::
5112 * i386 and x86-64 Options::
5114 * Intel 960 Options::
5115 * DEC Alpha Options::
5119 * System V Options::
5120 * TMS320C3x/C4x Options::
5128 * S/390 and zSeries Options::
5133 @node M680x0 Options
5134 @subsection M680x0 Options
5135 @cindex M680x0 options
5137 These are the @samp{-m} options defined for the 68000 series. The default
5138 values for these options depends on which style of 68000 was selected when
5139 the compiler was configured; the defaults for the most common choices are
5147 Generate output for a 68000. This is the default
5148 when the compiler is configured for 68000-based systems.
5150 Use this option for microcontrollers with a 68000 or EC000 core,
5151 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5157 Generate output for a 68020. This is the default
5158 when the compiler is configured for 68020-based systems.
5162 Generate output containing 68881 instructions for floating point.
5163 This is the default for most 68020 systems unless @option{--nfp} was
5164 specified when the compiler was configured.
5168 Generate output for a 68030. This is the default when the compiler is
5169 configured for 68030-based systems.
5173 Generate output for a 68040. This is the default when the compiler is
5174 configured for 68040-based systems.
5176 This option inhibits the use of 68881/68882 instructions that have to be
5177 emulated by software on the 68040. Use this option if your 68040 does not
5178 have code to emulate those instructions.
5182 Generate output for a 68060. This is the default when the compiler is
5183 configured for 68060-based systems.
5185 This option inhibits the use of 68020 and 68881/68882 instructions that
5186 have to be emulated by software on the 68060. Use this option if your 68060
5187 does not have code to emulate those instructions.
5191 Generate output for a CPU32. This is the default
5192 when the compiler is configured for CPU32-based systems.
5194 Use this option for microcontrollers with a
5195 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5196 68336, 68340, 68341, 68349 and 68360.
5200 Generate output for a 520X ``coldfire'' family cpu. This is the default
5201 when the compiler is configured for 520X-based systems.
5203 Use this option for microcontroller with a 5200 core, including
5204 the MCF5202, MCF5203, MCF5204 and MCF5202.
5209 Generate output for a 68040, without using any of the new instructions.
5210 This results in code which can run relatively efficiently on either a
5211 68020/68881 or a 68030 or a 68040. The generated code does use the
5212 68881 instructions that are emulated on the 68040.
5216 Generate output for a 68060, without using any of the new instructions.
5217 This results in code which can run relatively efficiently on either a
5218 68020/68881 or a 68030 or a 68040. The generated code does use the
5219 68881 instructions that are emulated on the 68060.
5223 Generate output containing Sun FPA instructions for floating point.
5226 @opindex msoft-float
5227 Generate output containing library calls for floating point.
5228 @strong{Warning:} the requisite libraries are not available for all m68k
5229 targets. Normally the facilities of the machine's usual C compiler are
5230 used, but this can't be done directly in cross-compilation. You must
5231 make your own arrangements to provide suitable library functions for
5232 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5233 @samp{m68k-*-coff} do provide software floating point support.
5237 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5240 @opindex mnobitfield
5241 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5242 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5246 Do use the bit-field instructions. The @option{-m68020} option implies
5247 @option{-mbitfield}. This is the default if you use a configuration
5248 designed for a 68020.
5252 Use a different function-calling convention, in which functions
5253 that take a fixed number of arguments return with the @code{rtd}
5254 instruction, which pops their arguments while returning. This
5255 saves one instruction in the caller since there is no need to pop
5256 the arguments there.
5258 This calling convention is incompatible with the one normally
5259 used on Unix, so you cannot use it if you need to call libraries
5260 compiled with the Unix compiler.
5262 Also, you must provide function prototypes for all functions that
5263 take variable numbers of arguments (including @code{printf});
5264 otherwise incorrect code will be generated for calls to those
5267 In addition, seriously incorrect code will result if you call a
5268 function with too many arguments. (Normally, extra arguments are
5269 harmlessly ignored.)
5271 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5272 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5275 @itemx -mno-align-int
5277 @opindex mno-align-int
5278 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5279 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5280 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5281 Aligning variables on 32-bit boundaries produces code that runs somewhat
5282 faster on processors with 32-bit busses at the expense of more memory.
5284 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5285 align structures containing the above types differently than
5286 most published application binary interface specifications for the m68k.
5290 Use the pc-relative addressing mode of the 68000 directly, instead of
5291 using a global offset table. At present, this option implies @option{-fpic},
5292 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5293 not presently supported with @option{-mpcrel}, though this could be supported for
5294 68020 and higher processors.
5296 @item -mno-strict-align
5297 @itemx -mstrict-align
5298 @opindex mno-strict-align
5299 @opindex mstrict-align
5300 Do not (do) assume that unaligned memory references will be handled by
5305 @node M68hc1x Options
5306 @subsection M68hc1x Options
5307 @cindex M68hc1x options
5309 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5310 microcontrollers. The default values for these options depends on
5311 which style of microcontroller was selected when the compiler was configured;
5312 the defaults for the most common choices are given below.
5319 Generate output for a 68HC11. This is the default
5320 when the compiler is configured for 68HC11-based systems.
5326 Generate output for a 68HC12. This is the default
5327 when the compiler is configured for 68HC12-based systems.
5330 @opindex mauto-incdec
5331 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5336 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5338 @item -msoft-reg-count=@var{count}
5339 @opindex msoft-reg-count
5340 Specify the number of pseudo-soft registers which are used for the
5341 code generation. The maximum number is 32. Using more pseudo-soft
5342 register may or may not result in better code depending on the program.
5343 The default is 4 for 68HC11 and 2 for 68HC12.
5348 @subsection VAX Options
5351 These @samp{-m} options are defined for the VAX:
5356 Do not output certain jump instructions (@code{aobleq} and so on)
5357 that the Unix assembler for the VAX cannot handle across long
5362 Do output those jump instructions, on the assumption that you
5363 will assemble with the GNU assembler.
5367 Output code for g-format floating point numbers instead of d-format.
5371 @subsection SPARC Options
5372 @cindex SPARC options
5374 These @samp{-m} switches are supported on the SPARC:
5379 @opindex mno-app-regs
5381 Specify @option{-mapp-regs} to generate output using the global registers
5382 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5385 To be fully SVR4 ABI compliant at the cost of some performance loss,
5386 specify @option{-mno-app-regs}. You should compile libraries and system
5387 software with this option.
5392 @opindex mhard-float
5393 Generate output containing floating point instructions. This is the
5399 @opindex msoft-float
5400 Generate output containing library calls for floating point.
5401 @strong{Warning:} the requisite libraries are not available for all SPARC
5402 targets. Normally the facilities of the machine's usual C compiler are
5403 used, but this cannot be done directly in cross-compilation. You must make
5404 your own arrangements to provide suitable library functions for
5405 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5406 @samp{sparclite-*-*} do provide software floating point support.
5408 @option{-msoft-float} changes the calling convention in the output file;
5409 therefore, it is only useful if you compile @emph{all} of a program with
5410 this option. In particular, you need to compile @file{libgcc.a}, the
5411 library that comes with GCC, with @option{-msoft-float} in order for
5414 @item -mhard-quad-float
5415 @opindex mhard-quad-float
5416 Generate output containing quad-word (long double) floating point
5419 @item -msoft-quad-float
5420 @opindex msoft-quad-float
5421 Generate output containing library calls for quad-word (long double)
5422 floating point instructions. The functions called are those specified
5423 in the SPARC ABI@. This is the default.
5425 As of this writing, there are no sparc implementations that have hardware
5426 support for the quad-word floating point instructions. They all invoke
5427 a trap handler for one of these instructions, and then the trap handler
5428 emulates the effect of the instruction. Because of the trap handler overhead,
5429 this is much slower than calling the ABI library routines. Thus the
5430 @option{-msoft-quad-float} option is the default.
5434 @opindex mno-epilogue
5436 With @option{-mepilogue} (the default), the compiler always emits code for
5437 function exit at the end of each function. Any function exit in
5438 the middle of the function (such as a return statement in C) will
5439 generate a jump to the exit code at the end of the function.
5441 With @option{-mno-epilogue}, the compiler tries to emit exit code inline
5442 at every function exit.
5448 With @option{-mflat}, the compiler does not generate save/restore instructions
5449 and will use a ``flat'' or single register window calling convention.
5450 This model uses %i7 as the frame pointer and is compatible with the normal
5451 register window model. Code from either may be intermixed.
5452 The local registers and the input registers (0--5) are still treated as
5453 ``call saved'' registers and will be saved on the stack as necessary.
5455 With @option{-mno-flat} (the default), the compiler emits save/restore
5456 instructions (except for leaf functions) and is the normal mode of operation.
5458 @item -mno-unaligned-doubles
5459 @itemx -munaligned-doubles
5460 @opindex mno-unaligned-doubles
5461 @opindex munaligned-doubles
5462 Assume that doubles have 8 byte alignment. This is the default.
5464 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5465 alignment only if they are contained in another type, or if they have an
5466 absolute address. Otherwise, it assumes they have 4 byte alignment.
5467 Specifying this option avoids some rare compatibility problems with code
5468 generated by other compilers. It is not the default because it results
5469 in a performance loss, especially for floating point code.
5471 @item -mno-faster-structs
5472 @itemx -mfaster-structs
5473 @opindex mno-faster-structs
5474 @opindex mfaster-structs
5475 With @option{-mfaster-structs}, the compiler assumes that structures
5476 should have 8 byte alignment. This enables the use of pairs of
5477 @code{ldd} and @code{std} instructions for copies in structure
5478 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5479 However, the use of this changed alignment directly violates the Sparc
5480 ABI@. Thus, it's intended only for use on targets where the developer
5481 acknowledges that their resulting code will not be directly in line with
5482 the rules of the ABI@.
5488 These two options select variations on the SPARC architecture.
5490 By default (unless specifically configured for the Fujitsu SPARClite),
5491 GCC generates code for the v7 variant of the SPARC architecture.
5493 @option{-mv8} will give you SPARC v8 code. The only difference from v7
5494 code is that the compiler emits the integer multiply and integer
5495 divide instructions which exist in SPARC v8 but not in SPARC v7.
5497 @option{-msparclite} will give you SPARClite code. This adds the integer
5498 multiply, integer divide step and scan (@code{ffs}) instructions which
5499 exist in SPARClite but not in SPARC v7.
5501 These options are deprecated and will be deleted in a future GCC release.
5502 They have been replaced with @option{-mcpu=xxx}.
5507 @opindex msupersparc
5508 These two options select the processor for which the code is optimised.
5510 With @option{-mcypress} (the default), the compiler optimizes code for the
5511 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
5512 This is also appropriate for the older SparcStation 1, 2, IPX etc.
5514 With @option{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
5515 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
5516 of the full SPARC v8 instruction set.
5518 These options are deprecated and will be deleted in a future GCC release.
5519 They have been replaced with @option{-mcpu=xxx}.
5521 @item -mcpu=@var{cpu_type}
5523 Set the instruction set, register set, and instruction scheduling parameters
5524 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
5525 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
5526 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
5527 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
5529 Default instruction scheduling parameters are used for values that select
5530 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
5531 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
5533 Here is a list of each supported architecture and their supported
5538 v8: supersparc, hypersparc
5539 sparclite: f930, f934, sparclite86x
5544 @item -mtune=@var{cpu_type}
5546 Set the instruction scheduling parameters for machine type
5547 @var{cpu_type}, but do not set the instruction set or register set that the
5548 option @option{-mcpu=@var{cpu_type}} would.
5550 The same values for @option{-mcpu=@var{cpu_type}} can be used for
5551 @option{-mtune=@var{cpu_type}}, but the only useful values are those
5552 that select a particular cpu implementation. Those are @samp{cypress},
5553 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
5554 @samp{sparclite86x}, @samp{tsc701}, and @samp{ultrasparc}.
5558 These @samp{-m} switches are supported in addition to the above
5559 on the SPARCLET processor.
5562 @item -mlittle-endian
5563 @opindex mlittle-endian
5564 Generate code for a processor running in little-endian mode.
5568 Treat register @code{%g0} as a normal register.
5569 GCC will continue to clobber it as necessary but will not assume
5570 it always reads as 0.
5572 @item -mbroken-saverestore
5573 @opindex mbroken-saverestore
5574 Generate code that does not use non-trivial forms of the @code{save} and
5575 @code{restore} instructions. Early versions of the SPARCLET processor do
5576 not correctly handle @code{save} and @code{restore} instructions used with
5577 arguments. They correctly handle them used without arguments. A @code{save}
5578 instruction used without arguments increments the current window pointer
5579 but does not allocate a new stack frame. It is assumed that the window
5580 overflow trap handler will properly handle this case as will interrupt
5584 These @samp{-m} switches are supported in addition to the above
5585 on SPARC V9 processors in 64-bit environments.
5588 @item -mlittle-endian
5589 @opindex mlittle-endian
5590 Generate code for a processor running in little-endian mode.
5596 Generate code for a 32-bit or 64-bit environment.
5597 The 32-bit environment sets int, long and pointer to 32 bits.
5598 The 64-bit environment sets int to 32 bits and long and pointer
5601 @item -mcmodel=medlow
5602 @opindex mcmodel=medlow
5603 Generate code for the Medium/Low code model: the program must be linked
5604 in the low 32 bits of the address space. Pointers are 64 bits.
5605 Programs can be statically or dynamically linked.
5607 @item -mcmodel=medmid
5608 @opindex mcmodel=medmid
5609 Generate code for the Medium/Middle code model: the program must be linked
5610 in the low 44 bits of the address space, the text segment must be less than
5611 2G bytes, and data segment must be within 2G of the text segment.
5612 Pointers are 64 bits.
5614 @item -mcmodel=medany
5615 @opindex mcmodel=medany
5616 Generate code for the Medium/Anywhere code model: the program may be linked
5617 anywhere in the address space, the text segment must be less than
5618 2G bytes, and data segment must be within 2G of the text segment.
5619 Pointers are 64 bits.
5621 @item -mcmodel=embmedany
5622 @opindex mcmodel=embmedany
5623 Generate code for the Medium/Anywhere code model for embedded systems:
5624 assume a 32-bit text and a 32-bit data segment, both starting anywhere
5625 (determined at link time). Register %g4 points to the base of the
5626 data segment. Pointers are still 64 bits.
5627 Programs are statically linked, PIC is not supported.
5630 @itemx -mno-stack-bias
5631 @opindex mstack-bias
5632 @opindex mno-stack-bias
5633 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
5634 frame pointer if present, are offset by @minus{}2047 which must be added back
5635 when making stack frame references.
5636 Otherwise, assume no such offset is present.
5639 @node Convex Options
5640 @subsection Convex Options
5641 @cindex Convex options
5643 These @samp{-m} options are defined for Convex:
5648 Generate output for C1. The code will run on any Convex machine.
5649 The preprocessor symbol @code{__convex__c1__} is defined.
5653 Generate output for C2. Uses instructions not available on C1.
5654 Scheduling and other optimizations are chosen for max performance on C2.
5655 The preprocessor symbol @code{__convex_c2__} is defined.
5659 Generate output for C32xx. Uses instructions not available on C1.
5660 Scheduling and other optimizations are chosen for max performance on C32.
5661 The preprocessor symbol @code{__convex_c32__} is defined.
5665 Generate output for C34xx. Uses instructions not available on C1.
5666 Scheduling and other optimizations are chosen for max performance on C34.
5667 The preprocessor symbol @code{__convex_c34__} is defined.
5671 Generate output for C38xx. Uses instructions not available on C1.
5672 Scheduling and other optimizations are chosen for max performance on C38.
5673 The preprocessor symbol @code{__convex_c38__} is defined.
5677 Generate code which puts an argument count in the word preceding each
5678 argument list. This is compatible with regular CC, and a few programs
5679 may need the argument count word. GDB and other source-level debuggers
5680 do not need it; this info is in the symbol table.
5683 @opindex mnoargcount
5684 Omit the argument count word. This is the default.
5686 @item -mvolatile-cache
5687 @opindex mvolatile-cache
5688 Allow volatile references to be cached. This is the default.
5690 @item -mvolatile-nocache
5691 @opindex mvolatile-nocache
5692 Volatile references bypass the data cache, going all the way to memory.
5693 This is only needed for multi-processor code that does not use standard
5694 synchronization instructions. Making non-volatile references to volatile
5695 locations will not necessarily work.
5699 Type long is 32 bits, the same as type int. This is the default.
5703 Type long is 64 bits, the same as type long long. This option is useless,
5704 because no library support exists for it.
5707 @node AMD29K Options
5708 @subsection AMD29K Options
5709 @cindex AMD29K options
5711 These @samp{-m} options are defined for the AMD Am29000:
5716 @cindex DW bit (29k)
5717 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5718 halfword operations are directly supported by the hardware. This is the
5723 Generate code that assumes the @code{DW} bit is not set.
5727 @cindex byte writes (29k)
5728 Generate code that assumes the system supports byte and halfword write
5729 operations. This is the default.
5733 Generate code that assumes the systems does not support byte and
5734 halfword write operations. @option{-mnbw} implies @option{-mndw}.
5738 @cindex memory model (29k)
5739 Use a small memory model that assumes that all function addresses are
5740 either within a single 256 KB segment or at an absolute address of less
5741 than 256k. This allows the @code{call} instruction to be used instead
5742 of a @code{const}, @code{consth}, @code{calli} sequence.
5746 Use the normal memory model: Generate @code{call} instructions only when
5747 calling functions in the same file and @code{calli} instructions
5748 otherwise. This works if each file occupies less than 256 KB but allows
5749 the entire executable to be larger than 256 KB@. This is the default.
5753 Always use @code{calli} instructions. Specify this option if you expect
5754 a single file to compile into more than 256 KB of code.
5758 @cindex processor selection (29k)
5759 Generate code for the Am29050.
5763 Generate code for the Am29000. This is the default.
5765 @item -mkernel-registers
5766 @opindex mkernel-registers
5767 @cindex kernel and user registers (29k)
5768 Generate references to registers @code{gr64-gr95} instead of to
5769 registers @code{gr96-gr127}. This option can be used when compiling
5770 kernel code that wants a set of global registers disjoint from that used
5773 Note that when this option is used, register names in @samp{-f} flags
5774 must use the normal, user-mode, names.
5776 @item -muser-registers
5777 @opindex muser-registers
5778 Use the normal set of global registers, @code{gr96-gr127}. This is the
5782 @itemx -mno-stack-check
5783 @opindex mstack-check
5784 @opindex mno-stack-check
5785 @cindex stack checks (29k)
5786 Insert (or do not insert) a call to @code{__msp_check} after each stack
5787 adjustment. This is often used for kernel code.
5790 @itemx -mno-storem-bug
5791 @opindex mstorem-bug
5792 @opindex mno-storem-bug
5793 @cindex storem bug (29k)
5794 @option{-mstorem-bug} handles 29k processors which cannot handle the
5795 separation of a mtsrim insn and a storem instruction (most 29000 chips
5796 to date, but not the 29050).
5798 @item -mno-reuse-arg-regs
5799 @itemx -mreuse-arg-regs
5800 @opindex mno-reuse-arg-regs
5801 @opindex mreuse-arg-regs
5802 @option{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5803 registers for copying out arguments. This helps detect calling a function
5804 with fewer arguments than it was declared with.
5806 @item -mno-impure-text
5807 @itemx -mimpure-text
5808 @opindex mno-impure-text
5809 @opindex mimpure-text
5810 @option{-mimpure-text}, used in addition to @option{-shared}, tells the compiler to
5811 not pass @option{-assert pure-text} to the linker when linking a shared object.
5814 @opindex msoft-float
5815 Generate output containing library calls for floating point.
5816 @strong{Warning:} the requisite libraries are not part of GCC@.
5817 Normally the facilities of the machine's usual C compiler are used, but
5818 this can't be done directly in cross-compilation. You must make your
5819 own arrangements to provide suitable library functions for
5824 Do not generate multm or multmu instructions. This is useful for some embedded
5825 systems which do not have trap handlers for these instructions.
5829 @subsection ARM Options
5832 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5837 @opindex mapcs-frame
5838 Generate a stack frame that is compliant with the ARM Procedure Call
5839 Standard for all functions, even if this is not strictly necessary for
5840 correct execution of the code. Specifying @option{-fomit-frame-pointer}
5841 with this option will cause the stack frames not to be generated for
5842 leaf functions. The default is @option{-mno-apcs-frame}.
5846 This is a synonym for @option{-mapcs-frame}.
5850 Generate code for a processor running with a 26-bit program counter,
5851 and conforming to the function calling standards for the APCS 26-bit
5852 option. This option replaces the @option{-m2} and @option{-m3} options
5853 of previous releases of the compiler.
5857 Generate code for a processor running with a 32-bit program counter,
5858 and conforming to the function calling standards for the APCS 32-bit
5859 option. This option replaces the @option{-m6} option of previous releases
5863 @c not currently implemented
5864 @item -mapcs-stack-check
5865 @opindex mapcs-stack-check
5866 Generate code to check the amount of stack space available upon entry to
5867 every function (that actually uses some stack space). If there is
5868 insufficient space available then either the function
5869 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5870 called, depending upon the amount of stack space required. The run time
5871 system is required to provide these functions. The default is
5872 @option{-mno-apcs-stack-check}, since this produces smaller code.
5874 @c not currently implemented
5876 @opindex mapcs-float
5877 Pass floating point arguments using the float point registers. This is
5878 one of the variants of the APCS@. This option is recommended if the
5879 target hardware has a floating point unit or if a lot of floating point
5880 arithmetic is going to be performed by the code. The default is
5881 @option{-mno-apcs-float}, since integer only code is slightly increased in
5882 size if @option{-mapcs-float} is used.
5884 @c not currently implemented
5885 @item -mapcs-reentrant
5886 @opindex mapcs-reentrant
5887 Generate reentrant, position independent code. The default is
5888 @option{-mno-apcs-reentrant}.
5891 @item -mthumb-interwork
5892 @opindex mthumb-interwork
5893 Generate code which supports calling between the ARM and Thumb
5894 instruction sets. Without this option the two instruction sets cannot
5895 be reliably used inside one program. The default is
5896 @option{-mno-thumb-interwork}, since slightly larger code is generated
5897 when @option{-mthumb-interwork} is specified.
5899 @item -mno-sched-prolog
5900 @opindex mno-sched-prolog
5901 Prevent the reordering of instructions in the function prolog, or the
5902 merging of those instruction with the instructions in the function's
5903 body. This means that all functions will start with a recognizable set
5904 of instructions (or in fact one of a choice from a small set of
5905 different function prologues), and this information can be used to
5906 locate the start if functions inside an executable piece of code. The
5907 default is @option{-msched-prolog}.
5910 @opindex mhard-float
5911 Generate output containing floating point instructions. This is the
5915 @opindex msoft-float
5916 Generate output containing library calls for floating point.
5917 @strong{Warning:} the requisite libraries are not available for all ARM
5918 targets. Normally the facilities of the machine's usual C compiler are
5919 used, but this cannot be done directly in cross-compilation. You must make
5920 your own arrangements to provide suitable library functions for
5923 @option{-msoft-float} changes the calling convention in the output file;
5924 therefore, it is only useful if you compile @emph{all} of a program with
5925 this option. In particular, you need to compile @file{libgcc.a}, the
5926 library that comes with GCC, with @option{-msoft-float} in order for
5929 @item -mlittle-endian
5930 @opindex mlittle-endian
5931 Generate code for a processor running in little-endian mode. This is
5932 the default for all standard configurations.
5935 @opindex mbig-endian
5936 Generate code for a processor running in big-endian mode; the default is
5937 to compile code for a little-endian processor.
5939 @item -mwords-little-endian
5940 @opindex mwords-little-endian
5941 This option only applies when generating code for big-endian processors.
5942 Generate code for a little-endian word order but a big-endian byte
5943 order. That is, a byte order of the form @samp{32107654}. Note: this
5944 option should only be used if you require compatibility with code for
5945 big-endian ARM processors generated by versions of the compiler prior to
5948 @item -malignment-traps
5949 @opindex malignment-traps
5950 Generate code that will not trap if the MMU has alignment traps enabled.
5951 On ARM architectures prior to ARMv4, there were no instructions to
5952 access half-word objects stored in memory. However, when reading from
5953 memory a feature of the ARM architecture allows a word load to be used,
5954 even if the address is unaligned, and the processor core will rotate the
5955 data as it is being loaded. This option tells the compiler that such
5956 misaligned accesses will cause a MMU trap and that it should instead
5957 synthesise the access as a series of byte accesses. The compiler can
5958 still use word accesses to load half-word data if it knows that the
5959 address is aligned to a word boundary.
5961 This option is ignored when compiling for ARM architecture 4 or later,
5962 since these processors have instructions to directly access half-word
5965 @item -mno-alignment-traps
5966 @opindex mno-alignment-traps
5967 Generate code that assumes that the MMU will not trap unaligned
5968 accesses. This produces better code when the target instruction set
5969 does not have half-word memory operations (i.e.@: implementations prior to
5972 Note that you cannot use this option to access unaligned word objects,
5973 since the processor will only fetch one 32-bit aligned object from
5976 The default setting for most targets is @option{-mno-alignment-traps}, since
5977 this produces better code when there are no half-word memory
5978 instructions available.
5980 @item -mshort-load-bytes
5981 @itemx -mno-short-load-words
5982 @opindex mshort-load-bytes
5983 @opindex mno-short-load-words
5984 These are deprecated aliases for @option{-malignment-traps}.
5986 @item -mno-short-load-bytes
5987 @itemx -mshort-load-words
5988 @opindex mno-short-load-bytes
5989 @opindex mshort-load-words
5990 This are deprecated aliases for @option{-mno-alignment-traps}.
5994 This option only applies to RISC iX@. Emulate the native BSD-mode
5995 compiler. This is the default if @option{-ansi} is not specified.
5999 This option only applies to RISC iX@. Emulate the native X/Open-mode
6002 @item -mno-symrename
6003 @opindex mno-symrename
6004 This option only applies to RISC iX@. Do not run the assembler
6005 post-processor, @samp{symrename}, after code has been assembled.
6006 Normally it is necessary to modify some of the standard symbols in
6007 preparation for linking with the RISC iX C library; this option
6008 suppresses this pass. The post-processor is never run when the
6009 compiler is built for cross-compilation.
6011 @item -mcpu=@var{name}
6013 This specifies the name of the target ARM processor. GCC uses this name
6014 to determine what kind of instructions it can emit when generating
6015 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
6016 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
6017 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
6018 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
6019 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
6020 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
6021 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
6022 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
6023 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
6024 @samp{arm1020t}, @samp{xscale}.
6026 @itemx -mtune=@var{name}
6028 This option is very similar to the @option{-mcpu=} option, except that
6029 instead of specifying the actual target processor type, and hence
6030 restricting which instructions can be used, it specifies that GCC should
6031 tune the performance of the code as if the target were of the type
6032 specified in this option, but still choosing the instructions that it
6033 will generate based on the cpu specified by a @option{-mcpu=} option.
6034 For some ARM implementations better performance can be obtained by using
6037 @item -march=@var{name}
6039 This specifies the name of the target ARM architecture. GCC uses this
6040 name to determine what kind of instructions it can emit when generating
6041 assembly code. This option can be used in conjunction with or instead
6042 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6043 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6044 @samp{armv5}, @samp{armv5t}, @samp{armv5te}.
6046 @item -mfpe=@var{number}
6047 @itemx -mfp=@var{number}
6050 This specifies the version of the floating point emulation available on
6051 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6052 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6054 @item -mstructure-size-boundary=@var{n}
6055 @opindex mstructure-size-boundary
6056 The size of all structures and unions will be rounded up to a multiple
6057 of the number of bits set by this option. Permissible values are 8 and
6058 32. The default value varies for different toolchains. For the COFF
6059 targeted toolchain the default value is 8. Specifying the larger number
6060 can produce faster, more efficient code, but can also increase the size
6061 of the program. The two values are potentially incompatible. Code
6062 compiled with one value cannot necessarily expect to work with code or
6063 libraries compiled with the other value, if they exchange information
6064 using structures or unions.
6066 @item -mabort-on-noreturn
6067 @opindex mabort-on-noreturn
6068 Generate a call to the function @code{abort} at the end of a
6069 @code{noreturn} function. It will be executed if the function tries to
6073 @itemx -mno-long-calls
6074 @opindex mlong-calls
6075 @opindex mno-long-calls
6076 Tells the compiler to perform function calls by first loading the
6077 address of the function into a register and then performing a subroutine
6078 call on this register. This switch is needed if the target function
6079 will lie outside of the 64 megabyte addressing range of the offset based
6080 version of subroutine call instruction.
6082 Even if this switch is enabled, not all function calls will be turned
6083 into long calls. The heuristic is that static functions, functions
6084 which have the @samp{short-call} attribute, functions that are inside
6085 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6086 definitions have already been compiled within the current compilation
6087 unit, will not be turned into long calls. The exception to this rule is
6088 that weak function definitions, functions with the @samp{long-call}
6089 attribute or the @samp{section} attribute, and functions that are within
6090 the scope of a @samp{#pragma long_calls} directive, will always be
6091 turned into long calls.
6093 This feature is not enabled by default. Specifying
6094 @option{-mno-long-calls} will restore the default behaviour, as will
6095 placing the function calls within the scope of a @samp{#pragma
6096 long_calls_off} directive. Note these switches have no effect on how
6097 the compiler generates code to handle function calls via function
6100 @item -mnop-fun-dllimport
6101 @opindex mnop-fun-dllimport
6102 Disable support for the @code{dllimport} attribute.
6104 @item -msingle-pic-base
6105 @opindex msingle-pic-base
6106 Treat the register used for PIC addressing as read-only, rather than
6107 loading it in the prologue for each function. The run-time system is
6108 responsible for initialising this register with an appropriate value
6109 before execution begins.
6111 @item -mpic-register=@var{reg}
6112 @opindex mpic-register
6113 Specify the register to be used for PIC addressing. The default is R10
6114 unless stack-checking is enabled, when R9 is used.
6116 @item -mpoke-function-name
6117 @opindex mpoke-function-name
6118 Write the name of each function into the text section, directly
6119 preceding the function prologue. The generated code is similar to this:
6123 .ascii "arm_poke_function_name", 0
6126 .word 0xff000000 + (t1 - t0)
6127 arm_poke_function_name
6129 stmfd sp!, @{fp, ip, lr, pc@}
6133 When performing a stack backtrace, code can inspect the value of
6134 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6135 location @code{pc - 12} and the top 8 bits are set, then we know that
6136 there is a function name embedded immediately preceding this location
6137 and has length @code{((pc[-3]) & 0xff000000)}.
6141 Generate code for the 16-bit Thumb instruction set. The default is to
6142 use the 32-bit ARM instruction set.
6145 @opindex mtpcs-frame
6146 Generate a stack frame that is compliant with the Thumb Procedure Call
6147 Standard for all non-leaf functions. (A leaf function is one that does
6148 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6150 @item -mtpcs-leaf-frame
6151 @opindex mtpcs-leaf-frame
6152 Generate a stack frame that is compliant with the Thumb Procedure Call
6153 Standard for all leaf functions. (A leaf function is one that does
6154 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6156 @item -mcallee-super-interworking
6157 @opindex mcallee-super-interworking
6158 Gives all externally visible functions in the file being compiled an ARM
6159 instruction set header which switches to Thumb mode before executing the
6160 rest of the function. This allows these functions to be called from
6161 non-interworking code.
6163 @item -mcaller-super-interworking
6164 @opindex mcaller-super-interworking
6165 Allows calls via function pointers (including virtual functions) to
6166 execute correctly regardless of whether the target code has been
6167 compiled for interworking or not. There is a small overhead in the cost
6168 of executing a function pointer if this option is enabled.
6172 @node MN10200 Options
6173 @subsection MN10200 Options
6174 @cindex MN10200 options
6175 These @option{-m} options are defined for Matsushita MN10200 architectures:
6180 Indicate to the linker that it should perform a relaxation optimization pass
6181 to shorten branches, calls and absolute memory addresses. This option only
6182 has an effect when used on the command line for the final link step.
6184 This option makes symbolic debugging impossible.
6187 @node MN10300 Options
6188 @subsection MN10300 Options
6189 @cindex MN10300 options
6190 These @option{-m} options are defined for Matsushita MN10300 architectures:
6195 Generate code to avoid bugs in the multiply instructions for the MN10300
6196 processors. This is the default.
6199 @opindex mno-mult-bug
6200 Do not generate code to avoid bugs in the multiply instructions for the
6205 Generate code which uses features specific to the AM33 processor.
6209 Do not generate code which uses features specific to the AM33 processor. This
6214 Do not link in the C run-time initialization object file.
6218 Indicate to the linker that it should perform a relaxation optimization pass
6219 to shorten branches, calls and absolute memory addresses. This option only
6220 has an effect when used on the command line for the final link step.
6222 This option makes symbolic debugging impossible.
6226 @node M32R/D Options
6227 @subsection M32R/D Options
6228 @cindex M32R/D options
6230 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6235 Generate code for the M32R/X. The default is to generate code for the M32R.
6237 @item -mcode-model=small
6238 @opindex mcode-model=small
6239 Assume all objects live in the lower 16MB of memory (so that their addresses
6240 can be loaded with the @code{ld24} instruction), and assume all subroutines
6241 are reachable with the @code{bl} instruction.
6242 This is the default.
6244 The addressability of a particular object can be set with the
6245 @code{model} attribute.
6247 @item -mcode-model=medium
6248 @opindex mcode-model=medium
6249 Assume objects may be anywhere in the 32-bit address space (the compiler
6250 will generate @code{seth/add3} instructions to load their addresses), and
6251 assume all subroutines are reachable with the @code{bl} instruction.
6253 @item -mcode-model=large
6254 @opindex mcode-model=large
6255 Assume objects may be anywhere in the 32-bit address space (the compiler
6256 will generate @code{seth/add3} instructions to load their addresses), and
6257 assume subroutines may not be reachable with the @code{bl} instruction
6258 (the compiler will generate the much slower @code{seth/add3/jl}
6259 instruction sequence).
6262 @opindex msdata=none
6263 Disable use of the small data area. Variables will be put into
6264 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6265 @code{section} attribute has been specified).
6266 This is the default.
6268 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6269 Objects may be explicitly put in the small data area with the
6270 @code{section} attribute using one of these sections.
6273 @opindex msdata=sdata
6274 Put small global and static data in the small data area, but do not
6275 generate special code to reference them.
6279 Put small global and static data in the small data area, and generate
6280 special instructions to reference them.
6284 @cindex smaller data references
6285 Put global and static objects less than or equal to @var{num} bytes
6286 into the small data or bss sections instead of the normal data or bss
6287 sections. The default value of @var{num} is 8.
6288 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6289 for this option to have any effect.
6291 All modules should be compiled with the same @option{-G @var{num}} value.
6292 Compiling with different values of @var{num} may or may not work; if it
6293 doesn't the linker will give an error message---incorrect code will not be
6299 @subsection M88K Options
6300 @cindex M88k options
6302 These @samp{-m} options are defined for Motorola 88k architectures:
6307 Generate code that works well on both the m88100 and the
6312 Generate code that works best for the m88100, but that also
6317 Generate code that works best for the m88110, and may not run
6322 Obsolete option to be removed from the next revision.
6325 @item -midentify-revision
6326 @opindex midentify-revision
6327 @cindex identifying source, compiler (88k)
6328 Include an @code{ident} directive in the assembler output recording the
6329 source file name, compiler name and version, timestamp, and compilation
6332 @item -mno-underscores
6333 @opindex mno-underscores
6334 @cindex underscores, avoiding (88k)
6335 In assembler output, emit symbol names without adding an underscore
6336 character at the beginning of each name. The default is to use an
6337 underscore as prefix on each name.
6339 @item -mocs-debug-info
6340 @itemx -mno-ocs-debug-info
6341 @opindex mocs-debug-info
6342 @opindex mno-ocs-debug-info
6344 @cindex debugging, 88k OCS
6345 Include (or omit) additional debugging information (about registers used
6346 in each stack frame) as specified in the 88open Object Compatibility
6347 Standard, ``OCS''@. This extra information allows debugging of code that
6348 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
6349 Delta 88 SVr3.2 is to include this information; other 88k configurations
6350 omit this information by default.
6352 @item -mocs-frame-position
6353 @opindex mocs-frame-position
6354 @cindex register positions in frame (88k)
6355 When emitting COFF debugging information for automatic variables and
6356 parameters stored on the stack, use the offset from the canonical frame
6357 address, which is the stack pointer (register 31) on entry to the
6358 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
6359 @option{-mocs-frame-position}; other 88k configurations have the default
6360 @option{-mno-ocs-frame-position}.
6362 @item -mno-ocs-frame-position
6363 @opindex mno-ocs-frame-position
6364 @cindex register positions in frame (88k)
6365 When emitting COFF debugging information for automatic variables and
6366 parameters stored on the stack, use the offset from the frame pointer
6367 register (register 30). When this option is in effect, the frame
6368 pointer is not eliminated when debugging information is selected by the
6371 @item -moptimize-arg-area
6372 @opindex moptimize-arg-area
6373 @cindex arguments in frame (88k)
6374 Save space by reorganizing the stack frame. This option generates code
6375 that does not agree with the 88open specifications, but uses less
6378 @itemx -mno-optimize-arg-area
6379 @opindex mno-optimize-arg-area
6380 Do not reorganize the stack frame to save space. This is the default.
6381 The generated conforms to the specification, but uses more memory.
6383 @item -mshort-data-@var{num}
6384 @opindex mshort-data
6385 @cindex smaller data references (88k)
6386 @cindex r0-relative references (88k)
6387 Generate smaller data references by making them relative to @code{r0},
6388 which allows loading a value using a single instruction (rather than the
6389 usual two). You control which data references are affected by
6390 specifying @var{num} with this option. For example, if you specify
6391 @option{-mshort-data-512}, then the data references affected are those
6392 involving displacements of less than 512 bytes.
6393 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6396 @item -mserialize-volatile
6397 @opindex mserialize-volatile
6398 @itemx -mno-serialize-volatile
6399 @opindex mno-serialize-volatile
6400 @cindex sequential consistency on 88k
6401 Do, or don't, generate code to guarantee sequential consistency
6402 of volatile memory references. By default, consistency is
6405 The order of memory references made by the MC88110 processor does
6406 not always match the order of the instructions requesting those
6407 references. In particular, a load instruction may execute before
6408 a preceding store instruction. Such reordering violates
6409 sequential consistency of volatile memory references, when there
6410 are multiple processors. When consistency must be guaranteed,
6411 GCC generates special instructions, as needed, to force
6412 execution in the proper order.
6414 The MC88100 processor does not reorder memory references and so
6415 always provides sequential consistency. However, by default, GCC
6416 generates the special instructions to guarantee consistency
6417 even when you use @option{-m88100}, so that the code may be run on an
6418 MC88110 processor. If you intend to run your code only on the
6419 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6421 The extra code generated to guarantee consistency may affect the
6422 performance of your application. If you know that you can safely
6423 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6429 @cindex assembler syntax, 88k
6431 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6432 related to System V release 4 (SVr4). This controls the following:
6436 Which variant of the assembler syntax to emit.
6438 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6439 that is used on System V release 4.
6441 @option{-msvr4} makes GCC issue additional declaration directives used in
6445 @option{-msvr4} is the default for the m88k-motorola-sysv4 and
6446 m88k-dg-dgux m88k configurations. @option{-msvr3} is the default for all
6447 other m88k configurations.
6449 @item -mversion-03.00
6450 @opindex mversion-03.00
6451 This option is obsolete, and is ignored.
6452 @c ??? which asm syntax better for GAS? option there too?
6454 @item -mno-check-zero-division
6455 @itemx -mcheck-zero-division
6456 @opindex mno-check-zero-division
6457 @opindex mcheck-zero-division
6458 @cindex zero division on 88k
6459 Do, or don't, generate code to guarantee that integer division by
6460 zero will be detected. By default, detection is guaranteed.
6462 Some models of the MC88100 processor fail to trap upon integer
6463 division by zero under certain conditions. By default, when
6464 compiling code that might be run on such a processor, GCC
6465 generates code that explicitly checks for zero-valued divisors
6466 and traps with exception number 503 when one is detected. Use of
6467 @option{-mno-check-zero-division} suppresses such checking for code
6468 generated to run on an MC88100 processor.
6470 GCC assumes that the MC88110 processor correctly detects all instances
6471 of integer division by zero. When @option{-m88110} is specified, no
6472 explicit checks for zero-valued divisors are generated, and both
6473 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6476 @item -muse-div-instruction
6477 @opindex muse-div-instruction
6478 @cindex divide instruction, 88k
6479 Use the div instruction for signed integer division on the
6480 MC88100 processor. By default, the div instruction is not used.
6482 On the MC88100 processor the signed integer division instruction
6483 div) traps to the operating system on a negative operand. The
6484 operating system transparently completes the operation, but at a
6485 large cost in execution time. By default, when compiling code
6486 that might be run on an MC88100 processor, GCC emulates signed
6487 integer division using the unsigned integer division instruction
6488 divu), thereby avoiding the large penalty of a trap to the
6489 operating system. Such emulation has its own, smaller, execution
6490 cost in both time and space. To the extent that your code's
6491 important signed integer division operations are performed on two
6492 nonnegative operands, it may be desirable to use the div
6493 instruction directly.
6495 On the MC88110 processor the div instruction (also known as the
6496 divs instruction) processes negative operands without trapping to
6497 the operating system. When @option{-m88110} is specified,
6498 @option{-muse-div-instruction} is ignored, and the div instruction is used
6499 for signed integer division.
6501 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6502 particular, the behavior of such a division with and without
6503 @option{-muse-div-instruction} may differ.
6505 @item -mtrap-large-shift
6506 @itemx -mhandle-large-shift
6507 @opindex mtrap-large-shift
6508 @opindex mhandle-large-shift
6509 @cindex bit shift overflow (88k)
6510 @cindex large bit shifts (88k)
6511 Include code to detect bit-shifts of more than 31 bits; respectively,
6512 trap such shifts or emit code to handle them properly. By default GCC
6513 makes no special provision for large bit shifts.
6515 @item -mwarn-passed-structs
6516 @opindex mwarn-passed-structs
6517 @cindex structure passing (88k)
6518 Warn when a function passes a struct as an argument or result.
6519 Structure-passing conventions have changed during the evolution of the C
6520 language, and are often the source of portability problems. By default,
6521 GCC issues no such warning.
6524 @c break page here to avoid unsightly interparagraph stretch.
6528 @node RS/6000 and PowerPC Options
6529 @subsection IBM RS/6000 and PowerPC Options
6530 @cindex RS/6000 and PowerPC Options
6531 @cindex IBM RS/6000 and PowerPC Options
6533 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6541 @itemx -mpowerpc-gpopt
6542 @itemx -mno-powerpc-gpopt
6543 @itemx -mpowerpc-gfxopt
6544 @itemx -mno-powerpc-gfxopt
6546 @itemx -mno-powerpc64
6552 @opindex mno-powerpc
6553 @opindex mpowerpc-gpopt
6554 @opindex mno-powerpc-gpopt
6555 @opindex mpowerpc-gfxopt
6556 @opindex mno-powerpc-gfxopt
6558 @opindex mno-powerpc64
6559 GCC supports two related instruction set architectures for the
6560 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6561 instructions supported by the @samp{rios} chip set used in the original
6562 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6563 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6564 the IBM 4xx microprocessors.
6566 Neither architecture is a subset of the other. However there is a
6567 large common subset of instructions supported by both. An MQ
6568 register is included in processors supporting the POWER architecture.
6570 You use these options to specify which instructions are available on the
6571 processor you are using. The default value of these options is
6572 determined when configuring GCC@. Specifying the
6573 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6574 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6575 rather than the options listed above.
6577 The @option{-mpower} option allows GCC to generate instructions that
6578 are found only in the POWER architecture and to use the MQ register.
6579 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6580 to generate instructions that are present in the POWER2 architecture but
6581 not the original POWER architecture.
6583 The @option{-mpowerpc} option allows GCC to generate instructions that
6584 are found only in the 32-bit subset of the PowerPC architecture.
6585 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6586 GCC to use the optional PowerPC architecture instructions in the
6587 General Purpose group, including floating-point square root. Specifying
6588 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6589 use the optional PowerPC architecture instructions in the Graphics
6590 group, including floating-point select.
6592 The @option{-mpowerpc64} option allows GCC to generate the additional
6593 64-bit instructions that are found in the full PowerPC64 architecture
6594 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6595 @option{-mno-powerpc64}.
6597 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6598 will use only the instructions in the common subset of both
6599 architectures plus some special AIX common-mode calls, and will not use
6600 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6601 permits GCC to use any instruction from either architecture and to
6602 allow use of the MQ register; specify this for the Motorola MPC601.
6604 @item -mnew-mnemonics
6605 @itemx -mold-mnemonics
6606 @opindex mnew-mnemonics
6607 @opindex mold-mnemonics
6608 Select which mnemonics to use in the generated assembler code. With
6609 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6610 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6611 assembler mnemonics defined for the POWER architecture. Instructions
6612 defined in only one architecture have only one mnemonic; GCC uses that
6613 mnemonic irrespective of which of these options is specified.
6615 GCC defaults to the mnemonics appropriate for the architecture in
6616 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6617 value of these option. Unless you are building a cross-compiler, you
6618 should normally not specify either @option{-mnew-mnemonics} or
6619 @option{-mold-mnemonics}, but should instead accept the default.
6621 @item -mcpu=@var{cpu_type}
6623 Set architecture type, register usage, choice of mnemonics, and
6624 instruction scheduling parameters for machine type @var{cpu_type}.
6625 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6626 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6627 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6628 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
6629 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
6630 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
6632 @option{-mcpu=common} selects a completely generic processor. Code
6633 generated under this option will run on any POWER or PowerPC processor.
6634 GCC will use only the instructions in the common subset of both
6635 architectures, and will not use the MQ register. GCC assumes a generic
6636 processor model for scheduling purposes.
6638 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
6639 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
6640 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
6641 types, with an appropriate, generic processor model assumed for
6642 scheduling purposes.
6644 The other options specify a specific processor. Code generated under
6645 those options will run best on that processor, and may not run at all on
6648 The @option{-mcpu} options automatically enable or disable other
6649 @option{-m} options as follows:
6653 @option{-mno-power}, @option{-mno-powerc}
6660 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
6675 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6678 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6683 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
6686 @item -mtune=@var{cpu_type}
6688 Set the instruction scheduling parameters for machine type
6689 @var{cpu_type}, but do not set the architecture type, register usage, or
6690 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
6691 values for @var{cpu_type} are used for @option{-mtune} as for
6692 @option{-mcpu}. If both are specified, the code generated will use the
6693 architecture, registers, and mnemonics set by @option{-mcpu}, but the
6694 scheduling parameters set by @option{-mtune}.
6699 @opindex mno-altivec
6700 These switches enable or disable the use of built-in functions that
6701 allow access to the AltiVec instruction set. You may also need to set
6702 @option{-mabi=altivec} to adjust the current ABI with AltiVec ABI
6706 @itemx -mno-fp-in-toc
6707 @itemx -mno-sum-in-toc
6708 @itemx -mminimal-toc
6710 @opindex mno-fp-in-toc
6711 @opindex mno-sum-in-toc
6712 @opindex mminimal-toc
6713 Modify generation of the TOC (Table Of Contents), which is created for
6714 every executable file. The @option{-mfull-toc} option is selected by
6715 default. In that case, GCC will allocate at least one TOC entry for
6716 each unique non-automatic variable reference in your program. GCC
6717 will also place floating-point constants in the TOC@. However, only
6718 16,384 entries are available in the TOC@.
6720 If you receive a linker error message that saying you have overflowed
6721 the available TOC space, you can reduce the amount of TOC space used
6722 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
6723 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
6724 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
6725 generate code to calculate the sum of an address and a constant at
6726 run-time instead of putting that sum into the TOC@. You may specify one
6727 or both of these options. Each causes GCC to produce very slightly
6728 slower and larger code at the expense of conserving TOC space.
6730 If you still run out of space in the TOC even when you specify both of
6731 these options, specify @option{-mminimal-toc} instead. This option causes
6732 GCC to make only one TOC entry for every file. When you specify this
6733 option, GCC will produce code that is slower and larger but which
6734 uses extremely little TOC space. You may wish to use this option
6735 only on files that contain less frequently executed code.
6741 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6742 @code{long} type, and the infrastructure needed to support them.
6743 Specifying @option{-maix64} implies @option{-mpowerpc64} and
6744 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
6745 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
6750 @opindex mno-xl-call
6751 On AIX, pass floating-point arguments to prototyped functions beyond the
6752 register save area (RSA) on the stack in addition to argument FPRs. The
6753 AIX calling convention was extended but not initially documented to
6754 handle an obscure K&R C case of calling a function that takes the
6755 address of its arguments with fewer arguments than declared. AIX XL
6756 compilers access floating point arguments which do not fit in the
6757 RSA from the stack when a subroutine is compiled without
6758 optimization. Because always storing floating-point arguments on the
6759 stack is inefficient and rarely needed, this option is not enabled by
6760 default and only is necessary when calling subroutines compiled by AIX
6761 XL compilers without optimization.
6765 Support @dfn{AIX Threads}. Link an application written to use
6766 @dfn{pthreads} with special libraries and startup code to enable the
6771 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
6772 application written to use message passing with special startup code to
6773 enable the application to run. The system must have PE installed in the
6774 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6775 must be overridden with the @option{-specs=} option to specify the
6776 appropriate directory location. The Parallel Environment does not
6777 support threads, so the @option{-mpe} option and the @option{-mthreads}
6778 option are incompatible.
6782 @opindex msoft-float
6783 @opindex mhard-float
6784 Generate code that does not use (uses) the floating-point register set.
6785 Software floating point emulation is provided if you use the
6786 @option{-msoft-float} option, and pass the option to GCC when linking.
6789 @itemx -mno-multiple
6791 @opindex mno-multiple
6792 Generate code that uses (does not use) the load multiple word
6793 instructions and the store multiple word instructions. These
6794 instructions are generated by default on POWER systems, and not
6795 generated on PowerPC systems. Do not use @option{-mmultiple} on little
6796 endian PowerPC systems, since those instructions do not work when the
6797 processor is in little endian mode. The exceptions are PPC740 and
6798 PPC750 which permit the instructions usage in little endian mode.
6804 Generate code that uses (does not use) the load string instructions
6805 and the store string word instructions to save multiple registers and
6806 do small block moves. These instructions are generated by default on
6807 POWER systems, and not generated on PowerPC systems. Do not use
6808 @option{-mstring} on little endian PowerPC systems, since those
6809 instructions do not work when the processor is in little endian mode.
6810 The exceptions are PPC740 and PPC750 which permit the instructions
6811 usage in little endian mode.
6817 Generate code that uses (does not use) the load or store instructions
6818 that update the base register to the address of the calculated memory
6819 location. These instructions are generated by default. If you use
6820 @option{-mno-update}, there is a small window between the time that the
6821 stack pointer is updated and the address of the previous frame is
6822 stored, which means code that walks the stack frame across interrupts or
6823 signals may get corrupted data.
6826 @itemx -mno-fused-madd
6827 @opindex mfused-madd
6828 @opindex mno-fused-madd
6829 Generate code that uses (does not use) the floating point multiply and
6830 accumulate instructions. These instructions are generated by default if
6831 hardware floating is used.
6833 @item -mno-bit-align
6835 @opindex mno-bit-align
6837 On System V.4 and embedded PowerPC systems do not (do) force structures
6838 and unions that contain bit-fields to be aligned to the base type of the
6841 For example, by default a structure containing nothing but 8
6842 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
6843 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
6844 the structure would be aligned to a 1 byte boundary and be one byte in
6847 @item -mno-strict-align
6848 @itemx -mstrict-align
6849 @opindex mno-strict-align
6850 @opindex mstrict-align
6851 On System V.4 and embedded PowerPC systems do not (do) assume that
6852 unaligned memory references will be handled by the system.
6855 @itemx -mno-relocatable
6856 @opindex mrelocatable
6857 @opindex mno-relocatable
6858 On embedded PowerPC systems generate code that allows (does not allow)
6859 the program to be relocated to a different address at runtime. If you
6860 use @option{-mrelocatable} on any module, all objects linked together must
6861 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
6863 @item -mrelocatable-lib
6864 @itemx -mno-relocatable-lib
6865 @opindex mrelocatable-lib
6866 @opindex mno-relocatable-lib
6867 On embedded PowerPC systems generate code that allows (does not allow)
6868 the program to be relocated to a different address at runtime. Modules
6869 compiled with @option{-mrelocatable-lib} can be linked with either modules
6870 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
6871 with modules compiled with the @option{-mrelocatable} options.
6877 On System V.4 and embedded PowerPC systems do not (do) assume that
6878 register 2 contains a pointer to a global area pointing to the addresses
6879 used in the program.
6882 @itemx -mlittle-endian
6884 @opindex mlittle-endian
6885 On System V.4 and embedded PowerPC systems compile code for the
6886 processor in little endian mode. The @option{-mlittle-endian} option is
6887 the same as @option{-mlittle}.
6892 @opindex mbig-endian
6893 On System V.4 and embedded PowerPC systems compile code for the
6894 processor in big endian mode. The @option{-mbig-endian} option is
6895 the same as @option{-mbig}.
6899 On System V.4 and embedded PowerPC systems compile code using calling
6900 conventions that adheres to the March 1995 draft of the System V
6901 Application Binary Interface, PowerPC processor supplement. This is the
6902 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
6904 @item -mcall-sysv-eabi
6905 @opindex mcall-sysv-eabi
6906 Specify both @option{-mcall-sysv} and @option{-meabi} options.
6908 @item -mcall-sysv-noeabi
6909 @opindex mcall-sysv-noeabi
6910 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
6914 On System V.4 and embedded PowerPC systems compile code using calling
6915 conventions that are similar to those used on AIX@. This is the
6916 default if you configured GCC using @samp{powerpc-*-eabiaix}.
6918 @item -mcall-solaris
6919 @opindex mcall-solaris
6920 On System V.4 and embedded PowerPC systems compile code for the Solaris
6924 @opindex mcall-linux
6925 On System V.4 and embedded PowerPC systems compile code for the
6926 Linux-based GNU system.
6929 @opindex mcall-netbsd
6930 On System V.4 and embedded PowerPC systems compile code for the
6931 NetBSD operating system.
6934 @opindex mabi=altivec
6935 Extend the current ABI with AltiVec ABI extensions. This does not
6936 change the default ABI, instead it adds the AltiVec ABI extensions to
6940 @itemx -mno-prototype
6942 @opindex mno-prototype
6943 On System V.4 and embedded PowerPC systems assume that all calls to
6944 variable argument functions are properly prototyped. Otherwise, the
6945 compiler must insert an instruction before every non prototyped call to
6946 set or clear bit 6 of the condition code register (@var{CR}) to
6947 indicate whether floating point values were passed in the floating point
6948 registers in case the function takes a variable arguments. With
6949 @option{-mprototype}, only calls to prototyped variable argument functions
6950 will set or clear the bit.
6954 On embedded PowerPC systems, assume that the startup module is called
6955 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
6956 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
6961 On embedded PowerPC systems, assume that the startup module is called
6962 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6967 On embedded PowerPC systems, assume that the startup module is called
6968 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6972 @opindex myellowknife
6973 On embedded PowerPC systems, assume that the startup module is called
6974 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6979 On System V.4 and embedded PowerPC systems, specify that you are
6980 compiling for a VxWorks system.
6984 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6985 header to indicate that @samp{eabi} extended relocations are used.
6991 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6992 Embedded Applications Binary Interface (eabi) which is a set of
6993 modifications to the System V.4 specifications. Selecting @option{-meabi}
6994 means that the stack is aligned to an 8 byte boundary, a function
6995 @code{__eabi} is called to from @code{main} to set up the eabi
6996 environment, and the @option{-msdata} option can use both @code{r2} and
6997 @code{r13} to point to two separate small data areas. Selecting
6998 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6999 do not call an initialization function from @code{main}, and the
7000 @option{-msdata} option will only use @code{r13} to point to a single
7001 small data area. The @option{-meabi} option is on by default if you
7002 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
7005 @opindex msdata=eabi
7006 On System V.4 and embedded PowerPC systems, put small initialized
7007 @code{const} global and static data in the @samp{.sdata2} section, which
7008 is pointed to by register @code{r2}. Put small initialized
7009 non-@code{const} global and static data in the @samp{.sdata} section,
7010 which is pointed to by register @code{r13}. Put small uninitialized
7011 global and static data in the @samp{.sbss} section, which is adjacent to
7012 the @samp{.sdata} section. The @option{-msdata=eabi} option is
7013 incompatible with the @option{-mrelocatable} option. The
7014 @option{-msdata=eabi} option also sets the @option{-memb} option.
7017 @opindex msdata=sysv
7018 On System V.4 and embedded PowerPC systems, put small global and static
7019 data in the @samp{.sdata} section, which is pointed to by register
7020 @code{r13}. Put small uninitialized global and static data in the
7021 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
7022 The @option{-msdata=sysv} option is incompatible with the
7023 @option{-mrelocatable} option.
7025 @item -msdata=default
7027 @opindex msdata=default
7029 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
7030 compile code the same as @option{-msdata=eabi}, otherwise compile code the
7031 same as @option{-msdata=sysv}.
7034 @opindex msdata-data
7035 On System V.4 and embedded PowerPC systems, put small global and static
7036 data in the @samp{.sdata} section. Put small uninitialized global and
7037 static data in the @samp{.sbss} section. Do not use register @code{r13}
7038 to address small data however. This is the default behavior unless
7039 other @option{-msdata} options are used.
7043 @opindex msdata=none
7045 On embedded PowerPC systems, put all initialized global and static data
7046 in the @samp{.data} section, and all uninitialized data in the
7047 @samp{.bss} section.
7051 @cindex smaller data references (PowerPC)
7052 @cindex .sdata/.sdata2 references (PowerPC)
7053 On embedded PowerPC systems, put global and static items less than or
7054 equal to @var{num} bytes into the small data or bss sections instead of
7055 the normal data or bss section. By default, @var{num} is 8. The
7056 @option{-G @var{num}} switch is also passed to the linker.
7057 All modules should be compiled with the same @option{-G @var{num}} value.
7060 @itemx -mno-regnames
7062 @opindex mno-regnames
7063 On System V.4 and embedded PowerPC systems do (do not) emit register
7064 names in the assembly language output using symbolic forms.
7069 @subsection IBM RT Options
7071 @cindex IBM RT options
7073 These @samp{-m} options are defined for the IBM RT PC:
7077 @opindex min-line-mul
7078 Use an in-line code sequence for integer multiplies. This is the
7081 @item -mcall-lib-mul
7082 @opindex mcall-lib-mul
7083 Call @code{lmul$$} for integer multiples.
7085 @item -mfull-fp-blocks
7086 @opindex mfull-fp-blocks
7087 Generate full-size floating point data blocks, including the minimum
7088 amount of scratch space recommended by IBM@. This is the default.
7090 @item -mminimum-fp-blocks
7091 @opindex mminimum-fp-blocks
7092 Do not include extra scratch space in floating point data blocks. This
7093 results in smaller code, but slower execution, since scratch space must
7094 be allocated dynamically.
7096 @cindex @file{varargs.h} and RT PC
7097 @cindex @file{stdarg.h} and RT PC
7098 @item -mfp-arg-in-fpregs
7099 @opindex mfp-arg-in-fpregs
7100 Use a calling sequence incompatible with the IBM calling convention in
7101 which floating point arguments are passed in floating point registers.
7102 Note that @code{varargs.h} and @code{stdarg.h} will not work with
7103 floating point operands if this option is specified.
7105 @item -mfp-arg-in-gregs
7106 @opindex mfp-arg-in-gregs
7107 Use the normal calling convention for floating point arguments. This is
7110 @item -mhc-struct-return
7111 @opindex mhc-struct-return
7112 Return structures of more than one word in memory, rather than in a
7113 register. This provides compatibility with the MetaWare HighC (hc)
7114 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7115 with the Portable C Compiler (pcc).
7117 @item -mnohc-struct-return
7118 @opindex mnohc-struct-return
7119 Return some structures of more than one word in registers, when
7120 convenient. This is the default. For compatibility with the
7121 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7122 option @option{-mhc-struct-return}.
7126 @subsection MIPS Options
7127 @cindex MIPS options
7129 These @samp{-m} options are defined for the MIPS family of computers:
7133 @item -march=@var{cpu-type}
7135 Assume the defaults for the machine type @var{cpu-type} when generating
7136 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7137 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7138 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7139 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7140 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7141 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc.
7143 @item -mtune=@var{cpu-type}
7145 Assume the defaults for the machine type @var{cpu-type} when scheduling
7146 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7147 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7148 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7149 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7150 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7151 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
7152 @var{cpu-type} will schedule things appropriately for that particular
7153 chip, the compiler will not generate any code that does not meet level 1
7154 of the MIPS ISA (instruction set architecture) without a @option{-mipsX}
7155 or @option{-mabi} switch being used.
7157 @item -mcpu=@var{cpu-type}
7159 This is identical to specifying both @option{-march} and @option{-mtune}.
7163 Issue instructions from level 1 of the MIPS ISA@. This is the default.
7164 @samp{r3000} is the default @var{cpu-type} at this ISA level.
7168 Issue instructions from level 2 of the MIPS ISA (branch likely, square
7169 root instructions). @samp{r6000} is the default @var{cpu-type} at this
7174 Issue instructions from level 3 of the MIPS ISA (64-bit instructions).
7175 @samp{r4000} is the default @var{cpu-type} at this ISA level.
7179 Issue instructions from level 4 of the MIPS ISA (conditional move,
7180 prefetch, enhanced FPU instructions). @samp{r8000} is the default
7181 @var{cpu-type} at this ISA level.
7185 Assume that 32 32-bit floating point registers are available. This is
7190 Assume that 32 64-bit floating point registers are available. This is
7191 the default when the @option{-mips3} option is used.
7194 @itemx -mno-fused-madd
7195 @opindex mfused-madd
7196 @opindex mno-fused-madd
7197 Generate code that uses (does not use) the floating point multiply and
7198 accumulate instructions, when they are available. These instructions
7199 are generated by default if they are available, but this may be
7200 undesirable if the extra precision causes problems or on certain chips
7201 in the mode where denormals are rounded to zero where denormals
7202 generated by multiply and accumulate instructions cause exceptions
7207 Assume that 32 32-bit general purpose registers are available. This is
7212 Assume that 32 64-bit general purpose registers are available. This is
7213 the default when the @option{-mips3} option is used.
7217 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7218 explanation of the default, and the width of pointers.
7222 Force long types to be 64 bits wide. See @option{-mlong32} for an
7223 explanation of the default, and the width of pointers.
7227 Force long, int, and pointer types to be 32 bits wide.
7229 If none of @option{-mlong32}, @option{-mlong64}, or @option{-mint64} are set,
7230 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
7231 For @option{-mabi=32}, and @option{-mabi=n32}, ints and longs are 32 bits
7232 wide. For @option{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
7233 For @option{-mabi=eabi} and either @option{-mips1} or @option{-mips2}, ints
7234 and longs are 32 bits wide. For @option{-mabi=eabi} and higher ISAs, ints
7235 are 32 bits, and longs are 64 bits wide. The width of pointer types is
7236 the smaller of the width of longs or the width of general purpose
7237 registers (which in turn depends on the ISA)@.
7249 Generate code for the indicated ABI@. The default instruction level is
7250 @option{-mips1} for @samp{32}, @option{-mips3} for @samp{n32}, and
7251 @option{-mips4} otherwise. Conversely, with @option{-mips1} or
7252 @option{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
7257 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7258 add normal debug information. This is the default for all
7259 platforms except for the OSF/1 reference platform, using the OSF/rose
7260 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7261 switches are used, the @file{mips-tfile} program will encapsulate the
7262 stabs within MIPS ECOFF@.
7266 Generate code for the GNU assembler. This is the default on the OSF/1
7267 reference platform, using the OSF/rose object format. Also, this is
7268 the default if the configure option @option{--with-gnu-as} is used.
7270 @item -msplit-addresses
7271 @itemx -mno-split-addresses
7272 @opindex msplit-addresses
7273 @opindex mno-split-addresses
7274 Generate code to load the high and low parts of address constants separately.
7275 This allows GCC to optimize away redundant loads of the high order
7276 bits of addresses. This optimization requires GNU as and GNU ld.
7277 This optimization is enabled by default for some embedded targets where
7278 GNU as and GNU ld are standard.
7284 The @option{-mrnames} switch says to output code using the MIPS software
7285 names for the registers, instead of the hardware names (ie, @var{a0}
7286 instead of @var{$4}). The only known assembler that supports this option
7287 is the Algorithmics assembler.
7293 The @option{-mgpopt} switch says to write all of the data declarations
7294 before the instructions in the text section, this allows the MIPS
7295 assembler to generate one word memory references instead of using two
7296 words for short global or static data items. This is on by default if
7297 optimization is selected.
7303 For each non-inline function processed, the @option{-mstats} switch
7304 causes the compiler to emit one line to the standard error file to
7305 print statistics about the program (number of registers saved, stack
7312 The @option{-mmemcpy} switch makes all block moves call the appropriate
7313 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7314 generating inline code.
7317 @itemx -mno-mips-tfile
7318 @opindex mmips-tfile
7319 @opindex mno-mips-tfile
7320 The @option{-mno-mips-tfile} switch causes the compiler not
7321 postprocess the object file with the @file{mips-tfile} program,
7322 after the MIPS assembler has generated it to add debug support. If
7323 @file{mips-tfile} is not run, then no local variables will be
7324 available to the debugger. In addition, @file{stage2} and
7325 @file{stage3} objects will have the temporary file names passed to the
7326 assembler embedded in the object file, which means the objects will
7327 not compare the same. The @option{-mno-mips-tfile} switch should only
7328 be used when there are bugs in the @file{mips-tfile} program that
7329 prevents compilation.
7332 @opindex msoft-float
7333 Generate output containing library calls for floating point.
7334 @strong{Warning:} the requisite libraries are not part of GCC@.
7335 Normally the facilities of the machine's usual C compiler are used, but
7336 this can't be done directly in cross-compilation. You must make your
7337 own arrangements to provide suitable library functions for
7341 @opindex mhard-float
7342 Generate output containing floating point instructions. This is the
7343 default if you use the unmodified sources.
7346 @itemx -mno-abicalls
7348 @opindex mno-abicalls
7349 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7350 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7351 position independent code.
7354 @itemx -mno-long-calls
7355 @opindex mlong-calls
7356 @opindex mno-long-calls
7357 Do all calls with the @samp{JALR} instruction, which requires
7358 loading up a function's address into a register before the call.
7359 You need to use this switch, if you call outside of the current
7360 512 megabyte segment to functions that are not through pointers.
7363 @itemx -mno-half-pic
7365 @opindex mno-half-pic
7366 Put pointers to extern references into the data section and load them
7367 up, rather than put the references in the text section.
7369 @item -membedded-pic
7370 @itemx -mno-embedded-pic
7371 @opindex membedded-pic
7372 @opindex mno-embedded-pic
7373 Generate PIC code suitable for some embedded systems. All calls are
7374 made using PC relative address, and all data is addressed using the $gp
7375 register. No more than 65536 bytes of global data may be used. This
7376 requires GNU as and GNU ld which do most of the work. This currently
7377 only works on targets which use ECOFF; it does not work with ELF@.
7379 @item -membedded-data
7380 @itemx -mno-embedded-data
7381 @opindex membedded-data
7382 @opindex mno-embedded-data
7383 Allocate variables to the read-only data section first if possible, then
7384 next in the small data section if possible, otherwise in data. This gives
7385 slightly slower code than the default, but reduces the amount of RAM required
7386 when executing, and thus may be preferred for some embedded systems.
7388 @item -muninit-const-in-rodata
7389 @itemx -mno-uninit-const-in-rodata
7390 @opindex muninit-const-in-rodata
7391 @opindex mno-uninit-const-in-rodata
7392 When used together with @option{-membedded-data}, it will always store uninitialized
7393 const variables in the read-only data section.
7395 @item -msingle-float
7396 @itemx -mdouble-float
7397 @opindex msingle-float
7398 @opindex mdouble-float
7399 The @option{-msingle-float} switch tells gcc to assume that the floating
7400 point coprocessor only supports single precision operations, as on the
7401 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
7402 double precision operations. This is the default.
7408 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7409 as on the @samp{r4650} chip.
7413 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
7414 @option{-mcpu=r4650}.
7420 Enable 16-bit instructions.
7424 Use the entry and exit pseudo ops. This option can only be used with
7429 Compile code for the processor in little endian mode.
7430 The requisite libraries are assumed to exist.
7434 Compile code for the processor in big endian mode.
7435 The requisite libraries are assumed to exist.
7439 @cindex smaller data references (MIPS)
7440 @cindex gp-relative references (MIPS)
7441 Put global and static items less than or equal to @var{num} bytes into
7442 the small data or bss sections instead of the normal data or bss
7443 section. This allows the assembler to emit one word memory reference
7444 instructions based on the global pointer (@var{gp} or @var{$28}),
7445 instead of the normal two words used. By default, @var{num} is 8 when
7446 the MIPS assembler is used, and 0 when the GNU assembler is used. The
7447 @option{-G @var{num}} switch is also passed to the assembler and linker.
7448 All modules should be compiled with the same @option{-G @var{num}}
7453 Tell the MIPS assembler to not run its preprocessor over user
7454 assembler files (with a @samp{.s} suffix) when assembling them.
7458 Pass an option to gas which will cause nops to be inserted if
7459 the read of the destination register of an mfhi or mflo instruction
7460 occurs in the following two instructions.
7464 Do not include the default crt0.
7466 @item -mflush-func=@var{func}
7467 @itemx -mno-flush-func
7468 @opindex mflush-func
7469 Specifies the function to call to flush the I and D caches, or to not
7470 call any such function. If called, the function must take the same
7471 arguments as the common @code{_flush_func()}, that is, the address of the
7472 memory range for which the cache is being flushed, the size of the
7473 memory range, and the number 3 (to flush both caches). The default
7474 depends on the target gcc was configured for, but commonly is either
7475 @samp{_flush_func} or @samp{__cpu_flush}.
7479 These options are defined by the macro
7480 @code{TARGET_SWITCHES} in the machine description. The default for the
7481 options is also defined by that macro, which enables you to change the
7485 @node i386 and x86-64 Options
7486 @subsection Intel 386 and AMD x86-64 Options
7487 @cindex i386 Options
7488 @cindex x86-64 Options
7489 @cindex Intel 386 Options
7490 @cindex AMD x86-64 Options
7492 These @samp{-m} options are defined for the i386 and x86-64 family of
7496 @item -mcpu=@var{cpu-type}
7498 Assume the defaults for the machine type @var{cpu-type} when scheduling
7499 instructions. The choices for @var{cpu-type} are @samp{i386},
7500 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
7501 @samp{pentiumpro}, @samp{pentium4}, @samp{k6}, and @samp{athlon}
7503 While picking a specific @var{cpu-type} will schedule things appropriately
7504 for that particular chip, the compiler will not generate any code that
7505 does not run on the i386 without the @option{-march=@var{cpu-type}} option
7506 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
7507 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
7508 AMD chips as opposed to the Intel ones.
7510 @item -march=@var{cpu-type}
7512 Generate instructions for the machine type @var{cpu-type}. The choices
7513 for @var{cpu-type} are the same as for @option{-mcpu}. Moreover,
7514 specifying @option{-march=@var{cpu-type}} implies @option{-mcpu=@var{cpu-type}}.
7523 @opindex mpentiumpro
7524 These options are synonyms for @option{-mcpu=i386}, @option{-mcpu=i486},
7525 @option{-mcpu=pentium}, and @option{-mcpu=pentiumpro} respectively.
7526 These synonyms are deprecated.
7528 @item -mintel-syntax
7529 @opindex mintel-syntax
7530 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
7535 @opindex mno-ieee-fp
7536 Control whether or not the compiler uses IEEE floating point
7537 comparisons. These handle correctly the case where the result of a
7538 comparison is unordered.
7541 @opindex msoft-float
7542 Generate output containing library calls for floating point.
7543 @strong{Warning:} the requisite libraries are not part of GCC@.
7544 Normally the facilities of the machine's usual C compiler are used, but
7545 this can't be done directly in cross-compilation. You must make your
7546 own arrangements to provide suitable library functions for
7549 On machines where a function returns floating point results in the 80387
7550 register stack, some floating point opcodes may be emitted even if
7551 @option{-msoft-float} is used.
7553 @item -mno-fp-ret-in-387
7554 @opindex mno-fp-ret-in-387
7555 Do not use the FPU registers for return values of functions.
7557 The usual calling convention has functions return values of types
7558 @code{float} and @code{double} in an FPU register, even if there
7559 is no FPU@. The idea is that the operating system should emulate
7562 The option @option{-mno-fp-ret-in-387} causes such values to be returned
7563 in ordinary CPU registers instead.
7565 @item -mno-fancy-math-387
7566 @opindex mno-fancy-math-387
7567 Some 387 emulators do not support the @code{sin}, @code{cos} and
7568 @code{sqrt} instructions for the 387. Specify this option to avoid
7569 generating those instructions. This option is the default on FreeBSD@.
7570 As of revision 2.6.1, these instructions are not generated unless you
7571 also use the @option{-funsafe-math-optimizations} switch.
7573 @item -malign-double
7574 @itemx -mno-align-double
7575 @opindex malign-double
7576 @opindex mno-align-double
7577 Control whether GCC aligns @code{double}, @code{long double}, and
7578 @code{long long} variables on a two word boundary or a one word
7579 boundary. Aligning @code{double} variables on a two word boundary will
7580 produce code that runs somewhat faster on a @samp{Pentium} at the
7581 expense of more memory.
7583 @item -m128bit-long-double
7584 @opindex m128bit-long-double
7585 Control the size of @code{long double} type. i386 application binary interface
7586 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
7587 prefer @code{long double} aligned to 8 or 16 byte boundary. This is
7588 impossible to reach with 12 byte long doubles in the array accesses.
7590 @strong{Warning:} if you use the @option{-m128bit-long-double} switch, the
7591 structures and arrays containing @code{long double} will change their size as
7592 well as function calling convention for function taking @code{long double}
7595 @item -m96bit-long-double
7596 @opindex m96bit-long-double
7597 Set the size of @code{long double} to 96 bits as required by the i386
7598 application binary interface. This is the default.
7601 @itemx -mno-svr3-shlib
7602 @opindex msvr3-shlib
7603 @opindex mno-svr3-shlib
7604 Control whether GCC places uninitialized local variables into the
7605 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
7606 into @code{bss}. These options are meaningful only on System V Release 3.
7610 Use a different function-calling convention, in which functions that
7611 take a fixed number of arguments return with the @code{ret} @var{num}
7612 instruction, which pops their arguments while returning. This saves one
7613 instruction in the caller since there is no need to pop the arguments
7616 You can specify that an individual function is called with this calling
7617 sequence with the function attribute @samp{stdcall}. You can also
7618 override the @option{-mrtd} option by using the function attribute
7619 @samp{cdecl}. @xref{Function Attributes}.
7621 @strong{Warning:} this calling convention is incompatible with the one
7622 normally used on Unix, so you cannot use it if you need to call
7623 libraries compiled with the Unix compiler.
7625 Also, you must provide function prototypes for all functions that
7626 take variable numbers of arguments (including @code{printf});
7627 otherwise incorrect code will be generated for calls to those
7630 In addition, seriously incorrect code will result if you call a
7631 function with too many arguments. (Normally, extra arguments are
7632 harmlessly ignored.)
7634 @item -mregparm=@var{num}
7636 Control how many registers are used to pass integer arguments. By
7637 default, no registers are used to pass arguments, and at most 3
7638 registers can be used. You can control this behavior for a specific
7639 function by using the function attribute @samp{regparm}.
7640 @xref{Function Attributes}.
7642 @strong{Warning:} if you use this switch, and
7643 @var{num} is nonzero, then you must build all modules with the same
7644 value, including any libraries. This includes the system libraries and
7647 @item -mpreferred-stack-boundary=@var{num}
7648 @opindex mpreferred-stack-boundary
7649 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
7650 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
7651 the default is 4 (16 bytes or 128 bits), except when optimizing for code
7652 size (@option{-Os}), in which case the default is the minimum correct
7653 alignment (4 bytes for x86, and 8 bytes for x86-64).
7655 On Pentium and PentiumPro, @code{double} and @code{long double} values
7656 should be aligned to an 8 byte boundary (see @option{-malign-double}) or
7657 suffer significant run time performance penalties. On Pentium III, the
7658 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
7659 penalties if it is not 16 byte aligned.
7661 To ensure proper alignment of this values on the stack, the stack boundary
7662 must be as aligned as that required by any value stored on the stack.
7663 Further, every function must be generated such that it keeps the stack
7664 aligned. Thus calling a function compiled with a higher preferred
7665 stack boundary from a function compiled with a lower preferred stack
7666 boundary will most likely misalign the stack. It is recommended that
7667 libraries that use callbacks always use the default setting.
7669 This extra alignment does consume extra stack space, and generally
7670 increases code size. Code that is sensitive to stack space usage, such
7671 as embedded systems and operating system kernels, may want to reduce the
7672 preferred alignment to @option{-mpreferred-stack-boundary=2}.
7686 These switches enable or disable the use of built-in functions that allow
7687 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
7689 The following machine modes are available for use with MMX built-in functions
7690 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32 bit integers,
7691 @code{V4HI} for a vector of four 16 bit integers, and @code{V8QI} for a
7692 vector of eight 8 bit integers. Some of the built-in functions operate on
7693 MMX registers as a whole 64 bit entity, these use @code{DI} as their mode.
7695 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
7696 of two 32 bit floating point values.
7698 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32 bit
7699 floating point values. Some instructions use a vector of four 32 bit
7700 integers, these use @code{V4SI}. Finally, some instructions operate on an
7701 entire vector register, interpreting it as a 128 bit integer, these use mode
7704 The following built-in functions are made available by @option{-mmmx}:
7706 @item v8qi __builtin_ia32_paddb (v8qi, v8qi)
7707 Generates the @code{paddb} machine instruction.
7708 @item v4hi __builtin_ia32_paddw (v4hi, v4hi)
7709 Generates the @code{paddw} machine instruction.
7710 @item v2si __builtin_ia32_paddd (v2si, v2si)
7711 Generates the @code{paddd} machine instruction.
7712 @item v8qi __builtin_ia32_psubb (v8qi, v8qi)
7713 Generates the @code{psubb} machine instruction.
7714 @item v4hi __builtin_ia32_psubw (v4hi, v4hi)
7715 Generates the @code{psubw} machine instruction.
7716 @item v2si __builtin_ia32_psubd (v2si, v2si)
7717 Generates the @code{psubd} machine instruction.
7719 @item v8qi __builtin_ia32_paddsb (v8qi, v8qi)
7720 Generates the @code{paddsb} machine instruction.
7721 @item v4hi __builtin_ia32_paddsw (v4hi, v4hi)
7722 Generates the @code{paddsw} machine instruction.
7723 @item v8qi __builtin_ia32_psubsb (v8qi, v8qi)
7724 Generates the @code{psubsb} machine instruction.
7725 @item v4hi __builtin_ia32_psubsw (v4hi, v4hi)
7726 Generates the @code{psubsw} machine instruction.
7728 @item v8qi __builtin_ia32_paddusb (v8qi, v8qi)
7729 Generates the @code{paddusb} machine instruction.
7730 @item v4hi __builtin_ia32_paddusw (v4hi, v4hi)
7731 Generates the @code{paddusw} machine instruction.
7732 @item v8qi __builtin_ia32_psubusb (v8qi, v8qi)
7733 Generates the @code{psubusb} machine instruction.
7734 @item v4hi __builtin_ia32_psubusw (v4hi, v4hi)
7735 Generates the @code{psubusw} machine instruction.
7737 @item v4hi __builtin_ia32_pmullw (v4hi, v4hi)
7738 Generates the @code{pmullw} machine instruction.
7739 @item v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
7740 Generates the @code{pmulhw} machine instruction.
7742 @item di __builtin_ia32_pand (di, di)
7743 Generates the @code{pand} machine instruction.
7744 @item di __builtin_ia32_pandn (di,di)
7745 Generates the @code{pandn} machine instruction.
7746 @item di __builtin_ia32_por (di, di)
7747 Generates the @code{por} machine instruction.
7748 @item di __builtin_ia32_pxor (di, di)
7749 Generates the @code{pxor} machine instruction.
7751 @item v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
7752 Generates the @code{pcmpeqb} machine instruction.
7753 @item v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
7754 Generates the @code{pcmpeqw} machine instruction.
7755 @item v2si __builtin_ia32_pcmpeqd (v2si, v2si)
7756 Generates the @code{pcmpeqd} machine instruction.
7757 @item v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
7758 Generates the @code{pcmpgtb} machine instruction.
7759 @item v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
7760 Generates the @code{pcmpgtw} machine instruction.
7761 @item v2si __builtin_ia32_pcmpgtd (v2si, v2si)
7762 Generates the @code{pcmpgtd} machine instruction.
7764 @item v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
7765 Generates the @code{punpckhbw} machine instruction.
7766 @item v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
7767 Generates the @code{punpckhwd} machine instruction.
7768 @item v2si __builtin_ia32_punpckhdq (v2si, v2si)
7769 Generates the @code{punpckhdq} machine instruction.
7770 @item v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
7771 Generates the @code{punpcklbw} machine instruction.
7772 @item v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
7773 Generates the @code{punpcklwd} machine instruction.
7774 @item v2si __builtin_ia32_punpckldq (v2si, v2si)
7775 Generates the @code{punpckldq} machine instruction.
7777 @item v8qi __builtin_ia32_packsswb (v4hi, v4hi)
7778 Generates the @code{packsswb} machine instruction.
7779 @item v4hi __builtin_ia32_packssdw (v2si, v2si)
7780 Generates the @code{packssdw} machine instruction.
7781 @item v8qi __builtin_ia32_packuswb (v4hi, v4hi)
7782 Generates the @code{packuswb} machine instruction.
7786 The following built-in functions are made available either with @option{-msse}, or
7787 with a combination of @option{-m3dnow} and @option{-march=athlon}.
7790 @item v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
7791 Generates the @code{pmulhuw} machine instruction.
7793 @item v8qi __builtin_ia32_pavgb (v8qi, v8qi)
7794 Generates the @code{pavgb} machine instruction.
7795 @item v4hi __builtin_ia32_pavgw (v4hi, v4hi)
7796 Generates the @code{pavgw} machine instruction.
7797 @item v4hi __builtin_ia32_psadbw (v8qi, v8qi)
7798 Generates the @code{psadbw} machine instruction.
7800 @item v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
7801 Generates the @code{pmaxub} machine instruction.
7802 @item v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
7803 Generates the @code{pmaxsw} machine instruction.
7804 @item v8qi __builtin_ia32_pminub (v8qi, v8qi)
7805 Generates the @code{pminub} machine instruction.
7806 @item v4hi __builtin_ia32_pminsw (v4hi, v4hi)
7807 Generates the @code{pminsw} machine instruction.
7809 @item int __builtin_ia32_pextrw (v4hi, int)
7810 Generates the @code{pextrw} machine instruction.
7811 @item v4hi __builtin_ia32_pinsrw (v4hi, int, int)
7812 Generates the @code{pinsrw} machine instruction.
7814 @item int __builtin_ia32_pmovmskb (v8qi)
7815 Generates the @code{pmovmskb} machine instruction.
7816 @item void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
7817 Generates the @code{maskmovq} machine instruction.
7818 @item void __builtin_ia32_movntq (di *, di)
7819 Generates the @code{movntq} machine instruction.
7820 @item void __builtin_ia32_sfence (void)
7821 Generates the @code{sfence} machine instruction.
7822 @item void __builtin_ia32_prefetch (char *, int selector)
7823 Generates a prefetch machine instruction, depending on the value of
7824 selector. If @code{selector} is 0, it generates @code{prefetchnta}; for
7825 a value of 1, it generates @code{prefetcht0}; for a value of 2, it generates
7826 @code{prefetcht1}; and for a value of 3 it generates @code{prefetcht2}.
7830 The following built-in functions are available when @option{-msse} is used.
7833 @item int __builtin_ia32_comieq (v4sf, v4sf)
7834 Generates the @code{comiss} machine instruction and performs an equality
7835 comparison. The return value is the truth value of that comparison.
7836 @item int __builtin_ia32_comineq (v4sf, v4sf)
7837 Generates the @code{comiss} machine instruction and performs an inequality
7838 comparison. The return value is the truth value of that comparison.
7839 @item int __builtin_ia32_comilt (v4sf, v4sf)
7840 Generates the @code{comiss} machine instruction and performs a ``less than''
7841 comparison. The return value is the truth value of that comparison.
7842 @item int __builtin_ia32_comile (v4sf, v4sf)
7843 Generates the @code{comiss} machine instruction and performs a ``less or
7844 equal'' comparison. The return value is the truth value of that comparison.
7845 @item int __builtin_ia32_comigt (v4sf, v4sf)
7846 Generates the @code{comiss} machine instruction and performs a ``greater than''
7847 comparison. The return value is the truth value of that comparison.
7848 @item int __builtin_ia32_comige (v4sf, v4sf)
7849 Generates the @code{comiss} machine instruction and performs a ``greater or
7850 equal'' comparison. The return value is the truth value of that comparison.
7852 @item int __builtin_ia32_ucomieq (v4sf, v4sf)
7853 Generates the @code{ucomiss} machine instruction and performs an equality
7854 comparison. The return value is the truth value of that comparison.
7855 @item int __builtin_ia32_ucomineq (v4sf, v4sf)
7856 Generates the @code{ucomiss} machine instruction and performs an inequality
7857 comparison. The return value is the truth value of that comparison.
7858 @item int __builtin_ia32_ucomilt (v4sf, v4sf)
7859 Generates the @code{ucomiss} machine instruction and performs a ``less than''
7860 comparison. The return value is the truth value of that comparison.
7861 @item int __builtin_ia32_ucomile (v4sf, v4sf)
7862 Generates the @code{ucomiss} machine instruction and performs a ``less or
7863 equal'' comparison. The return value is the truth value of that comparison.
7864 @item int __builtin_ia32_ucomigt (v4sf, v4sf)
7865 Generates the @code{ucomiss} machine instruction and performs a ``greater than''
7866 comparison. The return value is the truth value of that comparison.
7867 @item int __builtin_ia32_ucomige (v4sf, v4sf)
7868 Generates the @code{ucomiss} machine instruction and performs a ``greater or
7869 equal'' comparison. The return value is the truth value of that comparison.
7871 @item v4sf __builtin_ia32_addps (v4sf, v4sf)
7872 Generates the @code{addps} machine instruction.
7873 @item v4sf __builtin_ia32_addss (v4sf, v4sf)
7874 Generates the @code{addss} machine instruction.
7875 @item v4sf __builtin_ia32_subps (v4sf, v4sf)
7876 Generates the @code{subps} machine instruction.
7877 @item v4sf __builtin_ia32_subss (v4sf, v4sf)
7878 Generates the @code{subss} machine instruction.
7879 @item v4sf __builtin_ia32_mulps (v4sf, v4sf)
7880 Generates the @code{mulps} machine instruction.
7881 @item v4sf __builtin_ia32_mulss (v4sf, v4sf)
7882 Generates the @code{mulss} machine instruction.
7883 @item v4sf __builtin_ia32_divps (v4sf, v4sf)
7884 Generates the @code{divps} machine instruction.
7885 @item v4sf __builtin_ia32_divss (v4sf, v4sf)
7886 Generates the @code{divss} machine instruction.
7888 @item v4si __builtin_ia32_cmpeqps (v4sf, v4sf)
7889 Generates the @code{cmpeqps} machine instruction.
7890 @item v4si __builtin_ia32_cmplts (v4sf, v4sf)
7891 Generates the @code{cmpltps} machine instruction.
7892 @item v4si __builtin_ia32_cmpleps (v4sf, v4sf)
7893 Generates the @code{cmpleps} machine instruction.
7894 @item v4si __builtin_ia32_cmpgtps (v4sf, v4sf)
7895 Generates the @code{cmpgtps} machine instruction.
7896 @item v4si __builtin_ia32_cmpgeps (v4sf, v4sf)
7897 Generates the @code{cmpgeps} machine instruction.
7898 @item v4si __builtin_ia32_cmpunordps (v4sf, v4sf)
7899 Generates the @code{cmpunodps} machine instruction.
7900 @item v4si __builtin_ia32_cmpneqps (v4sf, v4sf)
7901 Generates the @code{cmpeqps} machine instruction.
7902 @item v4si __builtin_ia32_cmpnltps (v4sf, v4sf)
7903 Generates the @code{cmpltps} machine instruction.
7904 @item v4si __builtin_ia32_cmpnleps (v4sf, v4sf)
7905 Generates the @code{cmpleps} machine instruction.
7906 @item v4si __builtin_ia32_cmpngtps (v4sf, v4sf)
7907 Generates the @code{cmpgtps} machine instruction.
7908 @item v4si __builtin_ia32_cmpngeps (v4sf, v4sf)
7909 Generates the @code{cmpgeps} machine instruction.
7910 @item v4si __builtin_ia32_cmpordps (v4sf, v4sf)
7911 Generates the @code{cmpunodps} machine instruction.
7913 @item v4si __builtin_ia32_cmpeqss (v4sf, v4sf)
7914 Generates the @code{cmpeqss} machine instruction.
7915 @item v4si __builtin_ia32_cmpltss (v4sf, v4sf)
7916 Generates the @code{cmpltss} machine instruction.
7917 @item v4si __builtin_ia32_cmpless (v4sf, v4sf)
7918 Generates the @code{cmpless} machine instruction.
7919 @item v4si __builtin_ia32_cmpgtss (v4sf, v4sf)
7920 Generates the @code{cmpgtss} machine instruction.
7921 @item v4si __builtin_ia32_cmpgess (v4sf, v4sf)
7922 Generates the @code{cmpgess} machine instruction.
7923 @item v4si __builtin_ia32_cmpunordss (v4sf, v4sf)
7924 Generates the @code{cmpunodss} machine instruction.
7925 @item v4si __builtin_ia32_cmpneqss (v4sf, v4sf)
7926 Generates the @code{cmpeqss} machine instruction.
7927 @item v4si __builtin_ia32_cmpnlts (v4sf, v4sf)
7928 Generates the @code{cmpltss} machine instruction.
7929 @item v4si __builtin_ia32_cmpnless (v4sf, v4sf)
7930 Generates the @code{cmpless} machine instruction.
7931 @item v4si __builtin_ia32_cmpngtss (v4sf, v4sf)
7932 Generates the @code{cmpgtss} machine instruction.
7933 @item v4si __builtin_ia32_cmpngess (v4sf, v4sf)
7934 Generates the @code{cmpgess} machine instruction.
7935 @item v4si __builtin_ia32_cmpordss (v4sf, v4sf)
7936 Generates the @code{cmpunodss} machine instruction.
7938 @item v4sf __builtin_ia32_maxps (v4sf, v4sf)
7939 Generates the @code{maxps} machine instruction.
7940 @item v4sf __builtin_ia32_maxsss (v4sf, v4sf)
7941 Generates the @code{maxss} machine instruction.
7942 @item v4sf __builtin_ia32_minps (v4sf, v4sf)
7943 Generates the @code{minps} machine instruction.
7944 @item v4sf __builtin_ia32_minsss (v4sf, v4sf)
7945 Generates the @code{minss} machine instruction.
7947 @item ti __builtin_ia32_andps (ti, ti)
7948 Generates the @code{andps} machine instruction.
7949 @item ti __builtin_ia32_andnps (ti, ti)
7950 Generates the @code{andnps} machine instruction.
7951 @item ti __builtin_ia32_orps (ti, ti)
7952 Generates the @code{orps} machine instruction.
7953 @item ti __builtin_ia32_xorps (ti, ti)
7954 Generates the @code{xorps} machine instruction.
7956 @item v4sf __builtin_ia32_movps (v4sf, v4sf)
7957 Generates the @code{movps} machine instruction.
7958 @item v4sf __builtin_ia32_movhlps (v4sf, v4sf)
7959 Generates the @code{movhlps} machine instruction.
7960 @item v4sf __builtin_ia32_movlhps (v4sf, v4sf)
7961 Generates the @code{movlhps} machine instruction.
7962 @item v4sf __builtin_ia32_unpckhps (v4sf, v4sf)
7963 Generates the @code{unpckhps} machine instruction.
7964 @item v4sf __builtin_ia32_unpcklps (v4sf, v4sf)
7965 Generates the @code{unpcklps} machine instruction.
7967 @item v4sf __builtin_ia32_cvtpi2ps (v4sf, v2si)
7968 Generates the @code{cvtpi2ps} machine instruction.
7969 @item v2si __builtin_ia32_cvtps2pi (v4sf)
7970 Generates the @code{cvtps2pi} machine instruction.
7971 @item v4sf __builtin_ia32_cvtsi2ss (v4sf, int)
7972 Generates the @code{cvtsi2ss} machine instruction.
7973 @item int __builtin_ia32_cvtss2si (v4sf)
7974 Generates the @code{cvtsi2ss} machine instruction.
7975 @item v2si __builtin_ia32_cvttps2pi (v4sf)
7976 Generates the @code{cvttps2pi} machine instruction.
7977 @item int __builtin_ia32_cvttss2si (v4sf)
7978 Generates the @code{cvttsi2ss} machine instruction.
7980 @item v4sf __builtin_ia32_rcpps (v4sf)
7981 Generates the @code{rcpps} machine instruction.
7982 @item v4sf __builtin_ia32_rsqrtps (v4sf)
7983 Generates the @code{rsqrtps} machine instruction.
7984 @item v4sf __builtin_ia32_sqrtps (v4sf)
7985 Generates the @code{sqrtps} machine instruction.
7986 @item v4sf __builtin_ia32_rcpss (v4sf)
7987 Generates the @code{rcpss} machine instruction.
7988 @item v4sf __builtin_ia32_rsqrtss (v4sf)
7989 Generates the @code{rsqrtss} machine instruction.
7990 @item v4sf __builtin_ia32_sqrtss (v4sf)
7991 Generates the @code{sqrtss} machine instruction.
7993 @item v4sf __builtin_ia32_shufps (v4sf, v4sf, int)
7994 Generates the @code{shufps} machine instruction.
7996 @item v4sf __builtin_ia32_loadaps (float *)
7997 Generates the @code{movaps} machine instruction as a load from memory.
7998 @item void __builtin_ia32_storeaps (float *, v4sf)
7999 Generates the @code{movaps} machine instruction as a store to memory.
8000 @item v4sf __builtin_ia32_loadups (float *)
8001 Generates the @code{movups} machine instruction as a load from memory.
8002 @item void __builtin_ia32_storeups (float *, v4sf)
8003 Generates the @code{movups} machine instruction as a store to memory.
8004 @item v4sf __builtin_ia32_loadsss (float *)
8005 Generates the @code{movss} machine instruction as a load from memory.
8006 @item void __builtin_ia32_storess (float *, v4sf)
8007 Generates the @code{movss} machine instruction as a store to memory.
8009 @item v4sf __builtin_ia32_loadhps (v4sf, v2si *)
8010 Generates the @code{movhps} machine instruction as a load from memory.
8011 @item v4sf __builtin_ia32_loadlps (v4sf, v2si *)
8012 Generates the @code{movlps} machine instruction as a load from memory
8013 @item void __builtin_ia32_storehps (v4sf, v2si *)
8014 Generates the @code{movhps} machine instruction as a store to memory.
8015 @item void __builtin_ia32_storelps (v4sf, v2si *)
8016 Generates the @code{movlps} machine instruction as a store to memory.
8018 @item void __builtin_ia32_movntps (float *, v4sf)
8019 Generates the @code{movntps} machine instruction.
8020 @item int __builtin_ia32_movmskps (v4sf)
8021 Generates the @code{movntps} machine instruction.
8023 @item void __builtin_ia32_storeps1 (float *, v4sf)
8024 Generates the @code{movaps} machine instruction as a store to memory.
8025 Before storing, the value is modified with a @code{shufps} instruction
8026 so that the lowest of the four floating point elements is replicated
8027 across the entire vector that is stored.
8028 @item void __builtin_ia32_storerps (float *, v4sf)
8029 Generates the @code{movaps} machine instruction as a store to memory.
8030 Before storing, the value is modified with a @code{shufps} instruction
8031 so that the order of the four floating point elements in the vector is
8033 @item v4sf __builtin_ia32_loadps1 (float *)
8034 Generates a @code{movss} machine instruction to load a floating point
8035 value from memory, and a @code{shufps} instruction to replicate the
8036 loaded value across all four elements of the result vector.
8037 @item v4sf __builtin_ia32_loadrps (float *)
8038 Generates a @code{movaps} machine instruction to load a vector from
8039 memory, and a @code{shufps} instruction to reverse the order of the
8040 four floating point elements in the result vector.
8041 @item v4sf __builtin_ia32_setps (float, float, float, float)
8042 Constructs a vector from four single floating point values. The return
8043 value is equal to the value that would result from storing the four
8044 arguments into consecutive memory locations and then executing a
8045 @code{movaps} to load the vector from memory.
8046 @item v4sf __builtin_ia32_setps1 (float)
8047 Constructs a vector from a single floating point value by replicating
8048 it across all four elements of the result vector.
8052 @itemx -mno-push-args
8054 @opindex mno-push-args
8055 Use PUSH operations to store outgoing parameters. This method is shorter
8056 and usually equally fast as method using SUB/MOV operations and is enabled
8057 by default. In some cases disabling it may improve performance because of
8058 improved scheduling and reduced dependencies.
8060 @item -maccumulate-outgoing-args
8061 @opindex maccumulate-outgoing-args
8062 If enabled, the maximum amount of space required for outgoing arguments will be
8063 computed in the function prologue. This is faster on most modern CPUs
8064 because of reduced dependencies, improved scheduling and reduced stack usage
8065 when preferred stack boundary is not equal to 2. The drawback is a notable
8066 increase in code size. This switch implies @option{-mno-push-args}.
8070 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8071 on thread-safe exception handling must compile and link all code with the
8072 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8073 @option{-D_MT}; when linking, it links in a special thread helper library
8074 @option{-lmingwthrd} which cleans up per thread exception handling data.
8076 @item -mno-align-stringops
8077 @opindex mno-align-stringops
8078 Do not align destination of inlined string operations. This switch reduces
8079 code size and improves performance in case the destination is already aligned,
8080 but gcc don't know about it.
8082 @item -minline-all-stringops
8083 @opindex minline-all-stringops
8084 By default GCC inlines string operations only when destination is known to be
8085 aligned at least to 4 byte boundary. This enables more inlining, increase code
8086 size, but may improve performance of code that depends on fast memcpy, strlen
8087 and memset for short lengths.
8089 @item -momit-leaf-frame-pointer
8090 @opindex momit-leaf-frame-pointer
8091 Don't keep the frame pointer in a register for leaf functions. This
8092 avoids the instructions to save, set up and restore frame pointers and
8093 makes an extra register available in leaf functions. The option
8094 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8095 which might make debugging harder.
8098 These @samp{-m} switches are supported in addition to the above
8099 on AMD x86-64 processors in 64-bit environments.
8106 Generate code for a 32-bit or 64-bit environment.
8107 The 32-bit environment sets int, long and pointer to 32 bits and
8108 generates code that runs on any i386 system.
8109 The 64-bit environment sets int to 32 bits and long and pointer
8110 to 64 bits and generates code for AMD's x86-64 architecture.
8113 @opindex no-red-zone
8114 Do not use a so called red zone for x86-64 code. The red zone is mandated
8115 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8116 stack pointer that will not be modified by signal or interrupt handlers
8117 and therefore can be used for temporary data without adjusting the stack
8118 pointer. The flag @option{-mno-red-zone} disables this red zone.
8122 @subsection HPPA Options
8123 @cindex HPPA Options
8125 These @samp{-m} options are defined for the HPPA family of computers:
8128 @item -march=@var{architecture-type}
8130 Generate code for the specified architecture. The choices for
8131 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8132 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8133 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8134 architecture option for your machine. Code compiled for lower numbered
8135 architectures will run on higher numbered architectures, but not the
8138 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8139 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8143 @itemx -mpa-risc-1-1
8144 @itemx -mpa-risc-2-0
8145 @opindex mpa-risc-1-0
8146 @opindex mpa-risc-1-1
8147 @opindex mpa-risc-2-0
8148 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8151 @opindex mbig-switch
8152 Generate code suitable for big switch tables. Use this option only if
8153 the assembler/linker complain about out of range branches within a switch
8156 @item -mjump-in-delay
8157 @opindex mjump-in-delay
8158 Fill delay slots of function calls with unconditional jump instructions
8159 by modifying the return pointer for the function call to be the target
8160 of the conditional jump.
8162 @item -mdisable-fpregs
8163 @opindex mdisable-fpregs
8164 Prevent floating point registers from being used in any manner. This is
8165 necessary for compiling kernels which perform lazy context switching of
8166 floating point registers. If you use this option and attempt to perform
8167 floating point operations, the compiler will abort.
8169 @item -mdisable-indexing
8170 @opindex mdisable-indexing
8171 Prevent the compiler from using indexing address modes. This avoids some
8172 rather obscure problems when compiling MIG generated code under MACH@.
8174 @item -mno-space-regs
8175 @opindex mno-space-regs
8176 Generate code that assumes the target has no space registers. This allows
8177 GCC to generate faster indirect calls and use unscaled index address modes.
8179 Such code is suitable for level 0 PA systems and kernels.
8181 @item -mfast-indirect-calls
8182 @opindex mfast-indirect-calls
8183 Generate code that assumes calls never cross space boundaries. This
8184 allows GCC to emit code which performs faster indirect calls.
8186 This option will not work in the presence of shared libraries or nested
8189 @item -mlong-load-store
8190 @opindex mlong-load-store
8191 Generate 3-instruction load and store sequences as sometimes required by
8192 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8195 @item -mportable-runtime
8196 @opindex mportable-runtime
8197 Use the portable calling conventions proposed by HP for ELF systems.
8201 Enable the use of assembler directives only GAS understands.
8203 @item -mschedule=@var{cpu-type}
8205 Schedule code according to the constraints for the machine type
8206 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8207 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
8208 @file{/usr/lib/sched.models} on an HP-UX system to determine the
8209 proper scheduling option for your machine.
8212 @opindex mlinker-opt
8213 Enable the optimization pass in the HPUX linker. Note this makes symbolic
8214 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
8215 in which they give bogus error messages when linking some programs.
8218 @opindex msoft-float
8219 Generate output containing library calls for floating point.
8220 @strong{Warning:} the requisite libraries are not available for all HPPA
8221 targets. Normally the facilities of the machine's usual C compiler are
8222 used, but this cannot be done directly in cross-compilation. You must make
8223 your own arrangements to provide suitable library functions for
8224 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8225 does provide software floating point support.
8227 @option{-msoft-float} changes the calling convention in the output file;
8228 therefore, it is only useful if you compile @emph{all} of a program with
8229 this option. In particular, you need to compile @file{libgcc.a}, the
8230 library that comes with GCC, with @option{-msoft-float} in order for
8234 @node Intel 960 Options
8235 @subsection Intel 960 Options
8237 These @samp{-m} options are defined for the Intel 960 implementations:
8240 @item -m@var{cpu-type}
8248 Assume the defaults for the machine type @var{cpu-type} for some of
8249 the other options, including instruction scheduling, floating point
8250 support, and addressing modes. The choices for @var{cpu-type} are
8251 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8252 @samp{sa}, and @samp{sb}.
8259 @opindex msoft-float
8260 The @option{-mnumerics} option indicates that the processor does support
8261 floating-point instructions. The @option{-msoft-float} option indicates
8262 that floating-point support should not be assumed.
8264 @item -mleaf-procedures
8265 @itemx -mno-leaf-procedures
8266 @opindex mleaf-procedures
8267 @opindex mno-leaf-procedures
8268 Do (or do not) attempt to alter leaf procedures to be callable with the
8269 @code{bal} instruction as well as @code{call}. This will result in more
8270 efficient code for explicit calls when the @code{bal} instruction can be
8271 substituted by the assembler or linker, but less efficient code in other
8272 cases, such as calls via function pointers, or using a linker that doesn't
8273 support this optimization.
8276 @itemx -mno-tail-call
8278 @opindex mno-tail-call
8279 Do (or do not) make additional attempts (beyond those of the
8280 machine-independent portions of the compiler) to optimize tail-recursive
8281 calls into branches. You may not want to do this because the detection of
8282 cases where this is not valid is not totally complete. The default is
8283 @option{-mno-tail-call}.
8285 @item -mcomplex-addr
8286 @itemx -mno-complex-addr
8287 @opindex mcomplex-addr
8288 @opindex mno-complex-addr
8289 Assume (or do not assume) that the use of a complex addressing mode is a
8290 win on this implementation of the i960. Complex addressing modes may not
8291 be worthwhile on the K-series, but they definitely are on the C-series.
8292 The default is currently @option{-mcomplex-addr} for all processors except
8296 @itemx -mno-code-align
8297 @opindex mcode-align
8298 @opindex mno-code-align
8299 Align code to 8-byte boundaries for faster fetching (or don't bother).
8300 Currently turned on by default for C-series implementations only.
8303 @item -mclean-linkage
8304 @itemx -mno-clean-linkage
8305 @opindex mclean-linkage
8306 @opindex mno-clean-linkage
8307 These options are not fully implemented.
8311 @itemx -mic2.0-compat
8312 @itemx -mic3.0-compat
8314 @opindex mic2.0-compat
8315 @opindex mic3.0-compat
8316 Enable compatibility with iC960 v2.0 or v3.0.
8320 @opindex masm-compat
8322 Enable compatibility with the iC960 assembler.
8324 @item -mstrict-align
8325 @itemx -mno-strict-align
8326 @opindex mstrict-align
8327 @opindex mno-strict-align
8328 Do not permit (do permit) unaligned accesses.
8332 Enable structure-alignment compatibility with Intel's gcc release version
8333 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8335 @item -mlong-double-64
8336 @opindex mlong-double-64
8337 Implement type @samp{long double} as 64-bit floating point numbers.
8338 Without the option @samp{long double} is implemented by 80-bit
8339 floating point numbers. The only reason we have it because there is
8340 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8341 is only useful for people using soft-float targets. Otherwise, we
8342 should recommend against use of it.
8346 @node DEC Alpha Options
8347 @subsection DEC Alpha Options
8349 These @samp{-m} options are defined for the DEC Alpha implementations:
8352 @item -mno-soft-float
8354 @opindex mno-soft-float
8355 @opindex msoft-float
8356 Use (do not use) the hardware floating-point instructions for
8357 floating-point operations. When @option{-msoft-float} is specified,
8358 functions in @file{libgcc.a} will be used to perform floating-point
8359 operations. Unless they are replaced by routines that emulate the
8360 floating-point operations, or compiled in such a way as to call such
8361 emulations routines, these routines will issue floating-point
8362 operations. If you are compiling for an Alpha without floating-point
8363 operations, you must ensure that the library is built so as not to call
8366 Note that Alpha implementations without floating-point operations are
8367 required to have floating-point registers.
8372 @opindex mno-fp-regs
8373 Generate code that uses (does not use) the floating-point register set.
8374 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8375 register set is not used, floating point operands are passed in integer
8376 registers as if they were integers and floating-point results are passed
8377 in $0 instead of $f0. This is a non-standard calling sequence, so any
8378 function with a floating-point argument or return value called by code
8379 compiled with @option{-mno-fp-regs} must also be compiled with that
8382 A typical use of this option is building a kernel that does not use,
8383 and hence need not save and restore, any floating-point registers.
8387 The Alpha architecture implements floating-point hardware optimized for
8388 maximum performance. It is mostly compliant with the IEEE floating
8389 point standard. However, for full compliance, software assistance is
8390 required. This option generates code fully IEEE compliant code
8391 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8392 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8393 defined during compilation. The resulting code is less efficient but is
8394 able to correctly support denormalized numbers and exceptional IEEE
8395 values such as not-a-number and plus/minus infinity. Other Alpha
8396 compilers call this option @option{-ieee_with_no_inexact}.
8398 @item -mieee-with-inexact
8399 @opindex mieee-with-inexact
8400 This is like @option{-mieee} except the generated code also maintains
8401 the IEEE @var{inexact-flag}. Turning on this option causes the
8402 generated code to implement fully-compliant IEEE math. In addition to
8403 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8404 macro. On some Alpha implementations the resulting code may execute
8405 significantly slower than the code generated by default. Since there is
8406 very little code that depends on the @var{inexact-flag}, you should
8407 normally not specify this option. Other Alpha compilers call this
8408 option @option{-ieee_with_inexact}.
8410 @item -mfp-trap-mode=@var{trap-mode}
8411 @opindex mfp-trap-mode
8412 This option controls what floating-point related traps are enabled.
8413 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8414 The trap mode can be set to one of four values:
8418 This is the default (normal) setting. The only traps that are enabled
8419 are the ones that cannot be disabled in software (e.g., division by zero
8423 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8427 Like @samp{su}, but the instructions are marked to be safe for software
8428 completion (see Alpha architecture manual for details).
8431 Like @samp{su}, but inexact traps are enabled as well.
8434 @item -mfp-rounding-mode=@var{rounding-mode}
8435 @opindex mfp-rounding-mode
8436 Selects the IEEE rounding mode. Other Alpha compilers call this option
8437 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8442 Normal IEEE rounding mode. Floating point numbers are rounded towards
8443 the nearest machine number or towards the even machine number in case
8447 Round towards minus infinity.
8450 Chopped rounding mode. Floating point numbers are rounded towards zero.
8453 Dynamic rounding mode. A field in the floating point control register
8454 (@var{fpcr}, see Alpha architecture reference manual) controls the
8455 rounding mode in effect. The C library initializes this register for
8456 rounding towards plus infinity. Thus, unless your program modifies the
8457 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8460 @item -mtrap-precision=@var{trap-precision}
8461 @opindex mtrap-precision
8462 In the Alpha architecture, floating point traps are imprecise. This
8463 means without software assistance it is impossible to recover from a
8464 floating trap and program execution normally needs to be terminated.
8465 GCC can generate code that can assist operating system trap handlers
8466 in determining the exact location that caused a floating point trap.
8467 Depending on the requirements of an application, different levels of
8468 precisions can be selected:
8472 Program precision. This option is the default and means a trap handler
8473 can only identify which program caused a floating point exception.
8476 Function precision. The trap handler can determine the function that
8477 caused a floating point exception.
8480 Instruction precision. The trap handler can determine the exact
8481 instruction that caused a floating point exception.
8484 Other Alpha compilers provide the equivalent options called
8485 @option{-scope_safe} and @option{-resumption_safe}.
8487 @item -mieee-conformant
8488 @opindex mieee-conformant
8489 This option marks the generated code as IEEE conformant. You must not
8490 use this option unless you also specify @option{-mtrap-precision=i} and either
8491 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8492 is to emit the line @samp{.eflag 48} in the function prologue of the
8493 generated assembly file. Under DEC Unix, this has the effect that
8494 IEEE-conformant math library routines will be linked in.
8496 @item -mbuild-constants
8497 @opindex mbuild-constants
8498 Normally GCC examines a 32- or 64-bit integer constant to
8499 see if it can construct it from smaller constants in two or three
8500 instructions. If it cannot, it will output the constant as a literal and
8501 generate code to load it from the data segment at runtime.
8503 Use this option to require GCC to construct @emph{all} integer constants
8504 using code, even if it takes more instructions (the maximum is six).
8506 You would typically use this option to build a shared library dynamic
8507 loader. Itself a shared library, it must relocate itself in memory
8508 before it can find the variables and constants in its own data segment.
8514 Select whether to generate code to be assembled by the vendor-supplied
8515 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8529 Indicate whether GCC should generate code to use the optional BWX,
8530 CIX, and MAX instruction sets. The default is to use the instruction sets
8531 supported by the CPU type specified via @option{-mcpu=} option or that
8532 of the CPU on which GCC was built if none was specified.
8534 @item -mcpu=@var{cpu_type}
8536 Set the instruction set, register set, and instruction scheduling
8537 parameters for machine type @var{cpu_type}. You can specify either the
8538 @samp{EV} style name or the corresponding chip number. GCC
8539 supports scheduling parameters for the EV4 and EV5 family of processors
8540 and will choose the default values for the instruction set from
8541 the processor you specify. If you do not specify a processor type,
8542 GCC will default to the processor on which the compiler was built.
8544 Supported values for @var{cpu_type} are
8549 Schedules as an EV4 and has no instruction set extensions.
8553 Schedules as an EV5 and has no instruction set extensions.
8557 Schedules as an EV5 and supports the BWX extension.
8562 Schedules as an EV5 and supports the BWX and MAX extensions.
8566 Schedules as an EV5 (until Digital releases the scheduling parameters
8567 for the EV6) and supports the BWX, CIX, and MAX extensions.
8570 @item -mmemory-latency=@var{time}
8571 @opindex mmemory-latency
8572 Sets the latency the scheduler should assume for typical memory
8573 references as seen by the application. This number is highly
8574 dependent on the memory access patterns used by the application
8575 and the size of the external cache on the machine.
8577 Valid options for @var{time} are
8581 A decimal number representing clock cycles.
8587 The compiler contains estimates of the number of clock cycles for
8588 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8589 (also called Dcache, Scache, and Bcache), as well as to main memory.
8590 Note that L3 is only valid for EV5.
8595 @node Clipper Options
8596 @subsection Clipper Options
8598 These @samp{-m} options are defined for the Clipper implementations:
8603 Produce code for a C300 Clipper processor. This is the default.
8607 Produce code for a C400 Clipper processor, i.e.@: use floating point
8611 @node H8/300 Options
8612 @subsection H8/300 Options
8614 These @samp{-m} options are defined for the H8/300 implementations:
8619 Shorten some address references at link time, when possible; uses the
8620 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8621 ld.info, Using ld}, for a fuller description.
8625 Generate code for the H8/300H@.
8629 Generate code for the H8/S@.
8633 Generate code for the H8/S2600. This switch must be used with @option{-ms}.
8637 Make @code{int} data 32 bits by default.
8641 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
8642 The default for the H8/300H and H8/S is to align longs and floats on 4
8644 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8645 This option has no effect on the H8/300.
8649 @subsection SH Options
8651 These @samp{-m} options are defined for the SH implementations:
8656 Generate code for the SH1.
8660 Generate code for the SH2.
8664 Generate code for the SH3.
8668 Generate code for the SH3e.
8672 Generate code for the SH4 without a floating-point unit.
8674 @item -m4-single-only
8675 @opindex m4-single-only
8676 Generate code for the SH4 with a floating-point unit that only
8677 supports single-precision arithmetic.
8681 Generate code for the SH4 assuming the floating-point unit is in
8682 single-precision mode by default.
8686 Generate code for the SH4.
8690 Compile code for the processor in big endian mode.
8694 Compile code for the processor in little endian mode.
8698 Align doubles at 64-bit boundaries. Note that this changes the calling
8699 conventions, and thus some functions from the standard C library will
8700 not work unless you recompile it first with @option{-mdalign}.
8704 Shorten some address references at link time, when possible; uses the
8705 linker option @option{-relax}.
8709 Use 32-bit offsets in @code{switch} tables. The default is to use
8714 Enable the use of the instruction @code{fmovd}.
8718 Comply with the calling conventions defined by Hitachi.
8722 Mark the @code{MAC} register as call-clobbered, even if
8723 @option{-mhitachi} is given.
8727 Increase IEEE-compliance of floating-point code.
8731 Dump instruction size and location in the assembly code.
8735 This option is deprecated. It pads structures to multiple of 4 bytes,
8736 which is incompatible with the SH ABI@.
8740 Optimize for space instead of speed. Implied by @option{-Os}.
8744 When generating position-independent code, emit function calls using
8745 the Global Offset Table instead of the Procedure Linkage Table.
8749 Generate a library function call to invalidate instruction cache
8750 entries, after fixing up a trampoline. This library function call
8751 doesn't assume it can write to the whole memory address space. This
8752 is the default when the target is @code{sh-*-linux*}.
8755 @node System V Options
8756 @subsection Options for System V
8758 These additional options are available on System V Release 4 for
8759 compatibility with other compilers on those systems:
8764 Create a shared object.
8765 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
8769 Identify the versions of each tool used by the compiler, in a
8770 @code{.ident} assembler directive in the output.
8774 Refrain from adding @code{.ident} directives to the output file (this is
8777 @item -YP,@var{dirs}
8779 Search the directories @var{dirs}, and no others, for libraries
8780 specified with @option{-l}.
8784 Look in the directory @var{dir} to find the M4 preprocessor.
8785 The assembler uses this option.
8786 @c This is supposed to go with a -Yd for predefined M4 macro files, but
8787 @c the generic assembler that comes with Solaris takes just -Ym.
8790 @node TMS320C3x/C4x Options
8791 @subsection TMS320C3x/C4x Options
8792 @cindex TMS320C3x/C4x Options
8794 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
8798 @item -mcpu=@var{cpu_type}
8800 Set the instruction set, register set, and instruction scheduling
8801 parameters for machine type @var{cpu_type}. Supported values for
8802 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
8803 @samp{c44}. The default is @samp{c40} to generate code for the
8808 @itemx -msmall-memory
8810 @opindex mbig-memory
8812 @opindex msmall-memory
8814 Generates code for the big or small memory model. The small memory
8815 model assumed that all data fits into one 64K word page. At run-time
8816 the data page (DP) register must be set to point to the 64K page
8817 containing the .bss and .data program sections. The big memory model is
8818 the default and requires reloading of the DP register for every direct
8825 Allow (disallow) allocation of general integer operands into the block
8832 Enable (disable) generation of code using decrement and branch,
8833 DBcond(D), instructions. This is enabled by default for the C4x. To be
8834 on the safe side, this is disabled for the C3x, since the maximum
8835 iteration count on the C3x is @math{2^23 + 1} (but who iterates loops more than
8836 @math{2^23} times on the C3x?). Note that GCC will try to reverse a loop so
8837 that it can utilise the decrement and branch instruction, but will give
8838 up if there is more than one memory reference in the loop. Thus a loop
8839 where the loop counter is decremented can generate slightly more
8840 efficient code, in cases where the RPTB instruction cannot be utilised.
8842 @item -mdp-isr-reload
8844 @opindex mdp-isr-reload
8846 Force the DP register to be saved on entry to an interrupt service
8847 routine (ISR), reloaded to point to the data section, and restored on
8848 exit from the ISR@. This should not be required unless someone has
8849 violated the small memory model by modifying the DP register, say within
8856 For the C3x use the 24-bit MPYI instruction for integer multiplies
8857 instead of a library call to guarantee 32-bit results. Note that if one
8858 of the operands is a constant, then the multiplication will be performed
8859 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
8860 then squaring operations are performed inline instead of a library call.
8863 @itemx -mno-fast-fix
8865 @opindex mno-fast-fix
8866 The C3x/C4x FIX instruction to convert a floating point value to an
8867 integer value chooses the nearest integer less than or equal to the
8868 floating point value rather than to the nearest integer. Thus if the
8869 floating point number is negative, the result will be incorrectly
8870 truncated an additional code is necessary to detect and correct this
8871 case. This option can be used to disable generation of the additional
8872 code required to correct the result.
8878 Enable (disable) generation of repeat block sequences using the RPTB
8879 instruction for zero overhead looping. The RPTB construct is only used
8880 for innermost loops that do not call functions or jump across the loop
8881 boundaries. There is no advantage having nested RPTB loops due to the
8882 overhead required to save and restore the RC, RS, and RE registers.
8883 This is enabled by default with @option{-O2}.
8885 @item -mrpts=@var{count}
8889 Enable (disable) the use of the single instruction repeat instruction
8890 RPTS@. If a repeat block contains a single instruction, and the loop
8891 count can be guaranteed to be less than the value @var{count}, GCC will
8892 emit a RPTS instruction instead of a RPTB@. If no value is specified,
8893 then a RPTS will be emitted even if the loop count cannot be determined
8894 at compile time. Note that the repeated instruction following RPTS does
8895 not have to be reloaded from memory each iteration, thus freeing up the
8896 CPU buses for operands. However, since interrupts are blocked by this
8897 instruction, it is disabled by default.
8899 @item -mloop-unsigned
8900 @itemx -mno-loop-unsigned
8901 @opindex mloop-unsigned
8902 @opindex mno-loop-unsigned
8903 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
8904 is @math{2^31 + 1} since these instructions test if the iteration count is
8905 negative to terminate the loop. If the iteration count is unsigned
8906 there is a possibility than the @math{2^31 + 1} maximum iteration count may be
8907 exceeded. This switch allows an unsigned iteration count.
8911 Try to emit an assembler syntax that the TI assembler (asm30) is happy
8912 with. This also enforces compatibility with the API employed by the TI
8913 C3x C compiler. For example, long doubles are passed as structures
8914 rather than in floating point registers.
8920 Generate code that uses registers (stack) for passing arguments to functions.
8921 By default, arguments are passed in registers where possible rather
8922 than by pushing arguments on to the stack.
8924 @item -mparallel-insns
8925 @itemx -mno-parallel-insns
8926 @opindex mparallel-insns
8927 @opindex mno-parallel-insns
8928 Allow the generation of parallel instructions. This is enabled by
8929 default with @option{-O2}.
8931 @item -mparallel-mpy
8932 @itemx -mno-parallel-mpy
8933 @opindex mparallel-mpy
8934 @opindex mno-parallel-mpy
8935 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
8936 provided @option{-mparallel-insns} is also specified. These instructions have
8937 tight register constraints which can pessimize the code generation
8943 @subsection V850 Options
8944 @cindex V850 Options
8946 These @samp{-m} options are defined for V850 implementations:
8950 @itemx -mno-long-calls
8951 @opindex mlong-calls
8952 @opindex mno-long-calls
8953 Treat all calls as being far away (near). If calls are assumed to be
8954 far away, the compiler will always load the functions address up into a
8955 register, and call indirect through the pointer.
8961 Do not optimize (do optimize) basic blocks that use the same index
8962 pointer 4 or more times to copy pointer into the @code{ep} register, and
8963 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
8964 option is on by default if you optimize.
8966 @item -mno-prolog-function
8967 @itemx -mprolog-function
8968 @opindex mno-prolog-function
8969 @opindex mprolog-function
8970 Do not use (do use) external functions to save and restore registers at
8971 the prolog and epilog of a function. The external functions are slower,
8972 but use less code space if more than one function saves the same number
8973 of registers. The @option{-mprolog-function} option is on by default if
8978 Try to make the code as small as possible. At present, this just turns
8979 on the @option{-mep} and @option{-mprolog-function} options.
8983 Put static or global variables whose size is @var{n} bytes or less into
8984 the tiny data area that register @code{ep} points to. The tiny data
8985 area can hold up to 256 bytes in total (128 bytes for byte references).
8989 Put static or global variables whose size is @var{n} bytes or less into
8990 the small data area that register @code{gp} points to. The small data
8991 area can hold up to 64 kilobytes.
8995 Put static or global variables whose size is @var{n} bytes or less into
8996 the first 32 kilobytes of memory.
9000 Specify that the target processor is the V850.
9003 @opindex mbig-switch
9004 Generate code suitable for big switch tables. Use this option only if
9005 the assembler/linker complain about out of range branches within a switch
9010 @subsection ARC Options
9013 These options are defined for ARC implementations:
9018 Compile code for little endian mode. This is the default.
9022 Compile code for big endian mode.
9025 @opindex mmangle-cpu
9026 Prepend the name of the cpu to all public symbol names.
9027 In multiple-processor systems, there are many ARC variants with different
9028 instruction and register set characteristics. This flag prevents code
9029 compiled for one cpu to be linked with code compiled for another.
9030 No facility exists for handling variants that are ``almost identical''.
9031 This is an all or nothing option.
9033 @item -mcpu=@var{cpu}
9035 Compile code for ARC variant @var{cpu}.
9036 Which variants are supported depend on the configuration.
9037 All variants support @option{-mcpu=base}, this is the default.
9039 @item -mtext=@var{text-section}
9040 @itemx -mdata=@var{data-section}
9041 @itemx -mrodata=@var{readonly-data-section}
9045 Put functions, data, and readonly data in @var{text-section},
9046 @var{data-section}, and @var{readonly-data-section} respectively
9047 by default. This can be overridden with the @code{section} attribute.
9048 @xref{Variable Attributes}.
9053 @subsection NS32K Options
9054 @cindex NS32K options
9056 These are the @samp{-m} options defined for the 32000 series. The default
9057 values for these options depends on which style of 32000 was selected when
9058 the compiler was configured; the defaults for the most common choices are
9066 Generate output for a 32032. This is the default
9067 when the compiler is configured for 32032 and 32016 based systems.
9073 Generate output for a 32332. This is the default
9074 when the compiler is configured for 32332-based systems.
9080 Generate output for a 32532. This is the default
9081 when the compiler is configured for 32532-based systems.
9085 Generate output containing 32081 instructions for floating point.
9086 This is the default for all systems.
9090 Generate output containing 32381 instructions for floating point. This
9091 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9092 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9096 Try and generate multiply-add floating point instructions @code{polyF}
9097 and @code{dotF}. This option is only available if the @option{-m32381}
9098 option is in effect. Using these instructions requires changes to
9099 register allocation which generally has a negative impact on
9100 performance. This option should only be enabled when compiling code
9101 particularly likely to make heavy use of multiply-add instructions.
9104 @opindex mnomulti-add
9105 Do not try and generate multiply-add floating point instructions
9106 @code{polyF} and @code{dotF}. This is the default on all platforms.
9109 @opindex msoft-float
9110 Generate output containing library calls for floating point.
9111 @strong{Warning:} the requisite libraries may not be available.
9114 @opindex mnobitfield
9115 Do not use the bit-field instructions. On some machines it is faster to
9116 use shifting and masking operations. This is the default for the pc532.
9120 Do use the bit-field instructions. This is the default for all platforms
9125 Use a different function-calling convention, in which functions
9126 that take a fixed number of arguments return pop their
9127 arguments on return with the @code{ret} instruction.
9129 This calling convention is incompatible with the one normally
9130 used on Unix, so you cannot use it if you need to call libraries
9131 compiled with the Unix compiler.
9133 Also, you must provide function prototypes for all functions that
9134 take variable numbers of arguments (including @code{printf});
9135 otherwise incorrect code will be generated for calls to those
9138 In addition, seriously incorrect code will result if you call a
9139 function with too many arguments. (Normally, extra arguments are
9140 harmlessly ignored.)
9142 This option takes its name from the 680x0 @code{rtd} instruction.
9147 Use a different function-calling convention where the first two arguments
9148 are passed in registers.
9150 This calling convention is incompatible with the one normally
9151 used on Unix, so you cannot use it if you need to call libraries
9152 compiled with the Unix compiler.
9155 @opindex mnoregparam
9156 Do not pass any arguments in registers. This is the default for all
9161 It is OK to use the sb as an index register which is always loaded with
9162 zero. This is the default for the pc532-netbsd target.
9166 The sb register is not available for use or has not been initialized to
9167 zero by the run time system. This is the default for all targets except
9168 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9169 @option{-fpic} is set.
9173 Many ns32000 series addressing modes use displacements of up to 512MB@.
9174 If an address is above 512MB then displacements from zero can not be used.
9175 This option causes code to be generated which can be loaded above 512MB@.
9176 This may be useful for operating systems or ROM code.
9180 Assume code will be loaded in the first 512MB of virtual address space.
9181 This is the default for all platforms.
9187 @subsection AVR Options
9190 These options are defined for AVR implementations:
9193 @item -mmcu=@var{mcu}
9195 Specify ATMEL AVR instruction set or MCU type.
9197 Instruction set avr1 is for the minimal AVR core, not supported by the C
9198 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9199 attiny11, attiny12, attiny15, attiny28).
9201 Instruction set avr2 (default) is for the classic AVR core with up to
9202 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9203 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9204 at90c8534, at90s8535).
9206 Instruction set avr3 is for the classic AVR core with up to 128K program
9207 memory space (MCU types: atmega103, atmega603).
9209 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9210 memory space (MCU types: atmega83, atmega85).
9212 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9213 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
9217 Output instruction sizes to the asm file.
9219 @item -minit-stack=@var{N}
9220 @opindex minit-stack
9221 Specify the initial stack address, which may be a symbol or numeric value,
9222 @samp{__stack} is the default.
9224 @item -mno-interrupts
9225 @opindex mno-interrupts
9226 Generated code is not compatible with hardware interrupts.
9227 Code size will be smaller.
9229 @item -mcall-prologues
9230 @opindex mcall-prologues
9231 Functions prologues/epilogues expanded as call to appropriate
9232 subroutines. Code size will be smaller.
9234 @item -mno-tablejump
9235 @opindex mno-tablejump
9236 Do not generate tablejump insns which sometimes increase code size.
9239 @opindex mtiny-stack
9240 Change only the low 8 bits of the stack pointer.
9244 @subsection MCore Options
9245 @cindex MCore options
9247 These are the @samp{-m} options defined for the Motorola M*Core
9257 @opindex mno-hardlit
9258 Inline constants into the code stream if it can be done in two
9259 instructions or less.
9267 Use the divide instruction. (Enabled by default).
9269 @item -mrelax-immediate
9270 @itemx -mrelax-immediate
9271 @itemx -mno-relax-immediate
9272 @opindex mrelax-immediate
9273 @opindex mrelax-immediate
9274 @opindex mno-relax-immediate
9275 Allow arbitrary sized immediates in bit operations.
9277 @item -mwide-bitfields
9278 @itemx -mwide-bitfields
9279 @itemx -mno-wide-bitfields
9280 @opindex mwide-bitfields
9281 @opindex mwide-bitfields
9282 @opindex mno-wide-bitfields
9283 Always treat bit-fields as int-sized.
9285 @item -m4byte-functions
9286 @itemx -m4byte-functions
9287 @itemx -mno-4byte-functions
9288 @opindex m4byte-functions
9289 @opindex m4byte-functions
9290 @opindex mno-4byte-functions
9291 Force all functions to be aligned to a four byte boundary.
9293 @item -mcallgraph-data
9294 @itemx -mcallgraph-data
9295 @itemx -mno-callgraph-data
9296 @opindex mcallgraph-data
9297 @opindex mcallgraph-data
9298 @opindex mno-callgraph-data
9299 Emit callgraph information.
9303 @itemx -mno-slow-bytes
9304 @opindex mslow-bytes
9305 @opindex mslow-bytes
9306 @opindex mno-slow-bytes
9307 Prefer word access when reading byte quantities.
9309 @item -mlittle-endian
9310 @itemx -mlittle-endian
9312 @opindex mlittle-endian
9313 @opindex mlittle-endian
9314 @opindex mbig-endian
9315 Generate code for a little endian target.
9323 Generate code for the 210 processor.
9327 @subsection IA-64 Options
9328 @cindex IA-64 Options
9330 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9334 @opindex mbig-endian
9335 Generate code for a big endian target. This is the default for HPUX@.
9337 @item -mlittle-endian
9338 @opindex mlittle-endian
9339 Generate code for a little endian target. This is the default for AIX5
9346 Generate (or don't) code for the GNU assembler. This is the default.
9347 @c Also, this is the default if the configure option @option{--with-gnu-as}
9354 Generate (or don't) code for the GNU linker. This is the default.
9355 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9360 Generate code that does not use a global pointer register. The result
9361 is not position independent code, and violates the IA-64 ABI@.
9363 @item -mvolatile-asm-stop
9364 @itemx -mno-volatile-asm-stop
9365 @opindex mvolatile-asm-stop
9366 @opindex mno-volatile-asm-stop
9367 Generate (or don't) a stop bit immediately before and after volatile asm
9372 Generate code that works around Itanium B step errata.
9374 @item -mregister-names
9375 @itemx -mno-register-names
9376 @opindex mregister-names
9377 @opindex mno-register-names
9378 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9379 the stacked registers. This may make assembler output more readable.
9385 Disable (or enable) optimizations that use the small data section. This may
9386 be useful for working around optimizer bugs.
9389 @opindex mconstant-gp
9390 Generate code that uses a single constant global pointer value. This is
9391 useful when compiling kernel code.
9395 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9396 This is useful when compiling firmware code.
9398 @item -minline-divide-min-latency
9399 @opindex minline-divide-min-latency
9400 Generate code for inline divides using the minimum latency algorithm.
9402 @item -minline-divide-max-throughput
9403 @opindex minline-divide-max-throughput
9404 Generate code for inline divides using the maximum throughput algorithm.
9406 @item -mno-dwarf2-asm
9408 @opindex mno-dwarf2-asm
9409 @opindex mdwarf2-asm
9410 Don't (or do) generate assembler code for the DWARF2 line number debugging
9411 info. This may be useful when not using the GNU assembler.
9413 @item -mfixed-range=@var{register-range}
9414 @opindex mfixed-range
9415 Generate code treating the given register range as fixed registers.
9416 A fixed register is one that the register allocator can not use. This is
9417 useful when compiling kernel code. A register range is specified as
9418 two registers separated by a dash. Multiple register ranges can be
9419 specified separated by a comma.
9423 @subsection D30V Options
9424 @cindex D30V Options
9426 These @samp{-m} options are defined for D30V implementations:
9431 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
9432 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
9433 memory, which starts at location @code{0x80000000}.
9437 Same as the @option{-mextmem} switch.
9441 Link the @samp{.text} section into onchip text memory, which starts at
9442 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
9443 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
9444 into onchip data memory, which starts at location @code{0x20000000}.
9446 @item -mno-asm-optimize
9447 @itemx -masm-optimize
9448 @opindex mno-asm-optimize
9449 @opindex masm-optimize
9450 Disable (enable) passing @option{-O} to the assembler when optimizing.
9451 The assembler uses the @option{-O} option to automatically parallelize
9452 adjacent short instructions where possible.
9454 @item -mbranch-cost=@var{n}
9455 @opindex mbranch-cost
9456 Increase the internal costs of branches to @var{n}. Higher costs means
9457 that the compiler will issue more instructions to avoid doing a branch.
9460 @item -mcond-exec=@var{n}
9462 Specify the maximum number of conditionally executed instructions that
9463 replace a branch. The default is 4.
9466 @node S/390 and zSeries Options
9467 @subsection S/390 and zSeries Options
9468 @cindex S/390 and zSeries Options
9470 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9475 @opindex mhard-float
9476 @opindex msoft-float
9477 Use (do not use) the hardware floating-point instructions and registers
9478 for floating-point operations. When @option{-msoft-float} is specified,
9479 functions in @file{libgcc.a} will be used to perform floating-point
9480 operations. When @option{-mhard-float} is specified, the compiler
9481 generates IEEE floating-point instructions. This is the default.
9484 @itemx -mno-backchain
9486 @opindex mno-backchain
9487 Generate (or do not generate) code which maintains an explicit
9488 backchain within the stack frame that points to the caller's frame.
9489 This is currently needed to allow debugging. The default is to
9490 generate the backchain.
9493 @itemx -mno-small-exec
9494 @opindex msmall-exec
9495 @opindex mno-small-exec
9496 Generate (or do not generate) code using the @code{bras} instruction
9497 to do subroutine calls.
9498 This only works reliably if the total executable size does not
9499 exceed 64k. The default is to use the @code{basr} instruction instead,
9500 which does not have this limitation.
9506 When @option{-m31} is specified, generate code compliant to the
9507 Linux for S/390 ABI@. When @option{-m64} is specified, generate
9508 code compliant to the Linux for zSeries ABI@. This allows GCC in
9509 particular to generate 64-bit instructions. For the @samp{s390}
9510 targets, the default is @option{-m31}, while the @samp{s390x}
9511 targets default to @option{-m64}.
9517 Generate (or do not generate) code using the @code{mvcle} instruction
9518 to perform block moves. When @option{-mno-mvcle} is specifed,
9519 use a @code{mvc} loop instead. This is the default.
9525 Print (or do not print) additional debug information when compiling.
9526 The default is to not print debug information.
9531 @subsection CRIS Options
9532 @cindex CRIS Options
9534 These options are defined specifically for the CRIS ports.
9537 @item -march=@var{architecture-type}
9538 @itemx -mcpu=@var{architecture-type}
9541 Generate code for the specified architecture. The choices for
9542 @var{architecture-type} are @samp{v3}, @samp{v8} and @samp{v10} for
9543 respectively ETRAX@w{ }4, ETRAX@w{ }100, and ETRAX@w{ }100@w{ }LX.
9544 Default is @samp{v0} except for cris-axis-linux-gnu, where the default is
9547 @item -mtune=@var{architecture-type}
9549 Tune to @var{architecture-type} everything applicable about the generated
9550 code, except for the ABI and the set of available instructions. The
9551 choices for @var{architecture-type} are the same as for
9552 @option{-march=@var{architecture-type}}.
9554 @item -mmax-stack-frame=@var{n}
9555 @opindex mmax-stack-frame
9556 Warn when the stack frame of a function exceeds @var{n} bytes.
9558 @item -melinux-stacksize=@var{n}
9559 @opindex melinux-stacksize
9560 Only available with the @samp{cris-axis-aout} target. Arranges for
9561 indications in the program to the kernel loader that the stack of the
9562 program should be set to @var{n} bytes.
9568 The options @option{-metrax4} and @option{-metrax100} are synonyms for
9569 @option{-march=v3} and @option{-march=v8} respectively.
9573 Enable CRIS-specific verbose debug-related information in the assembly
9574 code. This option also has the effect to turn off the @samp{#NO_APP}
9575 formatted-code indicator to the assembler at the beginning of the
9580 Do not use condition-code results from previous instruction; always emit
9581 compare and test instructions before use of condition codes.
9583 @item -mno-side-effects
9584 @opindex mno-side-effects
9585 Do not emit instructions with side-effects in addressing modes other than
9589 @itemx -mno-stack-align
9591 @itemx -mno-data-align
9592 @itemx -mconst-align
9593 @itemx -mno-const-align
9594 @opindex mstack-align
9595 @opindex mno-stack-align
9596 @opindex mdata-align
9597 @opindex mno-data-align
9598 @opindex mconst-align
9599 @opindex mno-const-align
9600 These options (no-options) arranges (eliminate arrangements) for the
9601 stack-frame, individual data and constants to be aligned for the maximum
9602 single data access size for the chosen CPU model. The default is to
9603 arrange for 32-bit alignment. ABI details such as structure layout are
9604 not affected by these options.
9612 Similar to the stack- data- and const-align options above, these options
9613 arrange for stack-frame, writable data and constants to all be 32-bit,
9614 16-bit or 8-bit aligned. The default is 32-bit alignment.
9616 @item -mno-prologue-epilogue
9617 @itemx -mprologue-epilogue
9618 @opindex mno-prologue-epilogue
9619 @opindex mprologue-epilogue
9620 With @option{-mno-prologue-epilogue}, the normal function prologue and
9621 epilogue that sets up the stack-frame are omitted and no return
9622 instructions or return sequences are generated in the code. Use this
9623 option only together with visual inspection of the compiled code: no
9624 warnings or errors are generated when call-saved registers must be saved,
9625 or storage for local variable needs to be allocated.
9631 With @option{-fpic} and @option{-fPIC}, don't generate (do generate)
9632 instruction sequences that load addresses for functions from the PLT part
9633 of the GOT rather than (traditional on other architectures) calls to the
9634 PLT. The default is @option{-mgotplt}.
9638 Legacy no-op option only recognized with the cris-axis-aout target.
9642 Legacy no-op option only recognized with the cris-axis-elf and
9643 cris-axis-linux-gnu targets.
9647 Only recognized with the cris-axis-aout target, where it selects a
9648 GNU/linux-like multilib, include files and instruction set for
9653 Legacy no-op option only recognized with the cris-axis-linux-gnu target.
9657 This option, recognized for the cris-axis-aout and cris-axis-elf arranges
9658 to link with input-output functions from a simulator library. Code,
9659 initialized data and zero-initialized data are allocated consecutively.
9663 Like @option{-sim}, but pass linker options to locate initialized data at
9664 0x40000000 and zero-initialized data at 0x80000000.
9668 @subsection MMIX Options
9669 @cindex MMIX Options
9671 These options are defined for the MMIX:
9675 @itemx -mno-libfuncs
9676 Specify that intrinsic library functions are being compiled, passing all
9677 values in registers, no matter the size.
9681 Generate floating-point comparison instructions that compare with respect
9682 to the @code{rE} epsilon register.
9684 @item -mabi=mmixware
9686 Generate code that passes function parameters and return values that (in
9687 the called function) are seen as registers @code{$0} and up, as opposed to
9688 the GNU ABI which uses global registers @code{$231} and up.
9691 @item -mno-zero-extend
9692 When reading data from memory in sizes shorter than 64 bits, use (do not
9693 use) zero-extending load instructions by default, rather than
9694 sign-extending ones.
9697 @itemx -mno-knuthdiv
9698 Make the result of a division yielding a remainder have the same sign as
9699 the divisor. With the default, @option{-mno-knuthdiv}, the sign of the
9700 remainder follows the sign of the dividend. Both methods are
9701 arithmetically valid, the latter being almost exclusively used.
9703 @item -mtoplevel-symbols
9704 @itemx -mno-toplevel-symbols
9705 Prepend (do not prepend) a @samp{:} to all global symbols, so the assembly
9706 code can be used with the @code{PREFIX} assembly directive.
9709 Generate an executable in the ELF format, rather than the default
9710 @samp{mmo} format used by the @command{mmix} simulator.
9714 @node Code Gen Options
9715 @section Options for Code Generation Conventions
9716 @cindex code generation conventions
9717 @cindex options, code generation
9718 @cindex run-time options
9720 These machine-independent options control the interface conventions
9721 used in code generation.
9723 Most of them have both positive and negative forms; the negative form
9724 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
9725 one of the forms is listed---the one which is not the default. You
9726 can figure out the other form by either removing @samp{no-} or adding
9731 @opindex fexceptions
9732 Enable exception handling. Generates extra code needed to propagate
9733 exceptions. For some targets, this implies GCC will generate frame
9734 unwind information for all functions, which can produce significant data
9735 size overhead, although it does not affect execution. If you do not
9736 specify this option, GCC will enable it by default for languages like
9737 C++ which normally require exception handling, and disable it for
9738 languages like C that do not normally require it. However, you may need
9739 to enable this option when compiling C code that needs to interoperate
9740 properly with exception handlers written in C++. You may also wish to
9741 disable this option if you are compiling older C++ programs that don't
9742 use exception handling.
9744 @item -fnon-call-exceptions
9745 @opindex fnon-call-exceptions
9746 Generate code that allows trapping instructions to throw exceptions.
9747 Note that this requires platform-specific runtime support that does
9748 not exist everywhere. Moreover, it only allows @emph{trapping}
9749 instructions to throw exceptions, i.e.@: memory references or floating
9750 point instructions. It does not allow exceptions to be thrown from
9751 arbitrary signal handlers such as @code{SIGALRM}.
9753 @item -funwind-tables
9754 @opindex funwind-tables
9755 Similar to @option{-fexceptions}, except that it will just generate any needed
9756 static data, but will not affect the generated code in any other way.
9757 You will normally not enable this option; instead, a language processor
9758 that needs this handling would enable it on your behalf.
9760 @item -fasynchronous-unwind-tables
9761 @opindex funwind-tables
9762 Generate unwind table in dwarf2 format, if supported by target machine. The
9763 table is exact at each instruction boundary, so it can be used for stack
9764 unwinding from asynchronous events (such as debugger or garbage collector).
9766 @item -fpcc-struct-return
9767 @opindex fpcc-struct-return
9768 Return ``short'' @code{struct} and @code{union} values in memory like
9769 longer ones, rather than in registers. This convention is less
9770 efficient, but it has the advantage of allowing intercallability between
9771 GCC-compiled files and files compiled with other compilers.
9773 The precise convention for returning structures in memory depends
9774 on the target configuration macros.
9776 Short structures and unions are those whose size and alignment match
9777 that of some integer type.
9779 @item -freg-struct-return
9780 @opindex freg-struct-return
9781 Return @code{struct} and @code{union} values in registers when possible.
9782 This is more efficient for small structures than
9783 @option{-fpcc-struct-return}.
9785 If you specify neither @option{-fpcc-struct-return} nor
9786 @option{-freg-struct-return}, GCC defaults to whichever convention is
9787 standard for the target. If there is no standard convention, GCC
9788 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
9789 the principal compiler. In those cases, we can choose the standard, and
9790 we chose the more efficient register return alternative.
9793 @opindex fshort-enums
9794 Allocate to an @code{enum} type only as many bytes as it needs for the
9795 declared range of possible values. Specifically, the @code{enum} type
9796 will be equivalent to the smallest integer type which has enough room.
9798 @item -fshort-double
9799 @opindex fshort-double
9800 Use the same size for @code{double} as for @code{float}.
9803 @opindex fshared-data
9804 Requests that the data and non-@code{const} variables of this
9805 compilation be shared data rather than private data. The distinction
9806 makes sense only on certain operating systems, where shared data is
9807 shared between processes running the same program, while private data
9808 exists in one copy per process.
9812 In C, allocate even uninitialized global variables in the data section of the
9813 object file, rather than generating them as common blocks. This has the
9814 effect that if the same variable is declared (without @code{extern}) in
9815 two different compilations, you will get an error when you link them.
9816 The only reason this might be useful is if you wish to verify that the
9817 program will work on other systems which always work this way.
9821 Ignore the @samp{#ident} directive.
9823 @item -fno-gnu-linker
9824 @opindex fno-gnu-linker
9825 Do not output global initializations (such as C++ constructors and
9826 destructors) in the form used by the GNU linker (on systems where the GNU
9827 linker is the standard method of handling them). Use this option when
9828 you want to use a non-GNU linker, which also requires using the
9829 @command{collect2} program to make sure the system linker includes
9830 constructors and destructors. (@command{collect2} is included in the GCC
9831 distribution.) For systems which @emph{must} use @command{collect2}, the
9832 compiler driver @command{gcc} is configured to do this automatically.
9834 @item -finhibit-size-directive
9835 @opindex finhibit-size-directive
9836 Don't output a @code{.size} assembler directive, or anything else that
9837 would cause trouble if the function is split in the middle, and the
9838 two halves are placed at locations far apart in memory. This option is
9839 used when compiling @file{crtstuff.c}; you should not need to use it
9843 @opindex fverbose-asm
9844 Put extra commentary information in the generated assembly code to
9845 make it more readable. This option is generally only of use to those
9846 who actually need to read the generated assembly code (perhaps while
9847 debugging the compiler itself).
9849 @option{-fno-verbose-asm}, the default, causes the
9850 extra information to be omitted and is useful when comparing two assembler
9855 Consider all memory references through pointers to be volatile.
9857 @item -fvolatile-global
9858 @opindex fvolatile-global
9859 Consider all memory references to extern and global data items to
9860 be volatile. GCC does not consider static data items to be volatile
9861 because of this switch.
9863 @item -fvolatile-static
9864 @opindex fvolatile-static
9865 Consider all memory references to static data to be volatile.
9869 @cindex global offset table
9871 Generate position-independent code (PIC) suitable for use in a shared
9872 library, if supported for the target machine. Such code accesses all
9873 constant addresses through a global offset table (GOT)@. The dynamic
9874 loader resolves the GOT entries when the program starts (the dynamic
9875 loader is not part of GCC; it is part of the operating system). If
9876 the GOT size for the linked executable exceeds a machine-specific
9877 maximum size, you get an error message from the linker indicating that
9878 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
9879 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
9880 on the m68k and RS/6000. The 386 has no such limit.)
9882 Position-independent code requires special support, and therefore works
9883 only on certain machines. For the 386, GCC supports PIC for System V
9884 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
9885 position-independent.
9889 If supported for the target machine, emit position-independent code,
9890 suitable for dynamic linking and avoiding any limit on the size of the
9891 global offset table. This option makes a difference on the m68k, m88k,
9894 Position-independent code requires special support, and therefore works
9895 only on certain machines.
9897 @item -ffixed-@var{reg}
9899 Treat the register named @var{reg} as a fixed register; generated code
9900 should never refer to it (except perhaps as a stack pointer, frame
9901 pointer or in some other fixed role).
9903 @var{reg} must be the name of a register. The register names accepted
9904 are machine-specific and are defined in the @code{REGISTER_NAMES}
9905 macro in the machine description macro file.
9907 This flag does not have a negative form, because it specifies a
9910 @item -fcall-used-@var{reg}
9912 Treat the register named @var{reg} as an allocable register that is
9913 clobbered by function calls. It may be allocated for temporaries or
9914 variables that do not live across a call. Functions compiled this way
9915 will not save and restore the register @var{reg}.
9917 It is an error to used this flag with the frame pointer or stack pointer.
9918 Use of this flag for other registers that have fixed pervasive roles in
9919 the machine's execution model will produce disastrous results.
9921 This flag does not have a negative form, because it specifies a
9924 @item -fcall-saved-@var{reg}
9925 @opindex fcall-saved
9926 Treat the register named @var{reg} as an allocable register saved by
9927 functions. It may be allocated even for temporaries or variables that
9928 live across a call. Functions compiled this way will save and restore
9929 the register @var{reg} if they use it.
9931 It is an error to used this flag with the frame pointer or stack pointer.
9932 Use of this flag for other registers that have fixed pervasive roles in
9933 the machine's execution model will produce disastrous results.
9935 A different sort of disaster will result from the use of this flag for
9936 a register in which function values may be returned.
9938 This flag does not have a negative form, because it specifies a
9942 @opindex fpack-struct
9943 Pack all structure members together without holes. Usually you would
9944 not want to use this option, since it makes the code suboptimal, and
9945 the offsets of structure members won't agree with system libraries.
9947 @item -fcheck-memory-usage
9948 @opindex fcheck-memory-usage
9949 Generate extra code to check each memory access. GCC will generate
9950 code that is suitable for a detector of bad memory accesses such as
9953 Normally, you should compile all, or none, of your code with this option.
9955 If you do mix code compiled with and without this option,
9956 you must ensure that all code that has side effects
9957 and that is called by code compiled with this option
9958 is, itself, compiled with this option.
9959 If you do not, you might get erroneous messages from the detector.
9961 If you use functions from a library that have side-effects (such as
9962 @code{read}), you might not be able to recompile the library and
9963 specify this option. In that case, you can enable the
9964 @option{-fprefix-function-name} option, which requests GCC to encapsulate
9965 your code and make other functions look as if they were compiled with
9966 @option{-fcheck-memory-usage}. This is done by calling ``stubs'',
9967 which are provided by the detector. If you cannot find or build
9968 stubs for every function you call, you might have to specify
9969 @option{-fcheck-memory-usage} without @option{-fprefix-function-name}.
9971 If you specify this option, you can not use the @code{asm} or
9972 @code{__asm__} keywords in functions with memory checking enabled. GCC
9973 cannot understand what the @code{asm} statement may do, and therefore
9974 cannot generate the appropriate code, so it will reject it. However, if
9975 you specify the function attribute @code{no_check_memory_usage}
9976 (@pxref{Function Attributes}), GCC will disable memory checking within a
9977 function; you may use @code{asm} statements inside such functions. You
9978 may have an inline expansion of a non-checked function within a checked
9979 function; in that case GCC will not generate checks for the inlined
9980 function's memory accesses.
9982 If you move your @code{asm} statements to non-checked inline functions
9983 and they do access memory, you can add calls to the support code in your
9984 inline function, to indicate any reads, writes, or copies being done.
9985 These calls would be similar to those done in the stubs described above.
9987 @item -fprefix-function-name
9988 @opindex fprefix-function-name
9989 Request GCC to add a prefix to the symbols generated for function names.
9990 GCC adds a prefix to the names of functions defined as well as
9991 functions called. Code compiled with this option and code compiled
9992 without the option can't be linked together, unless stubs are used.
9994 If you compile the following code with @option{-fprefix-function-name}
9996 extern void bar (int);
10000 return bar (a + 5);
10005 GCC will compile the code as if it was written:
10007 extern void prefix_bar (int);
10011 return prefix_bar (a + 5);
10014 This option is designed to be used with @option{-fcheck-memory-usage}.
10016 @item -finstrument-functions
10017 @opindex finstrument-functions
10018 Generate instrumentation calls for entry and exit to functions. Just
10019 after function entry and just before function exit, the following
10020 profiling functions will be called with the address of the current
10021 function and its call site. (On some platforms,
10022 @code{__builtin_return_address} does not work beyond the current
10023 function, so the call site information may not be available to the
10024 profiling functions otherwise.)
10027 void __cyg_profile_func_enter (void *this_fn,
10029 void __cyg_profile_func_exit (void *this_fn,
10033 The first argument is the address of the start of the current function,
10034 which may be looked up exactly in the symbol table.
10036 This instrumentation is also done for functions expanded inline in other
10037 functions. The profiling calls will indicate where, conceptually, the
10038 inline function is entered and exited. This means that addressable
10039 versions of such functions must be available. If all your uses of a
10040 function are expanded inline, this may mean an additional expansion of
10041 code size. If you use @samp{extern inline} in your C code, an
10042 addressable version of such functions must be provided. (This is
10043 normally the case anyways, but if you get lucky and the optimizer always
10044 expands the functions inline, you might have gotten away without
10045 providing static copies.)
10047 A function may be given the attribute @code{no_instrument_function}, in
10048 which case this instrumentation will not be done. This can be used, for
10049 example, for the profiling functions listed above, high-priority
10050 interrupt routines, and any functions from which the profiling functions
10051 cannot safely be called (perhaps signal handlers, if the profiling
10052 routines generate output or allocate memory).
10054 @item -fstack-check
10055 @opindex fstack-check
10056 Generate code to verify that you do not go beyond the boundary of the
10057 stack. You should specify this flag if you are running in an
10058 environment with multiple threads, but only rarely need to specify it in
10059 a single-threaded environment since stack overflow is automatically
10060 detected on nearly all systems if there is only one stack.
10062 Note that this switch does not actually cause checking to be done; the
10063 operating system must do that. The switch causes generation of code
10064 to ensure that the operating system sees the stack being extended.
10066 @item -fstack-limit-register=@var{reg}
10067 @itemx -fstack-limit-symbol=@var{sym}
10068 @itemx -fno-stack-limit
10069 @opindex fstack-limit-register
10070 @opindex fstack-limit-symbol
10071 @opindex fno-stack-limit
10072 Generate code to ensure that the stack does not grow beyond a certain value,
10073 either the value of a register or the address of a symbol. If the stack
10074 would grow beyond the value, a signal is raised. For most targets,
10075 the signal is raised before the stack overruns the boundary, so
10076 it is possible to catch the signal without taking special precautions.
10078 For instance, if the stack starts at absolute address @samp{0x80000000}
10079 and grows downwards, you can use the flags
10080 @option{-fstack-limit-symbol=__stack_limit} and
10081 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
10082 of 128KB@. Note that this may only work with the GNU linker.
10084 @cindex aliasing of parameters
10085 @cindex parameters, aliased
10086 @item -fargument-alias
10087 @itemx -fargument-noalias
10088 @itemx -fargument-noalias-global
10089 @opindex fargument-alias
10090 @opindex fargument-noalias
10091 @opindex fargument-noalias-global
10092 Specify the possible relationships among parameters and between
10093 parameters and global data.
10095 @option{-fargument-alias} specifies that arguments (parameters) may
10096 alias each other and may alias global storage.@*
10097 @option{-fargument-noalias} specifies that arguments do not alias
10098 each other, but may alias global storage.@*
10099 @option{-fargument-noalias-global} specifies that arguments do not
10100 alias each other and do not alias global storage.
10102 Each language will automatically use whatever option is required by
10103 the language standard. You should not need to use these options yourself.
10105 @item -fleading-underscore
10106 @opindex fleading-underscore
10107 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
10108 change the way C symbols are represented in the object file. One use
10109 is to help link with legacy assembly code.
10111 Be warned that you should know what you are doing when invoking this
10112 option, and that not all targets provide complete support for it.
10117 @node Environment Variables
10118 @section Environment Variables Affecting GCC
10119 @cindex environment variables
10121 @c man begin ENVIRONMENT
10123 This section describes several environment variables that affect how GCC
10124 operates. Some of them work by specifying directories or prefixes to use
10125 when searching for various kinds of files. Some are used to specify other
10126 aspects of the compilation environment.
10129 Note that you can also specify places to search using options such as
10130 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
10131 take precedence over places specified using environment variables, which
10132 in turn take precedence over those specified by the configuration of GCC@.
10136 Note that you can also specify places to search using options such as
10137 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
10138 take precedence over places specified using environment variables, which
10139 in turn take precedence over those specified by the configuration of GCC@.
10146 @c @itemx LC_COLLATE
10148 @c @itemx LC_MONETARY
10149 @c @itemx LC_NUMERIC
10154 @c @findex LC_COLLATE
10155 @findex LC_MESSAGES
10156 @c @findex LC_MONETARY
10157 @c @findex LC_NUMERIC
10161 These environment variables control the way that GCC uses
10162 localization information that allow GCC to work with different
10163 national conventions. GCC inspects the locale categories
10164 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
10165 so. These locale categories can be set to any value supported by your
10166 installation. A typical value is @samp{en_UK} for English in the United
10169 The @env{LC_CTYPE} environment variable specifies character
10170 classification. GCC uses it to determine the character boundaries in
10171 a string; this is needed for some multibyte encodings that contain quote
10172 and escape characters that would otherwise be interpreted as a string
10175 The @env{LC_MESSAGES} environment variable specifies the language to
10176 use in diagnostic messages.
10178 If the @env{LC_ALL} environment variable is set, it overrides the value
10179 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
10180 and @env{LC_MESSAGES} default to the value of the @env{LANG}
10181 environment variable. If none of these variables are set, GCC
10182 defaults to traditional C English behavior.
10186 If @env{TMPDIR} is set, it specifies the directory to use for temporary
10187 files. GCC uses temporary files to hold the output of one stage of
10188 compilation which is to be used as input to the next stage: for example,
10189 the output of the preprocessor, which is the input to the compiler
10192 @item GCC_EXEC_PREFIX
10193 @findex GCC_EXEC_PREFIX
10194 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
10195 names of the subprograms executed by the compiler. No slash is added
10196 when this prefix is combined with the name of a subprogram, but you can
10197 specify a prefix that ends with a slash if you wish.
10199 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
10200 an appropriate prefix to use based on the pathname it was invoked with.
10202 If GCC cannot find the subprogram using the specified prefix, it
10203 tries looking in the usual places for the subprogram.
10205 The default value of @env{GCC_EXEC_PREFIX} is
10206 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
10207 of @code{prefix} when you ran the @file{configure} script.
10209 Other prefixes specified with @option{-B} take precedence over this prefix.
10211 This prefix is also used for finding files such as @file{crt0.o} that are
10214 In addition, the prefix is used in an unusual way in finding the
10215 directories to search for header files. For each of the standard
10216 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
10217 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
10218 replacing that beginning with the specified prefix to produce an
10219 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
10220 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
10221 These alternate directories are searched first; the standard directories
10224 @item COMPILER_PATH
10225 @findex COMPILER_PATH
10226 The value of @env{COMPILER_PATH} is a colon-separated list of
10227 directories, much like @env{PATH}. GCC tries the directories thus
10228 specified when searching for subprograms, if it can't find the
10229 subprograms using @env{GCC_EXEC_PREFIX}.
10232 @findex LIBRARY_PATH
10233 The value of @env{LIBRARY_PATH} is a colon-separated list of
10234 directories, much like @env{PATH}. When configured as a native compiler,
10235 GCC tries the directories thus specified when searching for special
10236 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
10237 using GCC also uses these directories when searching for ordinary
10238 libraries for the @option{-l} option (but directories specified with
10239 @option{-L} come first).
10241 @item C_INCLUDE_PATH
10242 @itemx CPLUS_INCLUDE_PATH
10243 @itemx OBJC_INCLUDE_PATH
10244 @findex C_INCLUDE_PATH
10245 @findex CPLUS_INCLUDE_PATH
10246 @findex OBJC_INCLUDE_PATH
10247 @c @itemx OBJCPLUS_INCLUDE_PATH
10248 These environment variables pertain to particular languages. Each
10249 variable's value is a colon-separated list of directories, much like
10250 @env{PATH}. When GCC searches for header files, it tries the
10251 directories listed in the variable for the language you are using, after
10252 the directories specified with @option{-I} but before the standard header
10255 @item DEPENDENCIES_OUTPUT
10256 @findex DEPENDENCIES_OUTPUT
10257 @cindex dependencies for make as output
10258 If this variable is set, its value specifies how to output dependencies
10259 for Make based on the header files processed by the compiler. This
10260 output looks much like the output from the @option{-M} option
10261 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
10262 in addition to the usual results of compilation.
10264 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
10265 which case the Make rules are written to that file, guessing the target
10266 name from the source file name. Or the value can have the form
10267 @samp{@var{file} @var{target}}, in which case the rules are written to
10268 file @var{file} using @var{target} as the target name.
10272 @cindex locale definition
10273 This variable is used to pass locale information to the compiler. One way in
10274 which this information is used is to determine the character set to be used
10275 when character literals, string literals and comments are parsed in C and C++.
10276 When the compiler is configured to allow multibyte characters,
10277 the following values for @env{LANG} are recognized:
10281 Recognize JIS characters.
10283 Recognize SJIS characters.
10285 Recognize EUCJP characters.
10288 If @env{LANG} is not defined, or if it has some other value, then the
10289 compiler will use mblen and mbtowc as defined by the default locale to
10290 recognize and translate multibyte characters.
10295 @node Running Protoize
10296 @section Running Protoize
10298 The program @code{protoize} is an optional part of GCC@. You can use
10299 it to add prototypes to a program, thus converting the program to ISO
10300 C in one respect. The companion program @code{unprotoize} does the
10301 reverse: it removes argument types from any prototypes that are found.
10303 When you run these programs, you must specify a set of source files as
10304 command line arguments. The conversion programs start out by compiling
10305 these files to see what functions they define. The information gathered
10306 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
10308 After scanning comes actual conversion. The specified files are all
10309 eligible to be converted; any files they include (whether sources or
10310 just headers) are eligible as well.
10312 But not all the eligible files are converted. By default,
10313 @code{protoize} and @code{unprotoize} convert only source and header
10314 files in the current directory. You can specify additional directories
10315 whose files should be converted with the @option{-d @var{directory}}
10316 option. You can also specify particular files to exclude with the
10317 @option{-x @var{file}} option. A file is converted if it is eligible, its
10318 directory name matches one of the specified directory names, and its
10319 name within the directory has not been excluded.
10321 Basic conversion with @code{protoize} consists of rewriting most
10322 function definitions and function declarations to specify the types of
10323 the arguments. The only ones not rewritten are those for varargs
10326 @code{protoize} optionally inserts prototype declarations at the
10327 beginning of the source file, to make them available for any calls that
10328 precede the function's definition. Or it can insert prototype
10329 declarations with block scope in the blocks where undeclared functions
10332 Basic conversion with @code{unprotoize} consists of rewriting most
10333 function declarations to remove any argument types, and rewriting
10334 function definitions to the old-style pre-ISO form.
10336 Both conversion programs print a warning for any function declaration or
10337 definition that they can't convert. You can suppress these warnings
10340 The output from @code{protoize} or @code{unprotoize} replaces the
10341 original source file. The original file is renamed to a name ending
10342 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
10343 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
10344 for DOS) file already exists, then the source file is simply discarded.
10346 @code{protoize} and @code{unprotoize} both depend on GCC itself to
10347 scan the program and collect information about the functions it uses.
10348 So neither of these programs will work until GCC is installed.
10350 Here is a table of the options you can use with @code{protoize} and
10351 @code{unprotoize}. Each option works with both programs unless
10355 @item -B @var{directory}
10356 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
10357 usual directory (normally @file{/usr/local/lib}). This file contains
10358 prototype information about standard system functions. This option
10359 applies only to @code{protoize}.
10361 @item -c @var{compilation-options}
10362 Use @var{compilation-options} as the options when running @code{gcc} to
10363 produce the @samp{.X} files. The special option @option{-aux-info} is
10364 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
10366 Note that the compilation options must be given as a single argument to
10367 @code{protoize} or @code{unprotoize}. If you want to specify several
10368 @code{gcc} options, you must quote the entire set of compilation options
10369 to make them a single word in the shell.
10371 There are certain @code{gcc} arguments that you cannot use, because they
10372 would produce the wrong kind of output. These include @option{-g},
10373 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
10374 the @var{compilation-options}, they are ignored.
10377 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
10378 systems) instead of @samp{.c}. This is convenient if you are converting
10379 a C program to C++. This option applies only to @code{protoize}.
10382 Add explicit global declarations. This means inserting explicit
10383 declarations at the beginning of each source file for each function
10384 that is called in the file and was not declared. These declarations
10385 precede the first function definition that contains a call to an
10386 undeclared function. This option applies only to @code{protoize}.
10388 @item -i @var{string}
10389 Indent old-style parameter declarations with the string @var{string}.
10390 This option applies only to @code{protoize}.
10392 @code{unprotoize} converts prototyped function definitions to old-style
10393 function definitions, where the arguments are declared between the
10394 argument list and the initial @samp{@{}. By default, @code{unprotoize}
10395 uses five spaces as the indentation. If you want to indent with just
10396 one space instead, use @option{-i " "}.
10399 Keep the @samp{.X} files. Normally, they are deleted after conversion
10403 Add explicit local declarations. @code{protoize} with @option{-l} inserts
10404 a prototype declaration for each function in each block which calls the
10405 function without any declaration. This option applies only to
10409 Make no real changes. This mode just prints information about the conversions
10410 that would have been done without @option{-n}.
10413 Make no @samp{.save} files. The original files are simply deleted.
10414 Use this option with caution.
10416 @item -p @var{program}
10417 Use the program @var{program} as the compiler. Normally, the name
10418 @file{gcc} is used.
10421 Work quietly. Most warnings are suppressed.
10424 Print the version number, just like @option{-v} for @code{gcc}.
10427 If you need special compiler options to compile one of your program's
10428 source files, then you should generate that file's @samp{.X} file
10429 specially, by running @code{gcc} on that source file with the
10430 appropriate options and the option @option{-aux-info}. Then run
10431 @code{protoize} on the entire set of files. @code{protoize} will use
10432 the existing @samp{.X} file because it is newer than the source file.
10436 gcc -Dfoo=bar file1.c -aux-info file1.X
10441 You need to include the special files along with the rest in the
10442 @code{protoize} command, even though their @samp{.X} files already
10443 exist, because otherwise they won't get converted.
10445 @xref{Protoize Caveats}, for more information on how to use
10446 @code{protoize} successfully.