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 make and distribute verbatim copies of this
12 manual provided the copyright notice and this permission notice are
13 preserved on all copies.
15 Permission is granted to copy and distribute modified versions of this
16 manual under the conditions for verbatim copying, provided also that the
17 entire resulting derived work is distributed under the terms of a
18 permission notice identical to this one.
20 Permission is granted to copy and distribute translations of this manual
21 into another language, under the above conditions for modified versions,
22 except that this permission notice may be included in translations
23 approved by the Free Software Foundation instead of in the original
26 @c Set file name and title for the man page.
28 @settitle GNU project C and C++ compiler
30 gcc [@option{-c}|@option{-S}|@option{-E}] [@option{-std=}@var{standard}]
31 [@option{-g}] [@option{-pg}] [@option{-O}@var{level}]
32 [@option{-W}@var{warn}@dots{}] [@option{-pedantic}]
33 [@option{-I}@var{dir}@dots{}] [@option{-L}@var{dir}@dots{}]
34 [@option{-D}@var{macro}[=@var{defn}]@dots{}] [@option{-U}@var{macro}]
35 [@option{-f}@var{option}@dots{}] [@option{-m}@var{machine-option}@dots{}]
36 [@option{-o} @var{outfile}] @var{infile}@dots{}
38 Only the most useful options are listed here; see below for the
39 remainder. @samp{g++} accepts mostly the same options as @samp{gcc}.
42 cpp(1), gcov(1), g77(1), as(1), ld(1), gdb(1), adb(1), dbx(1), sdb(1)
43 and the Info entries for @file{gcc}, @file{cpp}, @file{g77}, @file{as},
44 @file{ld}, @file{binutils} and @file{gdb}.
47 For instructions on reporting bugs, see
48 @w{@uref{http://gcc.gnu.org/bugs.html}}. Use of the @command{gccbug}
49 script to report bugs is recommended.
52 See the Info entry for @file{gcc}, or
53 @w{@uref{http://gcc.gnu.org/thanks.html}}, for contributors to GCC@.
58 @chapter GCC Command Options
59 @cindex GCC command options
60 @cindex command options
61 @cindex options, GCC command
63 @c man begin DESCRIPTION
65 When you invoke GCC, it normally does preprocessing, compilation,
66 assembly and linking. The ``overall options'' allow you to stop this
67 process at an intermediate stage. For example, the @option{-c} option
68 says not to run the linker. Then the output consists of object files
69 output by the assembler.
71 Other options are passed on to one stage of processing. Some options
72 control the preprocessor and others the compiler itself. Yet other
73 options control the assembler and linker; most of these are not
74 documented here, since you rarely need to use any of them.
76 @cindex C compilation options
77 Most of the command line options that you can use with GCC are useful
78 for C programs; when an option is only useful with another language
79 (usually C++), the explanation says so explicitly. If the description
80 for a particular option does not mention a source language, you can use
81 that option with all supported languages.
83 @cindex C++ compilation options
84 @xref{Invoking G++,,Compiling C++ Programs}, for a summary of special
85 options for compiling C++ programs.
87 @cindex grouping options
88 @cindex options, grouping
89 The @command{gcc} program accepts options and file names as operands. Many
90 options have multi-letter names; therefore multiple single-letter options
91 may @emph{not} be grouped: @option{-dr} is very different from @w{@samp{-d
94 @cindex order of options
95 @cindex options, order
96 You can mix options and other arguments. For the most part, the order
97 you use doesn't matter. Order does matter when you use several options
98 of the same kind; for example, if you specify @option{-L} more than once,
99 the directories are searched in the order specified.
101 Many options have long names starting with @samp{-f} or with
102 @samp{-W}---for example, @option{-fforce-mem},
103 @option{-fstrength-reduce}, @option{-Wformat} and so on. Most of
104 these have both positive and negative forms; the negative form of
105 @option{-ffoo} would be @option{-fno-foo}. This manual documents
106 only one of these two forms, whichever one is not the default.
110 @xref{Option Index}, for an index to GCC's options.
113 * Option Summary:: Brief list of all options, without explanations.
114 * Overall Options:: Controlling the kind of output:
115 an executable, object files, assembler files,
116 or preprocessed source.
117 * Invoking G++:: Compiling C++ programs.
118 * C Dialect Options:: Controlling the variant of C language compiled.
119 * C++ Dialect Options:: Variations on C++.
120 * Objective-C Dialect Options:: Variations on Objective-C.
121 * Language Independent Options:: Controlling how diagnostics should be
123 * Warning Options:: How picky should the compiler be?
124 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
125 * Optimize Options:: How much optimization?
126 * Preprocessor Options:: Controlling header files and macro definitions.
127 Also, getting dependency information for Make.
128 * Assembler Options:: Passing options to the assembler.
129 * Link Options:: Specifying libraries and so on.
130 * Directory Options:: Where to find header files and libraries.
131 Where to find the compiler executable files.
132 * Spec Files:: How to pass switches to sub-processes.
133 * Target Options:: Running a cross-compiler, or an old version of GCC.
134 * Submodel Options:: Specifying minor hardware or convention variations,
135 such as 68010 vs 68020.
136 * Code Gen Options:: Specifying conventions for function calls, data layout
138 * Environment Variables:: Env vars that affect GCC.
139 * Running Protoize:: Automatically adding or removing function prototypes.
145 @section Option Summary
147 Here is a summary of all the options, grouped by type. Explanations are
148 in the following sections.
151 @item Overall Options
152 @xref{Overall Options,,Options Controlling the Kind of Output}.
154 -c -S -E -o @var{file} -pipe -pass-exit-codes -x @var{language} @gol
155 -v --target-help --help}
157 @item C Language Options
158 @xref{C Dialect Options,,Options Controlling C Dialect}.
160 -ansi -std=@var{standard} -aux-info @var{filename} @gol
161 -fno-asm -fno-builtin @gol
162 -fhosted -ffreestanding @gol
163 -trigraphs -traditional -traditional-cpp @gol
164 -fallow-single-precision -fcond-mismatch @gol
165 -fsigned-bitfields -fsigned-char @gol
166 -funsigned-bitfields -funsigned-char @gol
167 -fwritable-strings -fshort-wchar}
169 @item C++ Language Options
170 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}.
172 -fno-access-control -fcheck-new -fconserve-space @gol
173 -fno-const-strings -fdollars-in-identifiers @gol
174 -fno-elide-constructors @gol
175 -fno-enforce-eh-specs -fexternal-templates @gol
176 -falt-external-templates @gol
177 -ffor-scope -fno-for-scope -fno-gnu-keywords @gol
178 -fno-implicit-templates @gol
179 -fno-implicit-inline-templates @gol
180 -fno-implement-inlines -fms-extensions @gol
181 -fno-nonansi-builtins -fno-operator-names @gol
182 -fno-optional-diags -fpermissive @gol
183 -frepo -fno-rtti -fstats -ftemplate-depth-@var{n} @gol
184 -fuse-cxa-atexit -fvtable-gc -fno-weak -nostdinc++ @gol
185 -fno-default-inline -Wctor-dtor-privacy @gol
186 -Wnon-virtual-dtor -Wreorder @gol
187 -Weffc++ -Wno-deprecated @gol
188 -Wno-non-template-friend -Wold-style-cast @gol
189 -Woverloaded-virtual -Wno-pmf-conversions @gol
190 -Wsign-promo -Wsynth}
192 @item Objective-C Language Options
193 @xref{Objective-C Dialect Options,,Options Controlling Objective-C Dialect}.
195 -fconstant-string-class=@var{class-name} @gol
196 -fgnu-runtime -fnext-runtime -gen-decls @gol
197 -Wno-protocol -Wselector}
199 @item Language Independent Options
200 @xref{Language Independent Options,,Options to Control Diagnostic Messages Formatting}.
202 -fmessage-length=@var{n} @gol
203 -fdiagnostics-show-location=@r{[}once@r{|}every-line@r{]}}
205 @item Warning Options
206 @xref{Warning Options,,Options to Request or Suppress Warnings}.
208 -fsyntax-only -pedantic -pedantic-errors @gol
209 -w -W -Wall -Waggregate-return @gol
210 -Wcast-align -Wcast-qual -Wchar-subscripts -Wcomment @gol
211 -Wconversion -Wdisabled-optimization -Werror @gol
212 -Wfloat-equal -Wformat -Wformat=2 @gol
213 -Wformat-nonliteral -Wformat-security @gol
214 -Wimplicit -Wimplicit-int @gol
215 -Wimplicit-function-declaration @gol
216 -Werror-implicit-function-declaration @gol
217 -Wimport -Winline @gol
218 -Wlarger-than-@var{len} -Wlong-long @gol
219 -Wmain -Wmissing-braces -Wmissing-declarations @gol
220 -Wmissing-format-attribute -Wmissing-noreturn @gol
221 -Wmultichar -Wno-format-extra-args -Wno-format-y2k @gol
222 -Wno-import -Wpacked -Wpadded @gol
223 -Wparentheses -Wpointer-arith -Wredundant-decls @gol
224 -Wreturn-type -Wsequence-point -Wshadow @gol
225 -Wsign-compare -Wswitch -Wsystem-headers @gol
226 -Wtrigraphs -Wundef -Wuninitialized @gol
227 -Wunknown-pragmas -Wunreachable-code @gol
228 -Wunused -Wunused-function -Wunused-label -Wunused-parameter @gol
229 -Wunused-value -Wunused-variable -Wwrite-strings}
231 @item C-only Warning Options
233 -Wbad-function-cast -Wmissing-prototypes -Wnested-externs @gol
234 -Wstrict-prototypes -Wtraditional}
236 @item Debugging Options
237 @xref{Debugging Options,,Options for Debugging Your Program or GCC}.
239 -a -ax -d@var{letters} -dumpspecs -dumpmachine -dumpversion @gol
240 -fdump-unnumbered -fdump-translation-unit@r{[}-@var{n}@r{]} @gol
241 -fdump-class-hierarchy@r{[}-@var{n}@r{]} @gol
242 -fdump-tree-original@r{[}-@var{n}@r{]} -fdump-tree-optimized@r{[}-@var{n}@r{]} @gol
243 -fdump-tree-inlined@r{[}-@var{n}@r{]} @gol
244 -fmem-report -fpretend-float @gol
245 -fprofile-arcs -ftest-coverage -ftime-report @gol
246 -g -g@var{level} -gcoff -gdwarf -gdwarf-1 -gdwarf-1+ -gdwarf-2 @gol
247 -ggdb -gstabs -gstabs+ -gxcoff -gxcoff+ @gol
248 -p -pg -print-file-name=@var{library} -print-libgcc-file-name @gol
249 -print-multi-directory -print-multi-lib @gol
250 -print-prog-name=@var{program} -print-search-dirs -Q @gol
253 @item Optimization Options
254 @xref{Optimize Options,,Options that Control Optimization}.
256 -falign-functions=@var{n} -falign-jumps=@var{n} @gol
257 -falign-labels=@var{n} -falign-loops=@var{n} @gol
258 -fbranch-probabilities -fcaller-saves @gol
259 -fcse-follow-jumps -fcse-skip-blocks -fdata-sections @gol
260 -fdelayed-branch -fdelete-null-pointer-checks @gol
261 -fexpensive-optimizations -ffast-math -ffloat-store @gol
262 -fforce-addr -fforce-mem -ffunction-sections @gol
263 -fgcse -fgcse-lm -fgcse-sm @gol
264 -finline-functions -finline-limit=@var{n} -fkeep-inline-functions @gol
265 -fkeep-static-consts -fmerge-constants -fmerge-all-constants @gol
266 -fmove-all-movables -fno-default-inline -fno-defer-pop @gol
267 -fno-function-cse -fno-guess-branch-probability @gol
268 -fno-inline -fno-math-errno -fno-peephole -fno-peephole2 @gol
269 -funsafe-math-optimizations -fno-trapping-math @gol
270 -fomit-frame-pointer -foptimize-register-move @gol
271 -foptimize-sibling-calls -freduce-all-givs @gol
272 -fregmove -frename-registers @gol
273 -frerun-cse-after-loop -frerun-loop-opt @gol
274 -fschedule-insns -fschedule-insns2 @gol
275 -fsingle-precision-constant -fssa -fssa-ccp -fssa-dce @gol
276 -fstrength-reduce -fstrict-aliasing -fthread-jumps -ftrapv @gol
277 -funroll-all-loops -funroll-loops @gol
278 --param @var{name}=@var{value}
279 -O -O0 -O1 -O2 -O3 -Os}
281 @item Preprocessor Options
282 @xref{Preprocessor Options,,Options Controlling the Preprocessor}.
284 -$ -A@var{question}=@var{answer} -A-@var{question}@r{[}=@var{answer}@r{]} @gol
285 -C -dD -dI -dM -dN @gol
286 -D@var{macro}@r{[}=@var{defn}@r{]} -E -H @gol
287 -idirafter @var{dir} @gol
288 -include @var{file} -imacros @var{file} @gol
289 -iprefix @var{file} -iwithprefix @var{dir} @gol
290 -iwithprefixbefore @var{dir} -isystem @var{dir} @gol
291 -M -MM -MF -MG -MP -MQ -MT -nostdinc -P -remap @gol
292 -trigraphs -undef -U@var{macro} -Wp,@var{option}}
294 @item Assembler Option
295 @xref{Assembler Options,,Passing Options to the Assembler}.
300 @xref{Link Options,,Options for Linking}.
302 @var{object-file-name} -l@var{library} @gol
303 -nostartfiles -nodefaultlibs -nostdlib @gol
304 -s -static -static-libgcc -shared -shared-libgcc -symbolic @gol
305 -Wl,@var{option} -Xlinker @var{option} @gol
308 @item Directory Options
309 @xref{Directory Options,,Options for Directory Search}.
311 -B@var{prefix} -I@var{dir} -I- -L@var{dir} -specs=@var{file}}
314 @c I wrote this xref this way to avoid overfull hbox. -- rms
315 @xref{Target Options}.
317 -b @var{machine} -V @var{version}}
319 @item Machine Dependent Options
320 @xref{Submodel Options,,Hardware Models and Configurations}.
321 @emph{M680x0 Options}
323 -m68000 -m68020 -m68020-40 -m68020-60 -m68030 -m68040 @gol
324 -m68060 -mcpu32 -m5200 -m68881 -mbitfield -mc68000 -mc68020 @gol
325 -mfpa -mnobitfield -mrtd -mshort -msoft-float -mpcrel @gol
326 -malign-int -mstrict-align}
328 @emph{M68hc1x Options}
330 -m6811 -m6812 -m68hc11 -m68hc12 @gol
331 -mauto-incdec -mshort -msoft-reg-count=@var{count}}
339 -mcpu=@var{cpu-type} @gol
340 -mtune=@var{cpu-type} @gol
341 -mcmodel=@var{code-model} @gol
343 -mapp-regs -mbroken-saverestore -mcypress @gol
344 -mepilogue -mfaster-structs -mflat @gol
345 -mfpu -mhard-float -mhard-quad-float @gol
346 -mimpure-text -mlive-g0 -mno-app-regs @gol
347 -mno-epilogue -mno-faster-structs -mno-flat -mno-fpu @gol
348 -mno-impure-text -mno-stack-bias -mno-unaligned-doubles @gol
349 -msoft-float -msoft-quad-float -msparclite -mstack-bias @gol
350 -msupersparc -munaligned-doubles -mv8}
352 @emph{Convex Options}
354 -mc1 -mc2 -mc32 -mc34 -mc38 @gol
355 -margcount -mnoargcount @gol
356 -mlong32 -mlong64 @gol
357 -mvolatile-cache -mvolatile-nocache}
359 @emph{AMD29K Options}
361 -m29000 -m29050 -mbw -mnbw -mdw -mndw @gol
362 -mlarge -mnormal -msmall @gol
363 -mkernel-registers -mno-reuse-arg-regs @gol
364 -mno-stack-check -mno-storem-bug @gol
365 -mreuse-arg-regs -msoft-float -mstack-check @gol
366 -mstorem-bug -muser-registers}
370 -mapcs-frame -mno-apcs-frame @gol
371 -mapcs-26 -mapcs-32 @gol
372 -mapcs-stack-check -mno-apcs-stack-check @gol
373 -mapcs-float -mno-apcs-float @gol
374 -mapcs-reentrant -mno-apcs-reentrant @gol
375 -msched-prolog -mno-sched-prolog @gol
376 -mlittle-endian -mbig-endian -mwords-little-endian @gol
377 -malignment-traps -mno-alignment-traps @gol
378 -msoft-float -mhard-float -mfpe @gol
379 -mthumb-interwork -mno-thumb-interwork @gol
380 -mcpu=@var{name} -march=@var{name} -mfpe=@var{name} @gol
381 -mstructure-size-boundary=@var{n} @gol
382 -mbsd -mxopen -mno-symrename @gol
383 -mabort-on-noreturn @gol
384 -mlong-calls -mno-long-calls @gol
385 -msingle-pic-base -mno-single-pic-base @gol
386 -mpic-register=@var{reg} @gol
387 -mnop-fun-dllimport @gol
388 -mpoke-function-name @gol
390 -mtpcs-frame -mtpcs-leaf-frame @gol
391 -mcaller-super-interworking -mcallee-super-interworking }
393 @emph{MN10200 Options}
397 @emph{MN10300 Options}
399 -mmult-bug -mno-mult-bug @gol
400 -mam33 -mno-am33 @gol
403 @emph{M32R/D Options}
405 -mcode-model=@var{model-type} -msdata=@var{sdata-type} @gol
410 -m88000 -m88100 -m88110 -mbig-pic @gol
411 -mcheck-zero-division -mhandle-large-shift @gol
412 -midentify-revision -mno-check-zero-division @gol
413 -mno-ocs-debug-info -mno-ocs-frame-position @gol
414 -mno-optimize-arg-area -mno-serialize-volatile @gol
415 -mno-underscores -mocs-debug-info @gol
416 -mocs-frame-position -moptimize-arg-area @gol
417 -mserialize-volatile -mshort-data-@var{num} -msvr3 @gol
418 -msvr4 -mtrap-large-shift -muse-div-instruction @gol
419 -mversion-03.00 -mwarn-passed-structs}
421 @emph{RS/6000 and PowerPC Options}
423 -mcpu=@var{cpu-type} @gol
424 -mtune=@var{cpu-type} @gol
425 -mpower -mno-power -mpower2 -mno-power2 @gol
426 -mpowerpc -mpowerpc64 -mno-powerpc @gol
427 -mpowerpc-gpopt -mno-powerpc-gpopt @gol
428 -mpowerpc-gfxopt -mno-powerpc-gfxopt @gol
429 -mnew-mnemonics -mold-mnemonics @gol
430 -mfull-toc -mminimal-toc -mno-fop-in-toc -mno-sum-in-toc @gol
431 -m64 -m32 -mxl-call -mno-xl-call -mthreads -mpe @gol
432 -msoft-float -mhard-float -mmultiple -mno-multiple @gol
433 -mstring -mno-string -mupdate -mno-update @gol
434 -mfused-madd -mno-fused-madd -mbit-align -mno-bit-align @gol
435 -mstrict-align -mno-strict-align -mrelocatable @gol
436 -mno-relocatable -mrelocatable-lib -mno-relocatable-lib @gol
437 -mtoc -mno-toc -mlittle -mlittle-endian -mbig -mbig-endian @gol
438 -mcall-aix -mcall-sysv -mcall-netbsd @gol
439 -mprototype -mno-prototype @gol
440 -msim -mmvme -mads -myellowknife -memb -msdata @gol
441 -msdata=@var{opt} -mvxworks -G @var{num}}
445 -mcall-lib-mul -mfp-arg-in-fpregs -mfp-arg-in-gregs @gol
446 -mfull-fp-blocks -mhc-struct-return -min-line-mul @gol
447 -mminimum-fp-blocks -mnohc-struct-return}
451 -mabicalls -march=@var{cpu-type} -mtune=@var{cpu=type} @gol
452 -mcpu=@var{cpu-type} -membedded-data -muninit-const-in-rodata @gol
453 -membedded-pic -mfp32 -mfp64 -mfused-madd -mno-fused-madd @gol
454 -mgas -mgp32 -mgp64 @gol
455 -mgpopt -mhalf-pic -mhard-float -mint64 -mips1 @gol
456 -mips2 -mips3 -mips4 -mlong64 -mlong32 -mlong-calls -mmemcpy @gol
457 -mmips-as -mmips-tfile -mno-abicalls @gol
458 -mno-embedded-data -mno-uninit-const-in-rodata @gol
459 -mno-embedded-pic -mno-gpopt -mno-long-calls @gol
460 -mno-memcpy -mno-mips-tfile -mno-rnames -mno-stats @gol
461 -mrnames -msoft-float @gol
462 -m4650 -msingle-float -mmad @gol
463 -mstats -EL -EB -G @var{num} -nocpp @gol
464 -mabi=32 -mabi=n32 -mabi=64 -mabi=eabi @gol
467 @emph{i386 and x86-64 Options}
469 -mcpu=@var{cpu-type} -march=@var{cpu-type} @gol
470 -mintel-syntax -mieee-fp -mno-fancy-math-387 @gol
471 -mno-fp-ret-in-387 -msoft-float -msvr3-shlib @gol
472 -mno-wide-multiply -mrtd -malign-double @gol
473 -mpreferred-stack-boundary=@var{num} @gol
474 -mmmx -msse -m3dnow @gol
475 -mthreads -mno-align-stringops -minline-all-stringops @gol
476 -mpush-args -maccumulate-outgoing-args -m128bit-long-double @gol
477 -m96bit-long-double -mregparm=@var{num} -momit-leaf-frame-pointer @gol
483 -march=@var{architecture-type} @gol
484 -mbig-switch -mdisable-fpregs -mdisable-indexing @gol
485 -mfast-indirect-calls -mgas -mjump-in-delay @gol
486 -mlong-load-store -mno-big-switch -mno-disable-fpregs @gol
487 -mno-disable-indexing -mno-fast-indirect-calls -mno-gas @gol
488 -mno-jump-in-delay -mno-long-load-store @gol
489 -mno-portable-runtime -mno-soft-float @gol
490 -mno-space-regs -msoft-float -mpa-risc-1-0 @gol
491 -mpa-risc-1-1 -mpa-risc-2-0 -mportable-runtime @gol
492 -mschedule=@var{cpu-type} -mspace-regs}
494 @emph{Intel 960 Options}
496 -m@var{cpu-type} -masm-compat -mclean-linkage @gol
497 -mcode-align -mcomplex-addr -mleaf-procedures @gol
498 -mic-compat -mic2.0-compat -mic3.0-compat @gol
499 -mintel-asm -mno-clean-linkage -mno-code-align @gol
500 -mno-complex-addr -mno-leaf-procedures @gol
501 -mno-old-align -mno-strict-align -mno-tail-call @gol
502 -mnumerics -mold-align -msoft-float -mstrict-align @gol
505 @emph{DEC Alpha Options}
507 -mfp-regs -mno-fp-regs -mno-soft-float -msoft-float @gol
508 -malpha-as -mgas @gol
509 -mieee -mieee-with-inexact -mieee-conformant @gol
510 -mfp-trap-mode=@var{mode} -mfp-rounding-mode=@var{mode} @gol
511 -mtrap-precision=@var{mode} -mbuild-constants @gol
512 -mcpu=@var{cpu-type} @gol
513 -mbwx -mno-bwx -mcix -mno-cix -mmax -mno-max @gol
514 -mmemory-latency=@var{time}}
516 @emph{Clipper Options}
520 @emph{H8/300 Options}
522 -mrelax -mh -ms -mint32 -malign-300}
526 -m1 -m2 -m3 -m3e @gol
527 -m4-nofpu -m4-single-only -m4-single -m4 @gol
528 -mb -ml -mdalign -mrelax @gol
529 -mbigtable -mfmovd -mhitachi -mnomacsave @gol
530 -mieee -misize -mpadstruct -mspace @gol
531 -mprefergot -musermode}
533 @emph{System V Options}
535 -Qy -Qn -YP,@var{paths} -Ym,@var{dir}}
540 -mmangle-cpu -mcpu=@var{cpu} -mtext=@var{text-section} @gol
541 -mdata=@var{data-section} -mrodata=@var{readonly-data-section}}
543 @emph{TMS320C3x/C4x Options}
545 -mcpu=@var{cpu} -mbig -msmall -mregparm -mmemparm @gol
546 -mfast-fix -mmpyi -mbk -mti -mdp-isr-reload @gol
547 -mrpts=@var{count} -mrptb -mdb -mloop-unsigned @gol
548 -mparallel-insns -mparallel-mpy -mpreserve-float}
552 -mlong-calls -mno-long-calls -mep -mno-ep @gol
553 -mprolog-function -mno-prolog-function -mspace @gol
554 -mtda=@var{n} -msda=@var{n} -mzda=@var{n} @gol
559 -m32032 -m32332 -m32532 -m32081 -m32381 @gol
560 -mmult-add -mnomult-add -msoft-float -mrtd -mnortd @gol
561 -mregparam -mnoregparam -msb -mnosb @gol
562 -mbitfield -mnobitfield -mhimem -mnohimem}
566 -mmcu=@var{mcu} -msize -minit-stack=@var{n} -mno-interrupts @gol
567 -mcall-prologues -mno-tablejump -mtiny-stack}
571 -mhardlit -mno-hardlit -mdiv -mno-div -mrelax-immediates @gol
572 -mno-relax-immediates -mwide-bitfields -mno-wide-bitfields @gol
573 -m4byte-functions -mno-4byte-functions -mcallgraph-data @gol
574 -mno-callgraph-data -mslow-bytes -mno-slow-bytes -mno-lsim @gol
575 -mlittle-endian -mbig-endian -m210 -m340 -mstack-increment}
579 -mbig-endian -mlittle-endian -mgnu-as -mgnu-ld -mno-pic @gol
580 -mvolatile-asm-stop -mb-step -mregister-names -mno-sdata @gol
581 -mconstant-gp -mauto-pic -minline-divide-min-latency @gol
582 -minline-divide-max-throughput -mno-dwarf2-asm @gol
583 -mfixed-range=@var{register-range}}
585 @emph{S/390 and zSeries Options}
587 -mhard-float -msoft-float -mbackchain -mno-backchain @gol
588 -msmall-exec -mno-small-exec -mmvcle -mno-mvcle @gol
589 -m64 -m31 -mdebug -mno-debug}
591 @item Code Generation Options
592 @xref{Code Gen Options,,Options for Code Generation Conventions}.
594 -fcall-saved-@var{reg} -fcall-used-@var{reg} @gol
595 -ffixed-@var{reg} -fexceptions @gol
596 -fnon-call-exceptions -funwind-tables @gol
597 -finhibit-size-directive -finstrument-functions @gol
598 -fcheck-memory-usage -fprefix-function-name @gol
599 -fno-common -fno-ident -fno-gnu-linker @gol
600 -fpcc-struct-return -fpic -fPIC @gol
601 -freg-struct-return -fshared-data -fshort-enums @gol
602 -fshort-double -fvolatile @gol
603 -fvolatile-global -fvolatile-static @gol
604 -fverbose-asm -fpack-struct -fstack-check @gol
605 -fstack-limit-register=@var{reg} -fstack-limit-symbol=@var{sym} @gol
606 -fargument-alias -fargument-noalias @gol
607 -fargument-noalias-global -fleading-underscore}
611 * Overall Options:: Controlling the kind of output:
612 an executable, object files, assembler files,
613 or preprocessed source.
614 * C Dialect Options:: Controlling the variant of C language compiled.
615 * C++ Dialect Options:: Variations on C++.
616 * Objective-C Dialect Options:: Variations on Objective-C.
617 * Language Independent Options:: Controlling how diagnostics should be
619 * Warning Options:: How picky should the compiler be?
620 * Debugging Options:: Symbol tables, measurements, and debugging dumps.
621 * Optimize Options:: How much optimization?
622 * Preprocessor Options:: Controlling header files and macro definitions.
623 Also, getting dependency information for Make.
624 * Assembler Options:: Passing options to the assembler.
625 * Link Options:: Specifying libraries and so on.
626 * Directory Options:: Where to find header files and libraries.
627 Where to find the compiler executable files.
628 * Spec Files:: How to pass switches to sub-processes.
629 * Target Options:: Running a cross-compiler, or an old version of GCC.
632 @node Overall Options
633 @section Options Controlling the Kind of Output
635 Compilation can involve up to four stages: preprocessing, compilation
636 proper, assembly and linking, always in that order. The first three
637 stages apply to an individual source file, and end by producing an
638 object file; linking combines all the object files (those newly
639 compiled, and those specified as input) into an executable file.
641 @cindex file name suffix
642 For any given input file, the file name suffix determines what kind of
647 C source code which must be preprocessed.
650 C source code which should not be preprocessed.
653 C++ source code which should not be preprocessed.
656 Objective-C source code. Note that you must link with the library
657 @file{libobjc.a} to make an Objective-C program work.
660 Objective-C source code which should not be preprocessed.
663 C header file (not to be compiled or linked).
667 @itemx @var{file}.cxx
668 @itemx @var{file}.cpp
669 @itemx @var{file}.c++
671 C++ source code which must be preprocessed. Note that in @samp{.cxx},
672 the last two letters must both be literally @samp{x}. Likewise,
673 @samp{.C} refers to a literal capital C@.
676 @itemx @var{file}.for
677 @itemx @var{file}.FOR
678 Fortran source code which should not be preprocessed.
681 @itemx @var{file}.fpp
682 @itemx @var{file}.FPP
683 Fortran source code which must be preprocessed (with the traditional
687 Fortran source code which must be preprocessed with a RATFOR
688 preprocessor (not included with GCC)@.
690 @xref{Overall Options,,Options Controlling the Kind of Output, g77,
691 Using and Porting GNU Fortran}, for more details of the handling of
694 @c FIXME: Descriptions of Java file types.
700 @c GCC also knows about some suffixes for languages not yet included:
710 @itemx @var{file}.chi
711 CHILL source code (preprocessed with the traditional preprocessor).
717 Assembler code which must be preprocessed.
720 An object file to be fed straight into linking.
721 Any file name with no recognized suffix is treated this way.
725 You can specify the input language explicitly with the @option{-x} option:
728 @item -x @var{language}
729 Specify explicitly the @var{language} for the following input files
730 (rather than letting the compiler choose a default based on the file
731 name suffix). This option applies to all following input files until
732 the next @option{-x} option. Possible values for @var{language} are:
734 c c-header cpp-output
736 objective-c objc-cpp-output
737 assembler assembler-with-cpp
738 f77 f77-cpp-input ratfor
741 @c Also f77-version, for internal use only.
744 Turn off any specification of a language, so that subsequent files are
745 handled according to their file name suffixes (as they are if @option{-x}
746 has not been used at all).
748 @item -pass-exit-codes
749 @opindex pass-exit-codes
750 Normally the @command{gcc} program will exit with the code of 1 if any
751 phase of the compiler returns a non-success return code. If you specify
752 @option{-pass-exit-codes}, the @command{gcc} program will instead return with
753 numerically highest error produced by any phase that returned an error
757 If you only want some of the stages of compilation, you can use
758 @option{-x} (or filename suffixes) to tell @command{gcc} where to start, and
759 one of the options @option{-c}, @option{-S}, or @option{-E} to say where
760 @command{gcc} is to stop. Note that some combinations (for example,
761 @samp{-x cpp-output -E}) instruct @command{gcc} to do nothing at all.
766 Compile or assemble the source files, but do not link. The linking
767 stage simply is not done. The ultimate output is in the form of an
768 object file for each source file.
770 By default, the object file name for a source file is made by replacing
771 the suffix @samp{.c}, @samp{.i}, @samp{.s}, etc., with @samp{.o}.
773 Unrecognized input files, not requiring compilation or assembly, are
778 Stop after the stage of compilation proper; do not assemble. The output
779 is in the form of an assembler code file for each non-assembler input
782 By default, the assembler file name for a source file is made by
783 replacing the suffix @samp{.c}, @samp{.i}, etc., with @samp{.s}.
785 Input files that don't require compilation are ignored.
789 Stop after the preprocessing stage; do not run the compiler proper. The
790 output is in the form of preprocessed source code, which is sent to the
793 Input files which don't require preprocessing are ignored.
795 @cindex output file option
798 Place output in file @var{file}. This applies regardless to whatever
799 sort of output is being produced, whether it be an executable file,
800 an object file, an assembler file or preprocessed C code.
802 Since only one output file can be specified, it does not make sense to
803 use @option{-o} when compiling more than one input file, unless you are
804 producing an executable file as output.
806 If @option{-o} is not specified, the default is to put an executable file
807 in @file{a.out}, the object file for @file{@var{source}.@var{suffix}} in
808 @file{@var{source}.o}, its assembler file in @file{@var{source}.s}, and
809 all preprocessed C source on standard output.
813 Print (on standard error output) the commands executed to run the stages
814 of compilation. Also print the version number of the compiler driver
815 program and of the preprocessor and the compiler proper.
819 Use pipes rather than temporary files for communication between the
820 various stages of compilation. This fails to work on some systems where
821 the assembler is unable to read from a pipe; but the GNU assembler has
826 Print (on the standard output) a description of the command line options
827 understood by @command{gcc}. If the @option{-v} option is also specified
828 then @option{--help} will also be passed on to the various processes
829 invoked by @command{gcc}, so that they can display the command line options
830 they accept. If the @option{-W} option is also specified then command
831 line options which have no documentation associated with them will also
836 Print (on the standard output) a description of target specific command
837 line options for each tool.
841 @section Compiling C++ Programs
843 @cindex suffixes for C++ source
844 @cindex C++ source file suffixes
845 C++ source files conventionally use one of the suffixes @samp{.C},
846 @samp{.cc}, @samp{.cpp}, @samp{.c++}, @samp{.cp}, or @samp{.cxx};
847 preprocessed C++ files use the suffix @samp{.ii}. GCC recognizes
848 files with these names and compiles them as C++ programs even if you
849 call the compiler the same way as for compiling C programs (usually with
850 the name @command{gcc}).
854 However, C++ programs often require class libraries as well as a
855 compiler that understands the C++ language---and under some
856 circumstances, you might want to compile programs from standard input,
857 or otherwise without a suffix that flags them as C++ programs.
858 @command{g++} is a program that calls GCC with the default language
859 set to C++, and automatically specifies linking against the C++
860 library. On many systems, @command{g++} is also
861 installed with the name @command{c++}.
863 @cindex invoking @command{g++}
864 When you compile C++ programs, you may specify many of the same
865 command-line options that you use for compiling programs in any
866 language; or command-line options meaningful for C and related
867 languages; or options that are meaningful only for C++ programs.
868 @xref{C Dialect Options,,Options Controlling C Dialect}, for
869 explanations of options for languages related to C@.
870 @xref{C++ Dialect Options,,Options Controlling C++ Dialect}, for
871 explanations of options that are meaningful only for C++ programs.
873 @node C Dialect Options
874 @section Options Controlling C Dialect
875 @cindex dialect options
876 @cindex language dialect options
877 @cindex options, dialect
879 The following options control the dialect of C (or languages derived
880 from C, such as C++ and Objective-C) that the compiler accepts:
887 In C mode, support all ISO C89 programs. In C++ mode,
888 remove GNU extensions that conflict with ISO C++.
890 This turns off certain features of GCC that are incompatible with ISO
891 C89 (when compiling C code), or of standard C++ (when compiling C++ code),
892 such as the @code{asm} and @code{typeof} keywords, and
893 predefined macros such as @code{unix} and @code{vax} that identify the
894 type of system you are using. It also enables the undesirable and
895 rarely used ISO trigraph feature. For the C compiler,
896 it disables recognition of C++ style @samp{//} comments as well as
897 the @code{inline} keyword.
899 The alternate keywords @code{__asm__}, @code{__extension__},
900 @code{__inline__} and @code{__typeof__} continue to work despite
901 @option{-ansi}. You would not want to use them in an ISO C program, of
902 course, but it is useful to put them in header files that might be included
903 in compilations done with @option{-ansi}. Alternate predefined macros
904 such as @code{__unix__} and @code{__vax__} are also available, with or
905 without @option{-ansi}.
907 The @option{-ansi} option does not cause non-ISO programs to be
908 rejected gratuitously. For that, @option{-pedantic} is required in
909 addition to @option{-ansi}. @xref{Warning Options}.
911 The macro @code{__STRICT_ANSI__} is predefined when the @option{-ansi}
912 option is used. Some header files may notice this macro and refrain
913 from declaring certain functions or defining certain macros that the
914 ISO standard doesn't call for; this is to avoid interfering with any
915 programs that might use these names for other things.
917 Functions which would normally be built in but do not have semantics
918 defined by ISO C (such as @code{alloca} and @code{ffs}) are not built-in
919 functions with @option{-ansi} is used. @xref{Other Builtins,,Other
920 built-in functions provided by GCC}, for details of the functions
925 Determine the language standard. This option is currently only
926 supported when compiling C@. A value for this option must be provided;
932 ISO C89 (same as @option{-ansi}).
935 ISO C89 as modified in amendment 1.
941 ISO C99. Note that this standard is not yet fully supported; see
942 @w{@uref{http://gcc.gnu.org/c99status.html}} for more information. The
943 names @samp{c9x} and @samp{iso9899:199x} are deprecated.
946 Default, ISO C89 plus GNU extensions (including some C99 features).
950 ISO C99 plus GNU extensions. When ISO C99 is fully implemented in GCC,
951 this will become the default. The name @samp{gnu9x} is deprecated.
955 Even when this option is not specified, you can still use some of the
956 features of newer standards in so far as they do not conflict with
957 previous C standards. For example, you may use @code{__restrict__} even
958 when @option{-std=c99} is not specified.
960 The @option{-std} options specifying some version of ISO C have the same
961 effects as @option{-ansi}, except that features that were not in ISO C89
962 but are in the specified version (for example, @samp{//} comments and
963 the @code{inline} keyword in ISO C99) are not disabled.
965 @xref{Standards,,Language Standards Supported by GCC}, for details of
966 these standard versions.
968 @item -aux-info @var{filename}
970 Output to the given filename prototyped declarations for all functions
971 declared and/or defined in a translation unit, including those in header
972 files. This option is silently ignored in any language other than C@.
974 Besides declarations, the file indicates, in comments, the origin of
975 each declaration (source file and line), whether the declaration was
976 implicit, prototyped or unprototyped (@samp{I}, @samp{N} for new or
977 @samp{O} for old, respectively, in the first character after the line
978 number and the colon), and whether it came from a declaration or a
979 definition (@samp{C} or @samp{F}, respectively, in the following
980 character). In the case of function definitions, a K&R-style list of
981 arguments followed by their declarations is also provided, inside
982 comments, after the declaration.
986 Do not recognize @code{asm}, @code{inline} or @code{typeof} as a
987 keyword, so that code can use these words as identifiers. You can use
988 the keywords @code{__asm__}, @code{__inline__} and @code{__typeof__}
989 instead. @option{-ansi} implies @option{-fno-asm}.
991 In C++, this switch only affects the @code{typeof} keyword, since
992 @code{asm} and @code{inline} are standard keywords. You may want to
993 use the @option{-fno-gnu-keywords} flag instead, which has the same
994 effect. In C99 mode (@option{-std=c99} or @option{-std=gnu99}), this
995 switch only affects the @code{asm} and @code{typeof} keywords, since
996 @code{inline} is a standard keyword in ISO C99.
1000 @cindex built-in functions
1001 Don't recognize built-in functions that do not begin with
1002 @samp{__builtin_} as prefix. @xref{Other Builtins,,Other built-in
1003 functions provided by GCC}, for details of the functions affected,
1004 including those which are not built-in functions when @option{-ansi} or
1005 @option{-std} options for strict ISO C conformance are used because they
1006 do not have an ISO standard meaning.
1008 GCC normally generates special code to handle certain built-in functions
1009 more efficiently; for instance, calls to @code{alloca} may become single
1010 instructions that adjust the stack directly, and calls to @code{memcpy}
1011 may become inline copy loops. The resulting code is often both smaller
1012 and faster, but since the function calls no longer appear as such, you
1013 cannot set a breakpoint on those calls, nor can you change the behavior
1014 of the functions by linking with a different library.
1016 In C++, @option{-fno-builtin} is always in effect. The @option{-fbuiltin}
1017 option has no effect. Therefore, in C++, the only way to get the
1018 optimization benefits of built-in functions is to call the function
1019 using the @samp{__builtin_} prefix. The GNU C++ Standard Library uses
1020 built-in functions to implement many functions (like
1021 @code{std::strchr}), so that you automatically get efficient code.
1025 @cindex hosted environment
1027 Assert that compilation takes place in a hosted environment. This implies
1028 @option{-fbuiltin}. A hosted environment is one in which the
1029 entire standard library is available, and in which @code{main} has a return
1030 type of @code{int}. Examples are nearly everything except a kernel.
1031 This is equivalent to @option{-fno-freestanding}.
1033 @item -ffreestanding
1034 @opindex ffreestanding
1035 @cindex hosted environment
1037 Assert that compilation takes place in a freestanding environment. This
1038 implies @option{-fno-builtin}. A freestanding environment
1039 is one in which the standard library may not exist, and program startup may
1040 not necessarily be at @code{main}. The most obvious example is an OS kernel.
1041 This is equivalent to @option{-fno-hosted}.
1043 @xref{Standards,,Language Standards Supported by GCC}, for details of
1044 freestanding and hosted environments.
1048 Support ISO C trigraphs. The @option{-ansi} option (and @option{-std}
1049 options for strict ISO C conformance) implies @option{-trigraphs}.
1051 @cindex traditional C language
1052 @cindex C language, traditional
1054 @opindex traditional
1055 Attempt to support some aspects of traditional C compilers.
1060 All @code{extern} declarations take effect globally even if they
1061 are written inside of a function definition. This includes implicit
1062 declarations of functions.
1065 The newer keywords @code{typeof}, @code{inline}, @code{signed}, @code{const}
1066 and @code{volatile} are not recognized. (You can still use the
1067 alternative keywords such as @code{__typeof__}, @code{__inline__}, and
1071 Comparisons between pointers and integers are always allowed.
1074 Integer types @code{unsigned short} and @code{unsigned char} promote
1075 to @code{unsigned int}.
1078 Out-of-range floating point literals are not an error.
1081 Certain constructs which ISO regards as a single invalid preprocessing
1082 number, such as @samp{0xe-0xd}, are treated as expressions instead.
1085 String ``constants'' are not necessarily constant; they are stored in
1086 writable space, and identical looking constants are allocated
1087 separately. (This is the same as the effect of
1088 @option{-fwritable-strings}.)
1090 @cindex @code{longjmp} and automatic variables
1092 All automatic variables not declared @code{register} are preserved by
1093 @code{longjmp}. Ordinarily, GNU C follows ISO C: automatic variables
1094 not declared @code{volatile} may be clobbered.
1099 @cindex escape sequences, traditional
1100 The character escape sequences @samp{\x} and @samp{\a} evaluate as the
1101 literal characters @samp{x} and @samp{a} respectively. Without
1102 @w{@option{-traditional}}, @samp{\x} is a prefix for the hexadecimal
1103 representation of a character, and @samp{\a} produces a bell.
1106 You may wish to use @option{-fno-builtin} as well as @option{-traditional}
1107 if your program uses names that are normally GNU C built-in functions for
1108 other purposes of its own.
1110 You cannot use @option{-traditional} if you include any header files that
1111 rely on ISO C features. Some vendors are starting to ship systems with
1112 ISO C header files and you cannot use @option{-traditional} on such
1113 systems to compile files that include any system headers.
1115 The @option{-traditional} option also enables @option{-traditional-cpp}.
1117 @item -traditional-cpp
1118 @opindex traditional-cpp
1119 Attempt to support some aspects of traditional C preprocessors.
1120 See the GNU CPP manual for details.
1122 @item -fcond-mismatch
1123 @opindex fcond-mismatch
1124 Allow conditional expressions with mismatched types in the second and
1125 third arguments. The value of such an expression is void. This option
1126 is not supported for C++.
1128 @item -funsigned-char
1129 @opindex funsigned-char
1130 Let the type @code{char} be unsigned, like @code{unsigned char}.
1132 Each kind of machine has a default for what @code{char} should
1133 be. It is either like @code{unsigned char} by default or like
1134 @code{signed char} by default.
1136 Ideally, a portable program should always use @code{signed char} or
1137 @code{unsigned char} when it depends on the signedness of an object.
1138 But many programs have been written to use plain @code{char} and
1139 expect it to be signed, or expect it to be unsigned, depending on the
1140 machines they were written for. This option, and its inverse, let you
1141 make such a program work with the opposite default.
1143 The type @code{char} is always a distinct type from each of
1144 @code{signed char} or @code{unsigned char}, even though its behavior
1145 is always just like one of those two.
1148 @opindex fsigned-char
1149 Let the type @code{char} be signed, like @code{signed char}.
1151 Note that this is equivalent to @option{-fno-unsigned-char}, which is
1152 the negative form of @option{-funsigned-char}. Likewise, the option
1153 @option{-fno-signed-char} is equivalent to @option{-funsigned-char}.
1155 @item -fsigned-bitfields
1156 @itemx -funsigned-bitfields
1157 @itemx -fno-signed-bitfields
1158 @itemx -fno-unsigned-bitfields
1159 @opindex fsigned-bitfields
1160 @opindex funsigned-bitfields
1161 @opindex fno-signed-bitfields
1162 @opindex fno-unsigned-bitfields
1163 These options control whether a bit-field is signed or unsigned, when the
1164 declaration does not use either @code{signed} or @code{unsigned}. By
1165 default, such a bit-field is signed, because this is consistent: the
1166 basic integer types such as @code{int} are signed types.
1168 However, when @option{-traditional} is used, bit-fields are all unsigned
1171 @item -fwritable-strings
1172 @opindex fwritable-strings
1173 Store string constants in the writable data segment and don't uniquize
1174 them. This is for compatibility with old programs which assume they can
1175 write into string constants. The option @option{-traditional} also has
1178 Writing into string constants is a very bad idea; ``constants'' should
1181 @item -fallow-single-precision
1182 @opindex fallow-single-precision
1183 Do not promote single precision math operations to double precision,
1184 even when compiling with @option{-traditional}.
1186 Traditional K&R C promotes all floating point operations to double
1187 precision, regardless of the sizes of the operands. On the
1188 architecture for which you are compiling, single precision may be faster
1189 than double precision. If you must use @option{-traditional}, but want
1190 to use single precision operations when the operands are single
1191 precision, use this option. This option has no effect when compiling
1192 with ISO or GNU C conventions (the default).
1195 @opindex fshort-wchar
1196 Override the underlying type for @samp{wchar_t} to be @samp{short
1197 unsigned int} instead of the default for the target. This option is
1198 useful for building programs to run under WINE@.
1201 @node C++ Dialect Options
1202 @section Options Controlling C++ Dialect
1204 @cindex compiler options, C++
1205 @cindex C++ options, command line
1206 @cindex options, C++
1207 This section describes the command-line options that are only meaningful
1208 for C++ programs; but you can also use most of the GNU compiler options
1209 regardless of what language your program is in. For example, you
1210 might compile a file @code{firstClass.C} like this:
1213 g++ -g -frepo -O -c firstClass.C
1217 In this example, only @option{-frepo} is an option meant
1218 only for C++ programs; you can use the other options with any
1219 language supported by GCC@.
1221 Here is a list of options that are @emph{only} for compiling C++ programs:
1224 @item -fno-access-control
1225 @opindex fno-access-control
1226 Turn off all access checking. This switch is mainly useful for working
1227 around bugs in the access control code.
1231 Check that the pointer returned by @code{operator new} is non-null
1232 before attempting to modify the storage allocated. The current Working
1233 Paper requires that @code{operator new} never return a null pointer, so
1234 this check is normally unnecessary.
1236 An alternative to using this option is to specify that your
1237 @code{operator new} does not throw any exceptions; if you declare it
1238 @samp{throw()}, G++ will check the return value. See also @samp{new
1241 @item -fconserve-space
1242 @opindex fconserve-space
1243 Put uninitialized or runtime-initialized global variables into the
1244 common segment, as C does. This saves space in the executable at the
1245 cost of not diagnosing duplicate definitions. If you compile with this
1246 flag and your program mysteriously crashes after @code{main()} has
1247 completed, you may have an object that is being destroyed twice because
1248 two definitions were merged.
1250 This option is no longer useful on most targets, now that support has
1251 been added for putting variables into BSS without making them common.
1253 @item -fno-const-strings
1254 @opindex fno-const-strings
1255 Give string constants type @code{char *} instead of type @code{const
1256 char *}. By default, G++ uses type @code{const char *} as required by
1257 the standard. Even if you use @option{-fno-const-strings}, you cannot
1258 actually modify the value of a string constant, unless you also use
1259 @option{-fwritable-strings}.
1261 This option might be removed in a future release of G++. For maximum
1262 portability, you should structure your code so that it works with
1263 string constants that have type @code{const char *}.
1265 @item -fdollars-in-identifiers
1266 @opindex fdollars-in-identifiers
1267 Accept @samp{$} in identifiers. You can also explicitly prohibit use of
1268 @samp{$} with the option @option{-fno-dollars-in-identifiers}. (GNU C allows
1269 @samp{$} by default on most target systems, but there are a few exceptions.)
1270 Traditional C allowed the character @samp{$} to form part of
1271 identifiers. However, ISO C and C++ forbid @samp{$} in identifiers.
1273 @item -fno-elide-constructors
1274 @opindex fno-elide-constructors
1275 The C++ standard allows an implementation to omit creating a temporary
1276 which is only used to initialize another object of the same type.
1277 Specifying this option disables that optimization, and forces G++ to
1278 call the copy constructor in all cases.
1280 @item -fno-enforce-eh-specs
1281 @opindex fno-enforce-eh-specs
1282 Don't check for violation of exception specifications at runtime. This
1283 option violates the C++ standard, but may be useful for reducing code
1284 size in production builds, much like defining @samp{NDEBUG}. The compiler
1285 will still optimize based on the exception specifications.
1287 @item -fexternal-templates
1288 @opindex fexternal-templates
1290 Cause @samp{#pragma interface} and @samp{implementation} to apply to
1291 template instantiation; template instances are emitted or not according
1292 to the location of the template definition. @xref{Template
1293 Instantiation}, for more information.
1295 This option is deprecated.
1297 @item -falt-external-templates
1298 @opindex falt-external-templates
1299 Similar to @option{-fexternal-templates}, but template instances are
1300 emitted or not according to the place where they are first instantiated.
1301 @xref{Template Instantiation}, for more information.
1303 This option is deprecated.
1306 @itemx -fno-for-scope
1308 @opindex fno-for-scope
1309 If @option{-ffor-scope} is specified, the scope of variables declared in
1310 a @i{for-init-statement} is limited to the @samp{for} loop itself,
1311 as specified by the C++ standard.
1312 If @option{-fno-for-scope} is specified, the scope of variables declared in
1313 a @i{for-init-statement} extends to the end of the enclosing scope,
1314 as was the case in old versions of G++, and other (traditional)
1315 implementations of C++.
1317 The default if neither flag is given to follow the standard,
1318 but to allow and give a warning for old-style code that would
1319 otherwise be invalid, or have different behavior.
1321 @item -fno-gnu-keywords
1322 @opindex fno-gnu-keywords
1323 Do not recognize @code{typeof} as a keyword, so that code can use this
1324 word as an identifier. You can use the keyword @code{__typeof__} instead.
1325 @option{-ansi} implies @option{-fno-gnu-keywords}.
1327 @item -fno-implicit-templates
1328 @opindex fno-implicit-templates
1329 Never emit code for non-inline templates which are instantiated
1330 implicitly (i.e.@: by use); only emit code for explicit instantiations.
1331 @xref{Template Instantiation}, for more information.
1333 @item -fno-implicit-inline-templates
1334 @opindex fno-implicit-inline-templates
1335 Don't emit code for implicit instantiations of inline templates, either.
1336 The default is to handle inlines differently so that compiles with and
1337 without optimization will need the same set of explicit instantiations.
1339 @item -fno-implement-inlines
1340 @opindex fno-implement-inlines
1341 To save space, do not emit out-of-line copies of inline functions
1342 controlled by @samp{#pragma implementation}. This will cause linker
1343 errors if these functions are not inlined everywhere they are called.
1345 @item -fms-extensions
1346 @opindex fms-extensions
1347 Disable pedantic warnings about constructs used in MFC, such as implicit
1348 int and getting a pointer to member function via non-standard syntax.
1350 @item -fno-nonansi-builtins
1351 @opindex fno-nonansi-builtins
1352 Disable built-in declarations of functions that are not mandated by
1353 ANSI/ISO C@. These include @code{ffs}, @code{alloca}, @code{_exit},
1354 @code{index}, @code{bzero}, @code{conjf}, and other related functions.
1356 @item -fno-operator-names
1357 @opindex fno-operator-names
1358 Do not treat the operator name keywords @code{and}, @code{bitand},
1359 @code{bitor}, @code{compl}, @code{not}, @code{or} and @code{xor} as
1360 synonyms as keywords.
1362 @item -fno-optional-diags
1363 @opindex fno-optional-diags
1364 Disable diagnostics that the standard says a compiler does not need to
1365 issue. Currently, the only such diagnostic issued by G++ is the one for
1366 a name having multiple meanings within a class.
1369 @opindex fpermissive
1370 Downgrade messages about nonconformant code from errors to warnings. By
1371 default, G++ effectively sets @option{-pedantic-errors} without
1372 @option{-pedantic}; this option reverses that. This behavior and this
1373 option are superseded by @option{-pedantic}, which works as it does for GNU C@.
1377 Enable automatic template instantiation at link time. This option also
1378 implies @option{-fno-implicit-templates}. @xref{Template
1379 Instantiation}, for more information.
1383 Disable generation of information about every class with virtual
1384 functions for use by the C++ runtime type identification features
1385 (@samp{dynamic_cast} and @samp{typeid}). If you don't use those parts
1386 of the language, you can save some space by using this flag. Note that
1387 exception handling uses the same information, but it will generate it as
1392 Emit statistics about front-end processing at the end of the compilation.
1393 This information is generally only useful to the G++ development team.
1395 @item -ftemplate-depth-@var{n}
1396 @opindex ftemplate-depth
1397 Set the maximum instantiation depth for template classes to @var{n}.
1398 A limit on the template instantiation depth is needed to detect
1399 endless recursions during template class instantiation. ANSI/ISO C++
1400 conforming programs must not rely on a maximum depth greater than 17.
1402 @item -fuse-cxa-atexit
1403 @opindex fuse-cxa-atexit
1404 Register destructors for objects with static storage duration with the
1405 @code{__cxa_atexit} function rather than the @code{atexit} function.
1406 This option is required for fully standards-compliant handling of static
1407 destructors, but will only work if your C library supports
1408 @code{__cxa_atexit}.
1412 Emit special relocations for vtables and virtual function references
1413 so that the linker can identify unused virtual functions and zero out
1414 vtable slots that refer to them. This is most useful with
1415 @option{-ffunction-sections} and @option{-Wl,--gc-sections}, in order to
1416 also discard the functions themselves.
1418 This optimization requires GNU as and GNU ld. Not all systems support
1419 this option. @option{-Wl,--gc-sections} is ignored without @option{-static}.
1423 Do not use weak symbol support, even if it is provided by the linker.
1424 By default, G++ will use weak symbols if they are available. This
1425 option exists only for testing, and should not be used by end-users;
1426 it will result in inferior code and has no benefits. This option may
1427 be removed in a future release of G++.
1431 Do not search for header files in the standard directories specific to
1432 C++, but do still search the other standard directories. (This option
1433 is used when building the C++ library.)
1436 In addition, these optimization, warning, and code generation options
1437 have meanings only for C++ programs:
1440 @item -fno-default-inline
1441 @opindex fno-default-inline
1442 Do not assume @samp{inline} for functions defined inside a class scope.
1443 @xref{Optimize Options,,Options That Control Optimization}. Note that these
1444 functions will have linkage like inline functions; they just won't be
1447 @item -Wctor-dtor-privacy @r{(C++ only)}
1448 @opindex Wctor-dtor-privacy
1449 Warn when a class seems unusable, because all the constructors or
1450 destructors in a class are private and the class has no friends or
1451 public static member functions.
1453 @item -Wnon-virtual-dtor @r{(C++ only)}
1454 @opindex Wnon-virtual-dtor
1455 Warn when a class declares a non-virtual destructor that should probably
1456 be virtual, because it looks like the class will be used polymorphically.
1458 @item -Wreorder @r{(C++ only)}
1460 @cindex reordering, warning
1461 @cindex warning for reordering of member initializers
1462 Warn when the order of member initializers given in the code does not
1463 match the order in which they must be executed. For instance:
1469 A(): j (0), i (1) @{ @}
1473 Here the compiler will warn that the member initializers for @samp{i}
1474 and @samp{j} will be rearranged to match the declaration order of the
1478 The following @option{-W@dots{}} options are not affected by @option{-Wall}.
1481 @item -Weffc++ @r{(C++ only)}
1483 Warn about violations of various style guidelines from Scott Meyers'
1484 @cite{Effective C++} books. If you use this option, you should be aware
1485 that the standard library headers do not obey all of these guidelines;
1486 you can use @samp{grep -v} to filter out those warnings.
1488 @item -Wno-deprecated @r{(C++ only)}
1489 @opindex Wno-deprecated
1490 Do not warn about usage of deprecated features. @xref{Deprecated Features}.
1492 @item -Wno-non-template-friend @r{(C++ only)}
1493 @opindex Wno-non-template-friend
1494 Disable warnings when non-templatized friend functions are declared
1495 within a template. With the advent of explicit template specification
1496 support in G++, if the name of the friend is an unqualified-id (i.e.,
1497 @samp{friend foo(int)}), the C++ language specification demands that the
1498 friend declare or define an ordinary, nontemplate function. (Section
1499 14.5.3). Before G++ implemented explicit specification, unqualified-ids
1500 could be interpreted as a particular specialization of a templatized
1501 function. Because this non-conforming behavior is no longer the default
1502 behavior for G++, @option{-Wnon-template-friend} allows the compiler to
1503 check existing code for potential trouble spots, and is on by default.
1504 This new compiler behavior can be turned off with
1505 @option{-Wno-non-template-friend} which keeps the conformant compiler code
1506 but disables the helpful warning.
1508 @item -Wold-style-cast @r{(C++ only)}
1509 @opindex Wold-style-cast
1510 Warn if an old-style (C-style) cast is used within a C++ program. The
1511 new-style casts (@samp{static_cast}, @samp{reinterpret_cast}, and
1512 @samp{const_cast}) are less vulnerable to unintended effects, and much
1515 @item -Woverloaded-virtual @r{(C++ only)}
1516 @opindex Woverloaded-virtual
1517 @cindex overloaded virtual fn, warning
1518 @cindex warning for overloaded virtual fn
1519 Warn when a function declaration hides virtual functions from a
1520 base class. For example, in:
1527 struct B: public A @{
1532 the @code{A} class version of @code{f} is hidden in @code{B}, and code
1540 will fail to compile.
1542 @item -Wno-pmf-conversions @r{(C++ only)}
1543 @opindex Wno-pmf-conversions
1544 Disable the diagnostic for converting a bound pointer to member function
1547 @item -Wsign-promo @r{(C++ only)}
1548 @opindex Wsign-promo
1549 Warn when overload resolution chooses a promotion from unsigned or
1550 enumeral type to a signed type over a conversion to an unsigned type of
1551 the same size. Previous versions of G++ would try to preserve
1552 unsignedness, but the standard mandates the current behavior.
1554 @item -Wsynth @r{(C++ only)}
1556 @cindex warning for synthesized methods
1557 @cindex synthesized methods, warning
1558 Warn when G++'s synthesis behavior does not match that of cfront. For
1564 A& operator = (int);
1574 In this example, G++ will synthesize a default @samp{A& operator =
1575 (const A&);}, while cfront will use the user-defined @samp{operator =}.
1578 @node Objective-C Dialect Options
1579 @section Options Controlling Objective-C Dialect
1581 @cindex compiler options, Objective-C
1582 @cindex Objective-C options, command line
1583 @cindex options, Objective-C
1584 This section describes the command-line options that are only meaningful
1585 for Objective-C programs; but you can also use most of the GNU compiler
1586 options regardless of what language your program is in. For example,
1587 you might compile a file @code{some_class.m} like this:
1590 gcc -g -fgnu-runtime -O -c some_class.m
1594 In this example, only @option{-fgnu-runtime} is an option meant only for
1595 Objective-C programs; you can use the other options with any language
1598 Here is a list of options that are @emph{only} for compiling Objective-C
1602 @item -fconstant-string-class=@var{class-name}
1603 @opindex fconstant-string-class
1604 Use @var{class-name} as the name of the class to instantiate for each
1605 literal string specified with the syntax @code{@@"@dots{}"}. The default
1606 class name is @code{NXConstantString}.
1609 @opindex fgnu-runtime
1610 Generate object code compatible with the standard GNU Objective-C
1611 runtime. This is the default for most types of systems.
1613 @item -fnext-runtime
1614 @opindex fnext-runtime
1615 Generate output compatible with the NeXT runtime. This is the default
1616 for NeXT-based systems, including Darwin and Mac OS X@.
1620 Dump interface declarations for all classes seen in the source file to a
1621 file named @file{@var{sourcename}.decl}.
1624 @opindex Wno-protocol
1625 Do not warn if methods required by a protocol are not implemented
1626 in the class adopting it.
1630 Warn if a selector has multiple methods of different types defined.
1632 @c not documented because only avail via -Wp
1633 @c @item -print-objc-runtime-info
1637 @node Language Independent Options
1638 @section Options to Control Diagnostic Messages Formatting
1639 @cindex options to control diagnostics formatting
1640 @cindex diagnostic messages
1641 @cindex message formatting
1643 Traditionally, diagnostic messages have been formatted irrespective of
1644 the output device's aspect (e.g.@: its width, @dots{}). The options described
1645 below can be used to control the diagnostic messages formatting
1646 algorithm, e.g.@: how many characters per line, how often source location
1647 information should be reported. Right now, only the C++ front end can
1648 honor these options. However it is expected, in the near future, that
1649 the remaining front ends would be able to digest them correctly.
1652 @item -fmessage-length=@var{n}
1653 @opindex fmessage-length
1654 Try to format error messages so that they fit on lines of about @var{n}
1655 characters. The default is 72 characters for @command{g++} and 0 for the rest of
1656 the front ends supported by GCC@. If @var{n} is zero, then no
1657 line-wrapping will be done; each error message will appear on a single
1660 @opindex fdiagnostics-show-location
1661 @item -fdiagnostics-show-location=once
1662 Only meaningful in line-wrapping mode. Instructs the diagnostic messages
1663 reporter to emit @emph{once} source location information; that is, in
1664 case the message is too long to fit on a single physical line and has to
1665 be wrapped, the source location won't be emitted (as prefix) again,
1666 over and over, in subsequent continuation lines. This is the default
1669 @item -fdiagnostics-show-location=every-line
1670 Only meaningful in line-wrapping mode. Instructs the diagnostic
1671 messages reporter to emit the same source location information (as
1672 prefix) for physical lines that result from the process of breaking
1673 a message which is too long to fit on a single line.
1677 @node Warning Options
1678 @section Options to Request or Suppress Warnings
1679 @cindex options to control warnings
1680 @cindex warning messages
1681 @cindex messages, warning
1682 @cindex suppressing warnings
1684 Warnings are diagnostic messages that report constructions which
1685 are not inherently erroneous but which are risky or suggest there
1686 may have been an error.
1688 You can request many specific warnings with options beginning @samp{-W},
1689 for example @option{-Wimplicit} to request warnings on implicit
1690 declarations. Each of these specific warning options also has a
1691 negative form beginning @samp{-Wno-} to turn off warnings;
1692 for example, @option{-Wno-implicit}. This manual lists only one of the
1693 two forms, whichever is not the default.
1695 These options control the amount and kinds of warnings produced by GCC:
1698 @cindex syntax checking
1700 @opindex fsyntax-only
1701 Check the code for syntax errors, but don't do anything beyond that.
1705 Issue all the warnings demanded by strict ISO C and ISO C++;
1706 reject all programs that use forbidden extensions, and some other
1707 programs that do not follow ISO C and ISO C++. For ISO C, follows the
1708 version of the ISO C standard specified by any @option{-std} option used.
1710 Valid ISO C and ISO C++ programs should compile properly with or without
1711 this option (though a rare few will require @option{-ansi} or a
1712 @option{-std} option specifying the required version of ISO C)@. However,
1713 without this option, certain GNU extensions and traditional C and C++
1714 features are supported as well. With this option, they are rejected.
1716 @option{-pedantic} does not cause warning messages for use of the
1717 alternate keywords whose names begin and end with @samp{__}. Pedantic
1718 warnings are also disabled in the expression that follows
1719 @code{__extension__}. However, only system header files should use
1720 these escape routes; application programs should avoid them.
1721 @xref{Alternate Keywords}.
1723 Some users try to use @option{-pedantic} to check programs for strict ISO
1724 C conformance. They soon find that it does not do quite what they want:
1725 it finds some non-ISO practices, but not all---only those for which
1726 ISO C @emph{requires} a diagnostic, and some others for which
1727 diagnostics have been added.
1729 A feature to report any failure to conform to ISO C might be useful in
1730 some instances, but would require considerable additional work and would
1731 be quite different from @option{-pedantic}. We don't have plans to
1732 support such a feature in the near future.
1734 Where the standard specified with @option{-std} represents a GNU
1735 extended dialect of C, such as @samp{gnu89} or @samp{gnu99}, there is a
1736 corresponding @dfn{base standard}, the version of ISO C on which the GNU
1737 extended dialect is based. Warnings from @option{-pedantic} are given
1738 where they are required by the base standard. (It would not make sense
1739 for such warnings to be given only for features not in the specified GNU
1740 C dialect, since by definition the GNU dialects of C include all
1741 features the compiler supports with the given option, and there would be
1742 nothing to warn about.)
1744 @item -pedantic-errors
1745 @opindex pedantic-errors
1746 Like @option{-pedantic}, except that errors are produced rather than
1751 Inhibit all warning messages.
1755 Inhibit warning messages about the use of @samp{#import}.
1757 @item -Wchar-subscripts
1758 @opindex Wchar-subscripts
1759 Warn if an array subscript has type @code{char}. This is a common cause
1760 of error, as programmers often forget that this type is signed on some
1765 Warn whenever a comment-start sequence @samp{/*} appears in a @samp{/*}
1766 comment, or whenever a Backslash-Newline appears in a @samp{//} comment.
1770 Check calls to @code{printf} and @code{scanf}, etc., to make sure that
1771 the arguments supplied have types appropriate to the format string
1772 specified, and that the conversions specified in the format string make
1773 sense. This includes standard functions, and others specified by format
1774 attributes (@pxref{Function Attributes}), in the @code{printf},
1775 @code{scanf}, @code{strftime} and @code{strfmon} (an X/Open extension,
1776 not in the C standard) families.
1778 The formats are checked against the format features supported by GNU
1779 libc version 2.2. These include all ISO C89 and C99 features, as well
1780 as features from the Single Unix Specification and some BSD and GNU
1781 extensions. Other library implementations may not support all these
1782 features; GCC does not support warning about features that go beyond a
1783 particular library's limitations. However, if @option{-pedantic} is used
1784 with @option{-Wformat}, warnings will be given about format features not
1785 in the selected standard version (but not for @code{strfmon} formats,
1786 since those are not in any version of the C standard). @xref{C Dialect
1787 Options,,Options Controlling C Dialect}.
1789 @option{-Wformat} is included in @option{-Wall}. For more control over some
1790 aspects of format checking, the options @option{-Wno-format-y2k},
1791 @option{-Wno-format-extra-args}, @option{-Wformat-nonliteral},
1792 @option{-Wformat-security} and @option{-Wformat=2} are available, but are
1793 not included in @option{-Wall}.
1795 @item -Wno-format-y2k
1796 @opindex Wno-format-y2k
1797 If @option{-Wformat} is specified, do not warn about @code{strftime}
1798 formats which may yield only a two-digit year.
1800 @item -Wno-format-extra-args
1801 @opindex Wno-format-extra-args
1802 If @option{-Wformat} is specified, do not warn about excess arguments to a
1803 @code{printf} or @code{scanf} format function. The C standard specifies
1804 that such arguments are ignored.
1806 @item -Wformat-nonliteral
1807 @opindex Wformat-nonliteral
1808 If @option{-Wformat} is specified, also warn if the format string is not a
1809 string literal and so cannot be checked, unless the format function
1810 takes its format arguments as a @code{va_list}.
1812 @item -Wformat-security
1813 @opindex Wformat-security
1814 If @option{-Wformat} is specified, also warn about uses of format
1815 functions that represent possible security problems. At present, this
1816 warns about calls to @code{printf} and @code{scanf} functions where the
1817 format string is not a string literal and there are no format arguments,
1818 as in @code{printf (foo);}. This may be a security hole if the format
1819 string came from untrusted input and contains @samp{%n}. (This is
1820 currently a subset of what @option{-Wformat-nonliteral} warns about, but
1821 in future warnings may be added to @option{-Wformat-security} that are not
1822 included in @option{-Wformat-nonliteral}.)
1826 Enable @option{-Wformat} plus format checks not included in
1827 @option{-Wformat}. Currently equivalent to @samp{-Wformat
1828 -Wformat-nonliteral -Wformat-security}.
1830 @item -Wimplicit-int
1831 @opindex Wimplicit-int
1832 Warn when a declaration does not specify a type.
1834 @item -Wimplicit-function-declaration
1835 @itemx -Werror-implicit-function-declaration
1836 @opindex Wimplicit-function-declaration
1837 @opindex Werror-implicit-function-declaration
1838 Give a warning (or error) whenever a function is used before being
1843 Same as @option{-Wimplicit-int} and @option{-Wimplicit-function-declaration}.
1847 Warn if the type of @samp{main} is suspicious. @samp{main} should be a
1848 function with external linkage, returning int, taking either zero
1849 arguments, two, or three arguments of appropriate types.
1851 @item -Wmissing-braces
1852 @opindex Wmissing-braces
1853 Warn if an aggregate or union initializer is not fully bracketed. In
1854 the following example, the initializer for @samp{a} is not fully
1855 bracketed, but that for @samp{b} is fully bracketed.
1858 int a[2][2] = @{ 0, 1, 2, 3 @};
1859 int b[2][2] = @{ @{ 0, 1 @}, @{ 2, 3 @} @};
1864 Warn if a multicharacter constant (@samp{'FOOF'}) is used. Usually they
1865 indicate a typo in the user's code, as they have implementation-defined
1866 values, and should not be used in portable code.
1869 @opindex Wparentheses
1870 Warn if parentheses are omitted in certain contexts, such
1871 as when there is an assignment in a context where a truth value
1872 is expected, or when operators are nested whose precedence people
1873 often get confused about.
1875 Also warn about constructions where there may be confusion to which
1876 @code{if} statement an @code{else} branch belongs. Here is an example of
1891 In C, every @code{else} branch belongs to the innermost possible @code{if}
1892 statement, which in this example is @code{if (b)}. This is often not
1893 what the programmer expected, as illustrated in the above example by
1894 indentation the programmer chose. When there is the potential for this
1895 confusion, GCC will issue a warning when this flag is specified.
1896 To eliminate the warning, add explicit braces around the innermost
1897 @code{if} statement so there is no way the @code{else} could belong to
1898 the enclosing @code{if}. The resulting code would look like this:
1914 @item -Wsequence-point
1915 @opindex Wsequence-point
1916 Warn about code that may have undefined semantics because of violations
1917 of sequence point rules in the C standard.
1919 The C standard defines the order in which expressions in a C program are
1920 evaluated in terms of @dfn{sequence points}, which represent a partial
1921 ordering between the execution of parts of the program: those executed
1922 before the sequence point, and those executed after it. These occur
1923 after the evaluation of a full expression (one which is not part of a
1924 larger expression), after the evaluation of the first operand of a
1925 @code{&&}, @code{||}, @code{? :} or @code{,} (comma) operator, before a
1926 function is called (but after the evaluation of its arguments and the
1927 expression denoting the called function), and in certain other places.
1928 Other than as expressed by the sequence point rules, the order of
1929 evaluation of subexpressions of an expression is not specified. All
1930 these rules describe only a partial order rather than a total order,
1931 since, for example, if two functions are called within one expression
1932 with no sequence point between them, the order in which the functions
1933 are called is not specified. However, the standards committee have
1934 ruled that function calls do not overlap.
1936 It is not specified when between sequence points modifications to the
1937 values of objects take effect. Programs whose behavior depends on this
1938 have undefined behavior; the C standard specifies that ``Between the
1939 previous and next sequence point an object shall have its stored value
1940 modified at most once by the evaluation of an expression. Furthermore,
1941 the prior value shall be read only to determine the value to be
1942 stored.''. If a program breaks these rules, the results on any
1943 particular implementation are entirely unpredictable.
1945 Examples of code with undefined behavior are @code{a = a++;}, @code{a[n]
1946 = b[n++]} and @code{a[i++] = i;}. Some more complicated cases are not
1947 diagnosed by this option, and it may give an occasional false positive
1948 result, but in general it has been found fairly effective at detecting
1949 this sort of problem in programs.
1951 The present implementation of this option only works for C programs. A
1952 future implementation may also work for C++ programs.
1954 The C standard is worded confusingly, therefore there is some debate
1955 over the precise meaning of the sequence point rules in subtle cases.
1956 Links to discussions of the problem, including proposed formal
1957 definitions, may be found on our readings page, at
1958 @w{@uref{http://gcc.gnu.org/readings.html}}.
1961 @opindex Wreturn-type
1962 Warn whenever a function is defined with a return-type that defaults to
1963 @code{int}. Also warn about any @code{return} statement with no
1964 return-value in a function whose return-type is not @code{void}.
1966 For C++, a function without return type always produces a diagnostic
1967 message, even when @option{-Wno-return-type} is specified. The only
1968 exceptions are @samp{main} and functions defined in system headers.
1972 Warn whenever a @code{switch} statement has an index of enumeral type
1973 and lacks a @code{case} for one or more of the named codes of that
1974 enumeration. (The presence of a @code{default} label prevents this
1975 warning.) @code{case} labels outside the enumeration range also
1976 provoke warnings when this option is used.
1980 Warn if any trigraphs are encountered that might change the meaning of
1981 the program (trigraphs within comments are not warned about).
1983 @item -Wunused-function
1984 @opindex Wunused-function
1985 Warn whenever a static function is declared but not defined or a
1986 non\-inline static function is unused.
1988 @item -Wunused-label
1989 @opindex Wunused-label
1990 Warn whenever a label is declared but not used.
1992 To suppress this warning use the @samp{unused} attribute
1993 (@pxref{Variable Attributes}).
1995 @item -Wunused-parameter
1996 @opindex Wunused-parameter
1997 Warn whenever a function parameter is unused aside from its declaration.
1999 To suppress this warning use the @samp{unused} attribute
2000 (@pxref{Variable Attributes}).
2002 @item -Wunused-variable
2003 @opindex Wunused-variable
2004 Warn whenever a local variable or non-constant static variable is unused
2005 aside from its declaration
2007 To suppress this warning use the @samp{unused} attribute
2008 (@pxref{Variable Attributes}).
2010 @item -Wunused-value
2011 @opindex Wunused-value
2012 Warn whenever a statement computes a result that is explicitly not used.
2014 To suppress this warning cast the expression to @samp{void}.
2018 All all the above @option{-Wunused} options combined.
2020 In order to get a warning about an unused function parameter, you must
2021 either specify @samp{-W -Wunused} or separately specify
2022 @option{-Wunused-parameter}.
2024 @item -Wuninitialized
2025 @opindex Wuninitialized
2026 Warn if an automatic variable is used without first being initialized or
2027 if a variable may be clobbered by a @code{setjmp} call.
2029 These warnings are possible only in optimizing compilation,
2030 because they require data flow information that is computed only
2031 when optimizing. If you don't specify @option{-O}, you simply won't
2034 These warnings occur only for variables that are candidates for
2035 register allocation. Therefore, they do not occur for a variable that
2036 is declared @code{volatile}, or whose address is taken, or whose size
2037 is other than 1, 2, 4 or 8 bytes. Also, they do not occur for
2038 structures, unions or arrays, even when they are in registers.
2040 Note that there may be no warning about a variable that is used only
2041 to compute a value that itself is never used, because such
2042 computations may be deleted by data flow analysis before the warnings
2045 These warnings are made optional because GCC is not smart
2046 enough to see all the reasons why the code might be correct
2047 despite appearing to have an error. Here is one example of how
2068 If the value of @code{y} is always 1, 2 or 3, then @code{x} is
2069 always initialized, but GCC doesn't know this. Here is
2070 another common case:
2075 if (change_y) save_y = y, y = new_y;
2077 if (change_y) y = save_y;
2082 This has no bug because @code{save_y} is used only if it is set.
2084 @cindex @code{longjmp} warnings
2085 This option also warns when a non-volatile automatic variable might be
2086 changed by a call to @code{longjmp}. These warnings as well are possible
2087 only in optimizing compilation.
2089 The compiler sees only the calls to @code{setjmp}. It cannot know
2090 where @code{longjmp} will be called; in fact, a signal handler could
2091 call it at any point in the code. As a result, you may get a warning
2092 even when there is in fact no problem because @code{longjmp} cannot
2093 in fact be called at the place which would cause a problem.
2095 Some spurious warnings can be avoided if you declare all the functions
2096 you use that never return as @code{noreturn}. @xref{Function
2099 @item -Wreorder @r{(C++ only)}
2101 @cindex reordering, warning
2102 @cindex warning for reordering of member initializers
2103 Warn when the order of member initializers given in the code does not
2104 match the order in which they must be executed. For instance:
2106 @item -Wunknown-pragmas
2107 @opindex Wunknown-pragmas
2108 @cindex warning for unknown pragmas
2109 @cindex unknown pragmas, warning
2110 @cindex pragmas, warning of unknown
2111 Warn when a #pragma directive is encountered which is not understood by
2112 GCC@. If this command line option is used, warnings will even be issued
2113 for unknown pragmas in system header files. This is not the case if
2114 the warnings were only enabled by the @option{-Wall} command line option.
2118 All of the above @samp{-W} options combined. This enables all the
2119 warnings about constructions that some users consider questionable, and
2120 that are easy to avoid (or modify to prevent the warning), even in
2121 conjunction with macros.
2123 @item -Wsystem-headers
2124 @opindex Wsystem-headers
2125 @cindex warnings from system headers
2126 @cindex system headers, warnings from
2127 Print warning messages for constructs found in system header files.
2128 Warnings from system headers are normally suppressed, on the assumption
2129 that they usually do not indicate real problems and would only make the
2130 compiler output harder to read. Using this command line option tells
2131 GCC to emit warnings from system headers as if they occurred in user
2132 code. However, note that using @option{-Wall} in conjunction with this
2133 option will @emph{not} warn about unknown pragmas in system
2134 headers---for that, @option{-Wunknown-pragmas} must also be used.
2137 The following @option{-W@dots{}} options are not implied by @option{-Wall}.
2138 Some of them warn about constructions that users generally do not
2139 consider questionable, but which occasionally you might wish to check
2140 for; others warn about constructions that are necessary or hard to avoid
2141 in some cases, and there is no simple way to modify the code to suppress
2147 Print extra warning messages for these events:
2151 A function can return either with or without a value. (Falling
2152 off the end of the function body is considered returning without
2153 a value.) For example, this function would evoke such a
2167 An expression-statement or the left-hand side of a comma expression
2168 contains no side effects.
2169 To suppress the warning, cast the unused expression to void.
2170 For example, an expression such as @samp{x[i,j]} will cause a warning,
2171 but @samp{x[(void)i,j]} will not.
2174 An unsigned value is compared against zero with @samp{<} or @samp{<=}.
2177 A comparison like @samp{x<=y<=z} appears; this is equivalent to
2178 @samp{(x<=y ? 1 : 0) <= z}, which is a different interpretation from
2179 that of ordinary mathematical notation.
2182 Storage-class specifiers like @code{static} are not the first things in
2183 a declaration. According to the C Standard, this usage is obsolescent.
2186 The return type of a function has a type qualifier such as @code{const}.
2187 Such a type qualifier has no effect, since the value returned by a
2188 function is not an lvalue. (But don't warn about the GNU extension of
2189 @code{volatile void} return types. That extension will be warned about
2190 if @option{-pedantic} is specified.)
2193 If @option{-Wall} or @option{-Wunused} is also specified, warn about unused
2197 A comparison between signed and unsigned values could produce an
2198 incorrect result when the signed value is converted to unsigned.
2199 (But don't warn if @option{-Wno-sign-compare} is also specified.)
2202 An aggregate has a partly bracketed initializer.
2203 For example, the following code would evoke such a warning,
2204 because braces are missing around the initializer for @code{x.h}:
2207 struct s @{ int f, g; @};
2208 struct t @{ struct s h; int i; @};
2209 struct t x = @{ 1, 2, 3 @};
2213 An aggregate has an initializer which does not initialize all members.
2214 For example, the following code would cause such a warning, because
2215 @code{x.h} would be implicitly initialized to zero:
2218 struct s @{ int f, g, h; @};
2219 struct s x = @{ 3, 4 @};
2224 @opindex Wfloat-equal
2225 Warn if floating point values are used in equality comparisons.
2227 The idea behind this is that sometimes it is convenient (for the
2228 programmer) to consider floating-point values as approximations to
2229 infinitely precise real numbers. If you are doing this, then you need
2230 to compute (by analysing the code, or in some other way) the maximum or
2231 likely maximum error that the computation introduces, and allow for it
2232 when performing comparisons (and when producing output, but that's a
2233 different problem). In particular, instead of testing for equality, you
2234 would check to see whether the two values have ranges that overlap; and
2235 this is done with the relational operators, so equality comparisons are
2238 @item -Wtraditional @r{(C only)}
2239 @opindex Wtraditional
2240 Warn about certain constructs that behave differently in traditional and
2241 ISO C@. Also warn about ISO C constructs that have no traditional C
2242 equivalent, and/or problematic constructs which should be avoided.
2246 Macro parameters that appear within string literals in the macro body.
2247 In traditional C macro replacement takes place within string literals,
2248 but does not in ISO C@.
2251 In traditional C, some preprocessor directives did not exist.
2252 Traditional preprocessors would only consider a line to be a directive
2253 if the @samp{#} appeared in column 1 on the line. Therefore
2254 @option{-Wtraditional} warns about directives that traditional C
2255 understands but would ignore because the @samp{#} does not appear as the
2256 first character on the line. It also suggests you hide directives like
2257 @samp{#pragma} not understood by traditional C by indenting them. Some
2258 traditional implementations would not recognise @samp{#elif}, so it
2259 suggests avoiding it altogether.
2262 A function-like macro that appears without arguments.
2265 The unary plus operator.
2268 The @samp{U} integer constant suffix, or the @samp{F} or @samp{L} floating point
2269 constant suffixes. (Traditional C does support the @samp{L} suffix on integer
2270 constants.) Note, these suffixes appear in macros defined in the system
2271 headers of most modern systems, e.g.@: the @samp{_MIN}/@samp{_MAX} macros in @code{<limits.h>}.
2272 Use of these macros in user code might normally lead to spurious
2273 warnings, however gcc's integrated preprocessor has enough context to
2274 avoid warning in these cases.
2277 A function declared external in one block and then used after the end of
2281 A @code{switch} statement has an operand of type @code{long}.
2284 A non-@code{static} function declaration follows a @code{static} one.
2285 This construct is not accepted by some traditional C compilers.
2288 The ISO type of an integer constant has a different width or
2289 signedness from its traditional type. This warning is only issued if
2290 the base of the constant is ten. I.e.@: hexadecimal or octal values, which
2291 typically represent bit patterns, are not warned about.
2294 Usage of ISO string concatenation is detected.
2297 Initialization of automatic aggregates.
2300 Identifier conflicts with labels. Traditional C lacks a separate
2301 namespace for labels.
2304 Initialization of unions. If the initializer is zero, the warning is
2305 omitted. This is done under the assumption that the zero initializer in
2306 user code appears conditioned on e.g.@: @code{__STDC__} to avoid missing
2307 initializer warnings and relies on default initialization to zero in the
2311 Conversions by prototypes between fixed/floating point values and vice
2312 versa. The absence of these prototypes when compiling with traditional
2313 C would cause serious problems. This is a subset of the possible
2314 conversion warnings, for the full set use @option{-Wconversion}.
2319 Warn if an undefined identifier is evaluated in an @samp{#if} directive.
2323 Warn whenever a local variable shadows another local variable, parameter or
2324 global variable or whenever a built-in function is shadowed.
2326 @item -Wlarger-than-@var{len}
2327 @opindex Wlarger-than
2328 Warn whenever an object of larger than @var{len} bytes is defined.
2330 @item -Wpointer-arith
2331 @opindex Wpointer-arith
2332 Warn about anything that depends on the ``size of'' a function type or
2333 of @code{void}. GNU C assigns these types a size of 1, for
2334 convenience in calculations with @code{void *} pointers and pointers
2337 @item -Wbad-function-cast @r{(C only)}
2338 @opindex Wbad-function-cast
2339 Warn whenever a function call is cast to a non-matching type.
2340 For example, warn if @code{int malloc()} is cast to @code{anything *}.
2344 Warn whenever a pointer is cast so as to remove a type qualifier from
2345 the target type. For example, warn if a @code{const char *} is cast
2346 to an ordinary @code{char *}.
2349 @opindex Wcast-align
2350 Warn whenever a pointer is cast such that the required alignment of the
2351 target is increased. For example, warn if a @code{char *} is cast to
2352 an @code{int *} on machines where integers can only be accessed at
2353 two- or four-byte boundaries.
2355 @item -Wwrite-strings
2356 @opindex Wwrite-strings
2357 When compiling C, give string constants the type @code{const
2358 char[@var{length}]} so that
2359 copying the address of one into a non-@code{const} @code{char *}
2360 pointer will get a warning; when compiling C++, warn about the
2361 deprecated conversion from string constants to @code{char *}.
2362 These warnings will help you find at
2363 compile time code that can try to write into a string constant, but
2364 only if you have been very careful about using @code{const} in
2365 declarations and prototypes. Otherwise, it will just be a nuisance;
2366 this is why we did not make @option{-Wall} request these warnings.
2369 @opindex Wconversion
2370 Warn if a prototype causes a type conversion that is different from what
2371 would happen to the same argument in the absence of a prototype. This
2372 includes conversions of fixed point to floating and vice versa, and
2373 conversions changing the width or signedness of a fixed point argument
2374 except when the same as the default promotion.
2376 Also, warn if a negative integer constant expression is implicitly
2377 converted to an unsigned type. For example, warn about the assignment
2378 @code{x = -1} if @code{x} is unsigned. But do not warn about explicit
2379 casts like @code{(unsigned) -1}.
2381 @item -Wsign-compare
2382 @opindex Wsign-compare
2383 @cindex warning for comparison of signed and unsigned values
2384 @cindex comparison of signed and unsigned values, warning
2385 @cindex signed and unsigned values, comparison warning
2386 Warn when a comparison between signed and unsigned values could produce
2387 an incorrect result when the signed value is converted to unsigned.
2388 This warning is also enabled by @option{-W}; to get the other warnings
2389 of @option{-W} without this warning, use @samp{-W -Wno-sign-compare}.
2391 @item -Waggregate-return
2392 @opindex Waggregate-return
2393 Warn if any functions that return structures or unions are defined or
2394 called. (In languages where you can return an array, this also elicits
2397 @item -Wstrict-prototypes @r{(C only)}
2398 @opindex Wstrict-prototypes
2399 Warn if a function is declared or defined without specifying the
2400 argument types. (An old-style function definition is permitted without
2401 a warning if preceded by a declaration which specifies the argument
2404 @item -Wmissing-prototypes @r{(C only)}
2405 @opindex Wmissing-prototypes
2406 Warn if a global function is defined without a previous prototype
2407 declaration. This warning is issued even if the definition itself
2408 provides a prototype. The aim is to detect global functions that fail
2409 to be declared in header files.
2411 @item -Wmissing-declarations
2412 @opindex Wmissing-declarations
2413 Warn if a global function is defined without a previous declaration.
2414 Do so even if the definition itself provides a prototype.
2415 Use this option to detect global functions that are not declared in
2418 @item -Wmissing-noreturn
2419 @opindex Wmissing-noreturn
2420 Warn about functions which might be candidates for attribute @code{noreturn}.
2421 Note these are only possible candidates, not absolute ones. Care should
2422 be taken to manually verify functions actually do not ever return before
2423 adding the @code{noreturn} attribute, otherwise subtle code generation
2424 bugs could be introduced. You will not get a warning for @code{main} in
2425 hosted C environments.
2427 @item -Wmissing-format-attribute
2428 @opindex Wmissing-format-attribute
2430 If @option{-Wformat} is enabled, also warn about functions which might be
2431 candidates for @code{format} attributes. Note these are only possible
2432 candidates, not absolute ones. GCC will guess that @code{format}
2433 attributes might be appropriate for any function that calls a function
2434 like @code{vprintf} or @code{vscanf}, but this might not always be the
2435 case, and some functions for which @code{format} attributes are
2436 appropriate may not be detected. This option has no effect unless
2437 @option{-Wformat} is enabled (possibly by @option{-Wall}).
2441 Warn if a structure is given the packed attribute, but the packed
2442 attribute has no effect on the layout or size of the structure.
2443 Such structures may be mis-aligned for little benefit. For
2444 instance, in this code, the variable @code{f.x} in @code{struct bar}
2445 will be misaligned even though @code{struct bar} does not itself
2446 have the packed attribute:
2453 @} __attribute__((packed));
2463 Warn if padding is included in a structure, either to align an element
2464 of the structure or to align the whole structure. Sometimes when this
2465 happens it is possible to rearrange the fields of the structure to
2466 reduce the padding and so make the structure smaller.
2468 @item -Wredundant-decls
2469 @opindex Wredundant-decls
2470 Warn if anything is declared more than once in the same scope, even in
2471 cases where multiple declaration is valid and changes nothing.
2473 @item -Wnested-externs @r{(C only)}
2474 @opindex Wnested-externs
2475 Warn if an @code{extern} declaration is encountered within a function.
2477 @item -Wunreachable-code
2478 @opindex Wunreachable-code
2479 Warn if the compiler detects that code will never be executed.
2481 This option is intended to warn when the compiler detects that at
2482 least a whole line of source code will never be executed, because
2483 some condition is never satisfied or because it is after a
2484 procedure that never returns.
2486 It is possible for this option to produce a warning even though there
2487 are circumstances under which part of the affected line can be executed,
2488 so care should be taken when removing apparently-unreachable code.
2490 For instance, when a function is inlined, a warning may mean that the
2491 line is unreachable in only one inlined copy of the function.
2493 This option is not made part of @option{-Wall} because in a debugging
2494 version of a program there is often substantial code which checks
2495 correct functioning of the program and is, hopefully, unreachable
2496 because the program does work. Another common use of unreachable
2497 code is to provide behaviour which is selectable at compile-time.
2501 Warn if a function can not be inlined and it was declared as inline.
2505 @opindex Wno-long-long
2506 Warn if @samp{long long} type is used. This is default. To inhibit
2507 the warning messages, use @option{-Wno-long-long}. Flags
2508 @option{-Wlong-long} and @option{-Wno-long-long} are taken into account
2509 only when @option{-pedantic} flag is used.
2511 @item -Wdisabled-optimization
2512 @opindex Wdisabled-optimization
2513 Warn if a requested optimization pass is disabled. This warning does
2514 not generally indicate that there is anything wrong with your code; it
2515 merely indicates that GCC's optimizers were unable to handle the code
2516 effectively. Often, the problem is that your code is too big or too
2517 complex; GCC will refuse to optimize programs when the optimization
2518 itself is likely to take inordinate amounts of time.
2522 Make all warnings into errors.
2525 @node Debugging Options
2526 @section Options for Debugging Your Program or GCC
2527 @cindex options, debugging
2528 @cindex debugging information options
2530 GCC has various special options that are used for debugging
2531 either your program or GCC:
2536 Produce debugging information in the operating system's native format
2537 (stabs, COFF, XCOFF, or DWARF)@. GDB can work with this debugging
2540 On most systems that use stabs format, @option{-g} enables use of extra
2541 debugging information that only GDB can use; this extra information
2542 makes debugging work better in GDB but will probably make other debuggers
2544 refuse to read the program. If you want to control for certain whether
2545 to generate the extra information, use @option{-gstabs+}, @option{-gstabs},
2546 @option{-gxcoff+}, @option{-gxcoff}, @option{-gdwarf-1+}, or @option{-gdwarf-1}
2549 Unlike most other C compilers, GCC allows you to use @option{-g} with
2550 @option{-O}. The shortcuts taken by optimized code may occasionally
2551 produce surprising results: some variables you declared may not exist
2552 at all; flow of control may briefly move where you did not expect it;
2553 some statements may not be executed because they compute constant
2554 results or their values were already at hand; some statements may
2555 execute in different places because they were moved out of loops.
2557 Nevertheless it proves possible to debug optimized output. This makes
2558 it reasonable to use the optimizer for programs that might have bugs.
2560 The following options are useful when GCC is generated with the
2561 capability for more than one debugging format.
2565 Produce debugging information for use by GDB@. This means to use the
2566 most expressive format available (DWARF 2, stabs, or the native format
2567 if neither of those are supported), including GDB extensions if at all
2572 Produce debugging information in stabs format (if that is supported),
2573 without GDB extensions. This is the format used by DBX on most BSD
2574 systems. On MIPS, Alpha and System V Release 4 systems this option
2575 produces stabs debugging output which is not understood by DBX or SDB@.
2576 On System V Release 4 systems this option requires the GNU assembler.
2580 Produce debugging information in stabs format (if that is supported),
2581 using GNU extensions understood only by the GNU debugger (GDB)@. The
2582 use of these extensions is likely to make other debuggers crash or
2583 refuse to read the program.
2587 Produce debugging information in COFF format (if that is supported).
2588 This is the format used by SDB on most System V systems prior to
2593 Produce debugging information in XCOFF format (if that is supported).
2594 This is the format used by the DBX debugger on IBM RS/6000 systems.
2598 Produce debugging information in XCOFF format (if that is supported),
2599 using GNU extensions understood only by the GNU debugger (GDB)@. The
2600 use of these extensions is likely to make other debuggers crash or
2601 refuse to read the program, and may cause assemblers other than the GNU
2602 assembler (GAS) to fail with an error.
2606 Produce debugging information in DWARF version 1 format (if that is
2607 supported). This is the format used by SDB on most System V Release 4
2612 Produce debugging information in DWARF version 1 format (if that is
2613 supported), using GNU extensions understood only by the GNU debugger
2614 (GDB)@. The use of these extensions is likely to make other debuggers
2615 crash or refuse to read the program.
2619 Produce debugging information in DWARF version 2 format (if that is
2620 supported). This is the format used by DBX on IRIX 6.
2623 @itemx -ggdb@var{level}
2624 @itemx -gstabs@var{level}
2625 @itemx -gcoff@var{level}
2626 @itemx -gxcoff@var{level}
2627 @itemx -gdwarf@var{level}
2628 @itemx -gdwarf-2@var{level}
2629 Request debugging information and also use @var{level} to specify how
2630 much information. The default level is 2.
2632 Level 1 produces minimal information, enough for making backtraces in
2633 parts of the program that you don't plan to debug. This includes
2634 descriptions of functions and external variables, but no information
2635 about local variables and no line numbers.
2637 Level 3 includes extra information, such as all the macro definitions
2638 present in the program. Some debuggers support macro expansion when
2639 you use @option{-g3}.
2644 Generate extra code to write profile information suitable for the
2645 analysis program @code{prof}. You must use this option when compiling
2646 the source files you want data about, and you must also use it when
2649 @cindex @code{gprof}
2652 Generate extra code to write profile information suitable for the
2653 analysis program @code{gprof}. You must use this option when compiling
2654 the source files you want data about, and you must also use it when
2660 Generate extra code to write profile information for basic blocks, which will
2661 record the number of times each basic block is executed, the basic block start
2662 address, and the function name containing the basic block. If @option{-g} is
2663 used, the line number and filename of the start of the basic block will also be
2664 recorded. If not overridden by the machine description, the default action is
2665 to append to the text file @file{bb.out}.
2667 This data could be analyzed by a program like @code{tcov}. Note,
2668 however, that the format of the data is not what @code{tcov} expects.
2669 Eventually GNU @code{gprof} should be extended to process this data.
2673 Makes the compiler print out each function name as it is compiled, and
2674 print some statistics about each pass when it finishes.
2677 @opindex ftime-report
2678 Makes the compiler print some statistics about the time consumed by each
2679 pass when it finishes.
2682 @opindex fmem-report
2683 Makes the compiler print some statistics about permanent memory
2684 allocation when it finishes.
2688 Generate extra code to profile basic blocks. Your executable will
2689 produce output that is a superset of that produced when @option{-a} is
2690 used. Additional output is the source and target address of the basic
2691 blocks where a jump takes place, the number of times a jump is executed,
2692 and (optionally) the complete sequence of basic blocks being executed.
2693 The output is appended to file @file{bb.out}.
2695 You can examine different profiling aspects without recompilation. Your
2696 executable will read a list of function names from file @file{bb.in}.
2697 Profiling starts when a function on the list is entered and stops when
2698 that invocation is exited. To exclude a function from profiling, prefix
2699 its name with @samp{-}. If a function name is not unique, you can
2700 disambiguate it by writing it in the form
2701 @samp{/path/filename.d:functionname}. Your executable will write the
2702 available paths and filenames in file @file{bb.out}.
2704 Several function names have a special meaning:
2707 Write source, target and frequency of jumps to file @file{bb.out}.
2708 @item __bb_hidecall__
2709 Exclude function calls from frequency count.
2710 @item __bb_showret__
2711 Include function returns in frequency count.
2713 Write the sequence of basic blocks executed to file @file{bbtrace.gz}.
2714 The file will be compressed using the program @samp{gzip}, which must
2715 exist in your @env{PATH}. On systems without the @samp{popen}
2716 function, the file will be named @file{bbtrace} and will not be
2717 compressed. @strong{Profiling for even a few seconds on these systems
2718 will produce a very large file.} Note: @code{__bb_hidecall__} and
2719 @code{__bb_showret__} will not affect the sequence written to
2723 Here's a short example using different profiling parameters
2724 in file @file{bb.in}. Assume function @code{foo} consists of basic blocks
2725 1 and 2 and is called twice from block 3 of function @code{main}. After
2726 the calls, block 3 transfers control to block 4 of @code{main}.
2728 With @code{__bb_trace__} and @code{main} contained in file @file{bb.in},
2729 the following sequence of blocks is written to file @file{bbtrace.gz}:
2730 0 3 1 2 1 2 4. The return from block 2 to block 3 is not shown, because
2731 the return is to a point inside the block and not to the top. The
2732 block address 0 always indicates, that control is transferred
2733 to the trace from somewhere outside the observed functions. With
2734 @samp{-foo} added to @file{bb.in}, the blocks of function
2735 @code{foo} are removed from the trace, so only 0 3 4 remains.
2737 With @code{__bb_jumps__} and @code{main} contained in file @file{bb.in},
2738 jump frequencies will be written to file @file{bb.out}. The
2739 frequencies are obtained by constructing a trace of blocks
2740 and incrementing a counter for every neighbouring pair of blocks
2741 in the trace. The trace 0 3 1 2 1 2 4 displays the following
2745 Jump from block 0x0 to block 0x3 executed 1 time(s)
2746 Jump from block 0x3 to block 0x1 executed 1 time(s)
2747 Jump from block 0x1 to block 0x2 executed 2 time(s)
2748 Jump from block 0x2 to block 0x1 executed 1 time(s)
2749 Jump from block 0x2 to block 0x4 executed 1 time(s)
2752 With @code{__bb_hidecall__}, control transfer due to call instructions
2753 is removed from the trace, that is the trace is cut into three parts: 0
2754 3 4, 0 1 2 and 0 1 2. With @code{__bb_showret__}, control transfer due
2755 to return instructions is added to the trace. The trace becomes: 0 3 1
2756 2 3 1 2 3 4. Note, that this trace is not the same, as the sequence
2757 written to @file{bbtrace.gz}. It is solely used for counting jump
2760 @item -fprofile-arcs
2761 @opindex fprofile-arcs
2762 Instrument @dfn{arcs} during compilation to generate coverage data
2763 or for profile-directed block ordering. During execution the program
2764 records how many times each branch is executed and how many times it is
2765 taken. When the compiled program exits it saves this data to a file
2766 called @file{@var{sourcename}.da} for each source file.
2768 For profile-directed block ordering, compile the program with
2769 @option{-fprofile-arcs} plus optimization and code generation options,
2770 generate the arc profile information by running the program on a
2771 selected workload, and then compile the program again with the same
2772 optimization and code generation options plus
2773 @option{-fbranch-probabilities} (@pxref{Optimize Options,,Options that
2774 Control Optimization}).
2776 The other use of @option{-fprofile-arcs} is for use with @code{gcov},
2777 when it is used with the @option{-ftest-coverage} option. GCC
2778 supports two methods of determining code coverage: the options that
2779 support @code{gcov}, and options @option{-a} and @option{-ax}, which
2780 write information to text files. The options that support @code{gcov}
2781 do not need to instrument every arc in the program, so a program compiled
2782 with them runs faster than a program compiled with @option{-a}, which
2783 adds instrumentation code to every basic block in the program. The
2784 tradeoff: since @code{gcov} does not have execution counts for all
2785 branches, it must start with the execution counts for the instrumented
2786 branches, and then iterate over the program flow graph until the entire
2787 graph has been solved. Hence, @code{gcov} runs a little more slowly than
2788 a program which uses information from @option{-a} and @option{-ax}.
2790 With @option{-fprofile-arcs}, for each function of your program GCC
2791 creates a program flow graph, then finds a spanning tree for the graph.
2792 Only arcs that are not on the spanning tree have to be instrumented: the
2793 compiler adds code to count the number of times that these arcs are
2794 executed. When an arc is the only exit or only entrance to a block, the
2795 instrumentation code can be added to the block; otherwise, a new basic
2796 block must be created to hold the instrumentation code.
2798 This option makes it possible to estimate branch probabilities and to
2799 calculate basic block execution counts. In general, basic block
2800 execution counts as provided by @option{-a} do not give enough
2801 information to estimate all branch probabilities.
2804 @item -ftest-coverage
2805 @opindex ftest-coverage
2806 Create data files for the @code{gcov} code-coverage utility
2807 (@pxref{Gcov,, @code{gcov}: a GCC Test Coverage Program}).
2808 The data file names begin with the name of your source file:
2811 @item @var{sourcename}.bb
2812 A mapping from basic blocks to line numbers, which @code{gcov} uses to
2813 associate basic block execution counts with line numbers.
2815 @item @var{sourcename}.bbg
2816 A list of all arcs in the program flow graph. This allows @code{gcov}
2817 to reconstruct the program flow graph, so that it can compute all basic
2818 block and arc execution counts from the information in the
2819 @code{@var{sourcename}.da} file.
2822 Use @option{-ftest-coverage} with @option{-fprofile-arcs}; the latter
2823 option adds instrumentation to the program, which then writes
2824 execution counts to another data file:
2827 @item @var{sourcename}.da
2828 Runtime arc execution counts, used in conjunction with the arc
2829 information in the file @code{@var{sourcename}.bbg}.
2832 Coverage data will map better to the source files if
2833 @option{-ftest-coverage} is used without optimization.
2835 @item -d@var{letters}
2837 Says to make debugging dumps during compilation at times specified by
2838 @var{letters}. This is used for debugging the compiler. The file names
2839 for most of the dumps are made by appending a pass number and a word to
2840 the source file name (e.g. @file{foo.c.00.rtl} or @file{foo.c.01.sibling}).
2841 Here are the possible letters for use in @var{letters}, and their meanings:
2846 Annotate the assembler output with miscellaneous debugging information.
2849 Dump after computing branch probabilities, to @file{@var{file}.14.bp}.
2852 Dump after block reordering, to @file{@var{file}.28.bbro}.
2855 Dump after instruction combination, to the file @file{@var{file}.16.combine}.
2858 Dump after the first if conversion, to the file @file{@var{file}.17.ce}.
2861 Dump after delayed branch scheduling, to @file{@var{file}.31.dbr}.
2864 Dump all macro definitions, at the end of preprocessing, in addition to
2868 Dump after SSA optimizations, to @file{@var{file}.04.ssa} and
2869 @file{@var{file}.07.ussa}.
2872 Dump after the second if conversion, to @file{@var{file}.26.ce2}.
2875 Dump after life analysis, to @file{@var{file}.15.life}.
2878 Dump after purging @code{ADDRESSOF} codes, to @file{@var{file}.09.addressof}.
2881 Dump after global register allocation, to @file{@var{file}.21.greg}.
2884 Dump after finalization of EH handling code, to @file{@var{file}.02.eh}.
2887 Dump after post-reload optimizations, to @file{@var{file}.22.postreload}.
2890 Dump after GCSE, to @file{@var{file}.10.gcse}.
2893 Dump after sibling call optimizations, to @file{@var{file}.01.sibling}.
2896 Dump after the first jump optimization, to @file{@var{file}.03.jump}.
2899 Dump after conversion from registers to stack, to @file{@var{file}.32.stack}.
2902 Dump after local register allocation, to @file{@var{file}.20.lreg}.
2905 Dump after loop optimization, to @file{@var{file}.11.loop}.
2908 Dump after performing the machine dependent reorganisation pass, to
2909 @file{@var{file}.30.mach}.
2912 Dump after register renumbering, to @file{@var{file}.25.rnreg}.
2915 Dump after the register move pass, to @file{@var{file}.18.regmove}.
2918 Dump after RTL generation, to @file{@var{file}.00.rtl}.
2921 Dump after the second scheduling pass, to @file{@var{file}.27.sched2}.
2924 Dump after CSE (including the jump optimization that sometimes follows
2925 CSE), to @file{@var{file}.08.cse}.
2928 Dump after the first scheduling pass, to @file{@var{file}.19.sched}.
2931 Dump after the second CSE pass (including the jump optimization that
2932 sometimes follows CSE), to @file{@var{file}.12.cse2}.
2935 Dump after the second flow pass, to @file{@var{file}.23.flow2}.
2938 Dump after SSA dead code elimination, to @file{@var{file}.06.ssadce}.
2941 Dump after the peephole pass, to @file{@var{file}.24.peephole2}.
2944 Produce all the dumps listed above.
2947 Print statistics on memory usage, at the end of the run, to
2951 Annotate the assembler output with a comment indicating which
2952 pattern and alternative was used. The length of each instruction is
2956 Dump the RTL in the assembler output as a comment before each instruction.
2957 Also turns on @option{-dp} annotation.
2960 For each of the other indicated dump files (except for
2961 @file{@var{file}.00.rtl}), dump a representation of the control flow graph
2962 suitable for viewing with VCG to @file{@var{file}.@var{pass}.vcg}.
2965 Just generate RTL for a function instead of compiling it. Usually used
2969 Dump debugging information during parsing, to standard error.
2972 @item -fdump-unnumbered
2973 @opindex fdump-unnumbered
2974 When doing debugging dumps (see @option{-d} option above), suppress instruction
2975 numbers and line number note output. This makes it more feasible to
2976 use diff on debugging dumps for compiler invocations with different
2977 options, in particular with and without @option{-g}.
2979 @item -fdump-translation-unit @r{(C and C++ only)}
2980 @itemx -fdump-translation-unit-@var{options} @r{(C and C++ only)}
2981 @opindex fdump-translation-unit
2982 Dump a representation of the tree structure for the entire translation
2983 unit to a file. The file name is made by appending @file{.tu} to the
2984 source file name. If the @samp{-@var{options}} form is used, @var{options}
2985 controls the details of the dump as described for the
2986 @option{-fdump-tree} options.
2988 @item -fdump-class-hierarchy @r{(C++ only)}
2989 @itemx -fdump-class-hierarchy-@var{options} @r{(C++ only)}
2990 @opindex fdump-class-hierarchy
2991 Dump a representation of each class's hierarchy and virtual function
2992 table layout to a file. The file name is made by appending @file{.class}
2993 to the source file name. If the @samp{-@var{options}} form is used,
2994 @var{options} controls the details of the dump as described for the
2995 @option{-fdump-tree} options.
2997 @item -fdump-tree-@var{switch} @r{(C++ only)}
2998 @itemx -fdump-tree-@var{switch}-@var{options} @r{(C++ only)}
3000 Control the dumping at various stages of processing the intermediate
3001 language tree to a file. The file name is generated by appending a switch
3002 specific suffix to the source file name. If the @samp{-@var{options}}
3003 form is used, @var{options} is a list of @samp{-} separated options that
3004 control the details of the dump. Not all options are applicable to all
3005 dumps, those which are not meaningful will be ignored. The following
3006 options are available
3010 Print the address of each node. Usually this is not meaningful as it
3011 changes according to the environment and source file. Its primary use
3012 is for tying up a dump file with a debug environment.
3014 Inhibit dumping of members of a scope or body of a function merely
3015 because that scope has been reached. Only dump such items when they
3016 are directly reachable by some other path.
3018 Turn on all options.
3021 The following tree dumps are possible:
3024 Dump before any tree based optimization, to @file{@var{file}.original}.
3026 Dump after all tree based optimization, to @file{@var{file}.optimized}.
3028 Dump after function inlining, to @file{@var{file}.inlined}.
3031 @item -fpretend-float
3032 @opindex fpretend-float
3033 When running a cross-compiler, pretend that the target machine uses the
3034 same floating point format as the host machine. This causes incorrect
3035 output of the actual floating constants, but the actual instruction
3036 sequence will probably be the same as GCC would make when running on
3041 Store the usual ``temporary'' intermediate files permanently; place them
3042 in the current directory and name them based on the source file. Thus,
3043 compiling @file{foo.c} with @samp{-c -save-temps} would produce files
3044 @file{foo.i} and @file{foo.s}, as well as @file{foo.o}. This creates a
3045 preprocessed @file{foo.i} output file even though the compiler now
3046 normally uses an integrated preprocessor.
3050 Report the CPU time taken by each subprocess in the compilation
3051 sequence. For C source files, this is the compiler proper and assembler
3052 (plus the linker if linking is done). The output looks like this:
3059 The first number on each line is the ``user time,'' that is time spent
3060 executing the program itself. The second number is ``system time,''
3061 time spent executing operating system routines on behalf of the program.
3062 Both numbers are in seconds.
3064 @item -print-file-name=@var{library}
3065 @opindex print-file-name
3066 Print the full absolute name of the library file @var{library} that
3067 would be used when linking---and don't do anything else. With this
3068 option, GCC does not compile or link anything; it just prints the
3071 @item -print-multi-directory
3072 @opindex print-multi-directory
3073 Print the directory name corresponding to the multilib selected by any
3074 other switches present in the command line. This directory is supposed
3075 to exist in @env{GCC_EXEC_PREFIX}.
3077 @item -print-multi-lib
3078 @opindex print-multi-lib
3079 Print the mapping from multilib directory names to compiler switches
3080 that enable them. The directory name is separated from the switches by
3081 @samp{;}, and each switch starts with an @samp{@@} instead of the
3082 @samp{-}, without spaces between multiple switches. This is supposed to
3083 ease shell-processing.
3085 @item -print-prog-name=@var{program}
3086 @opindex print-prog-name
3087 Like @option{-print-file-name}, but searches for a program such as @samp{cpp}.
3089 @item -print-libgcc-file-name
3090 @opindex print-libgcc-file-name
3091 Same as @option{-print-file-name=libgcc.a}.
3093 This is useful when you use @option{-nostdlib} or @option{-nodefaultlibs}
3094 but you do want to link with @file{libgcc.a}. You can do
3097 gcc -nostdlib @var{files}@dots{} `gcc -print-libgcc-file-name`
3100 @item -print-search-dirs
3101 @opindex print-search-dirs
3102 Print the name of the configured installation directory and a list of
3103 program and library directories gcc will search---and don't do anything else.
3105 This is useful when gcc prints the error message
3106 @samp{installation problem, cannot exec cpp0: No such file or directory}.
3107 To resolve this you either need to put @file{cpp0} and the other compiler
3108 components where gcc expects to find them, or you can set the environment
3109 variable @env{GCC_EXEC_PREFIX} to the directory where you installed them.
3110 Don't forget the trailing '/'.
3111 @xref{Environment Variables}.
3114 @opindex dumpmachine
3115 Print the compiler's target machine (for example,
3116 @samp{i686-pc-linux-gnu})---and don't do anything else.
3119 @opindex dumpversion
3120 Print the compiler version (for example, @samp{3.0})---and don't do
3125 Print the compiler's built-in specs---and don't do anything else. (This
3126 is used when GCC itself is being built.) @xref{Spec Files}.
3129 @node Optimize Options
3130 @section Options That Control Optimization
3131 @cindex optimize options
3132 @cindex options, optimization
3134 These options control various sorts of optimizations:
3141 Optimize. Optimizing compilation takes somewhat more time, and a lot
3142 more memory for a large function.
3144 Without @option{-O}, the compiler's goal is to reduce the cost of
3145 compilation and to make debugging produce the expected results.
3146 Statements are independent: if you stop the program with a breakpoint
3147 between statements, you can then assign a new value to any variable or
3148 change the program counter to any other statement in the function and
3149 get exactly the results you would expect from the source code.
3151 With @option{-O}, the compiler tries to reduce code size and execution
3152 time, without performing any optimizations that take a great deal of
3157 Optimize even more. GCC performs nearly all supported optimizations
3158 that do not involve a space-speed tradeoff. The compiler does not
3159 perform loop unrolling or function inlining when you specify @option{-O2}.
3160 As compared to @option{-O}, this option increases both compilation time
3161 and the performance of the generated code.
3163 @option{-O2} turns on all optional optimizations except for loop unrolling,
3164 function inlining, and register renaming. It also turns on the
3165 @option{-fforce-mem} option on all machines and frame pointer elimination
3166 on machines where doing so does not interfere with debugging.
3168 Please note the warning under @option{-fgcse} about
3169 invoking @option{-O2} on programs that use computed gotos.
3173 Optimize yet more. @option{-O3} turns on all optimizations specified by
3174 @option{-O2} and also turns on the @option{-finline-functions} and
3175 @option{-frename-registers} options.
3183 Optimize for size. @option{-Os} enables all @option{-O2} optimizations that
3184 do not typically increase code size. It also performs further
3185 optimizations designed to reduce code size.
3187 If you use multiple @option{-O} options, with or without level numbers,
3188 the last such option is the one that is effective.
3191 Options of the form @option{-f@var{flag}} specify machine-independent
3192 flags. Most flags have both positive and negative forms; the negative
3193 form of @option{-ffoo} would be @option{-fno-foo}. In the table below,
3194 only one of the forms is listed---the one which is not the default.
3195 You can figure out the other form by either removing @samp{no-} or
3200 @opindex ffloat-store
3201 Do not store floating point variables in registers, and inhibit other
3202 options that might change whether a floating point value is taken from a
3205 @cindex floating point precision
3206 This option prevents undesirable excess precision on machines such as
3207 the 68000 where the floating registers (of the 68881) keep more
3208 precision than a @code{double} is supposed to have. Similarly for the
3209 x86 architecture. For most programs, the excess precision does only
3210 good, but a few programs rely on the precise definition of IEEE floating
3211 point. Use @option{-ffloat-store} for such programs, after modifying
3212 them to store all pertinent intermediate computations into variables.
3214 @item -fno-default-inline
3215 @opindex fno-default-inline
3216 Do not make member functions inline by default merely because they are
3217 defined inside the class scope (C++ only). Otherwise, when you specify
3218 @w{@option{-O}}, member functions defined inside class scope are compiled
3219 inline by default; i.e., you don't need to add @samp{inline} in front of
3220 the member function name.
3222 @item -fno-defer-pop
3223 @opindex fno-defer-pop
3224 Always pop the arguments to each function call as soon as that function
3225 returns. For machines which must pop arguments after a function call,
3226 the compiler normally lets arguments accumulate on the stack for several
3227 function calls and pops them all at once.
3231 Force memory operands to be copied into registers before doing
3232 arithmetic on them. This produces better code by making all memory
3233 references potential common subexpressions. When they are not common
3234 subexpressions, instruction combination should eliminate the separate
3235 register-load. The @option{-O2} option turns on this option.
3238 @opindex fforce-addr
3239 Force memory address constants to be copied into registers before
3240 doing arithmetic on them. This may produce better code just as
3241 @option{-fforce-mem} may.
3243 @item -fomit-frame-pointer
3244 @opindex fomit-frame-pointer
3245 Don't keep the frame pointer in a register for functions that
3246 don't need one. This avoids the instructions to save, set up and
3247 restore frame pointers; it also makes an extra register available
3248 in many functions. @strong{It also makes debugging impossible on
3252 On some machines, such as the VAX, this flag has no effect, because
3253 the standard calling sequence automatically handles the frame pointer
3254 and nothing is saved by pretending it doesn't exist. The
3255 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3256 whether a target machine supports this flag. @xref{Registers}.
3259 On some machines, such as the VAX, this flag has no effect, because
3260 the standard calling sequence automatically handles the frame pointer
3261 and nothing is saved by pretending it doesn't exist. The
3262 machine-description macro @code{FRAME_POINTER_REQUIRED} controls
3263 whether a target machine supports this flag. @xref{Registers,,Register
3264 Usage, gcc.info, Using and Porting GCC}.
3267 @item -foptimize-sibling-calls
3268 @opindex foptimize-sibling-calls
3269 Optimize sibling and tail recursive calls.
3273 This option generates traps for signed overflow on addition, subtraction,
3274 multiplication operations.
3278 Don't pay attention to the @code{inline} keyword. Normally this option
3279 is used to keep the compiler from expanding any functions inline.
3280 Note that if you are not optimizing, no functions can be expanded inline.
3282 @item -finline-functions
3283 @opindex finline-functions
3284 Integrate all simple functions into their callers. The compiler
3285 heuristically decides which functions are simple enough to be worth
3286 integrating in this way.
3288 If all calls to a given function are integrated, and the function is
3289 declared @code{static}, then the function is normally not output as
3290 assembler code in its own right.
3292 @item -finline-limit=@var{n}
3293 @opindex finline-limit
3294 By default, gcc limits the size of functions that can be inlined. This flag
3295 allows the control of this limit for functions that are explicitly marked as
3296 inline (ie marked with the inline keyword or defined within the class
3297 definition in c++). @var{n} is the size of functions that can be inlined in
3298 number of pseudo instructions (not counting parameter handling). The default
3299 value of @var{n} is 600.
3300 Increasing this value can result in more inlined code at
3301 the cost of compilation time and memory consumption. Decreasing usually makes
3302 the compilation faster and less code will be inlined (which presumably
3303 means slower programs). This option is particularly useful for programs that
3304 use inlining heavily such as those based on recursive templates with C++.
3306 @emph{Note:} pseudo instruction represents, in this particular context, an
3307 abstract measurement of function's size. In no way, it represents a count
3308 of assembly instructions and as such its exact meaning might change from one
3309 release to an another.
3311 @item -fkeep-inline-functions
3312 @opindex fkeep-inline-functions
3313 Even if all calls to a given function are integrated, and the function
3314 is declared @code{static}, nevertheless output a separate run-time
3315 callable version of the function. This switch does not affect
3316 @code{extern inline} functions.
3318 @item -fkeep-static-consts
3319 @opindex fkeep-static-consts
3320 Emit variables declared @code{static const} when optimization isn't turned
3321 on, even if the variables aren't referenced.
3323 GCC enables this option by default. If you want to force the compiler to
3324 check if the variable was referenced, regardless of whether or not
3325 optimization is turned on, use the @option{-fno-keep-static-consts} option.
3327 @item -fmerge-constants
3328 Attempt to merge identical constants (string constants and floating point
3329 constants) accross compilation units.
3331 This option is default for optimized compilation if assembler and linker
3332 support it. Use @option{-fno-merge-constants} to inhibit this behaviour.
3334 @item -fmerge-all-constants
3335 Attempt to merge identical constants and identical variables.
3337 This option implies @option{-fmerge-constants}. In addition to
3338 @option{-fmerge-constants} this considers e.g. even constant initialized
3339 arrays or initialized constant variables with integral or floating point
3340 types. Languages like C or C++ require each non-automatic variable to
3341 have distinct location, so using this option will result in non-conforming
3344 @item -fno-function-cse
3345 @opindex fno-function-cse
3346 Do not put function addresses in registers; make each instruction that
3347 calls a constant function contain the function's address explicitly.
3349 This option results in less efficient code, but some strange hacks
3350 that alter the assembler output may be confused by the optimizations
3351 performed when this option is not used.
3355 Sets @option{-fno-math-errno}, @option{-funsafe-math-optimizations}, and @*
3356 @option{-fno-trapping-math}.
3358 This option causes the preprocessor macro @code{__FAST_MATH__} to be defined.
3360 This option should never be turned on by any @option{-O} option since
3361 it can result in incorrect output for programs which depend on
3362 an exact implementation of IEEE or ISO rules/specifications for
3365 @item -fno-math-errno
3366 @opindex fno-math-errno
3367 Do not set ERRNO after calling math functions that are executed
3368 with a single instruction, e.g., sqrt. A program that relies on
3369 IEEE exceptions for math error handling may want to use this flag
3370 for speed while maintaining IEEE arithmetic compatibility.
3372 This option should never be turned on by any @option{-O} option since
3373 it can result in incorrect output for programs which depend on
3374 an exact implementation of IEEE or ISO rules/specifications for
3377 The default is @option{-fmath-errno}.
3379 @item -funsafe-math-optimizations
3380 @opindex funsafe-math-optimizations
3381 Allow optimizations for floating-point arithmetic that (a) assume
3382 that arguments and results are valid and (b) may violate IEEE or
3383 ANSI standards. When used at link-time, it may include libraries
3384 or startup files that change the default FPU control word or other
3385 similar optimizations.
3387 This option should never be turned on by any @option{-O} option since
3388 it can result in incorrect output for programs which depend on
3389 an exact implementation of IEEE or ISO rules/specifications for
3392 The default is @option{-fno-unsafe-math-optimizations}.
3394 @item -fno-trapping-math
3395 @opindex fno-trapping-math
3396 Compile code assuming that floating-point operations cannot generate
3397 user-visible traps. Setting this option may allow faster code
3398 if one relies on ``non-stop'' IEEE arithmetic, for example.
3400 This option should never be turned on by any @option{-O} option since
3401 it can result in incorrect output for programs which depend on
3402 an exact implementation of IEEE or ISO rules/specifications for
3405 The default is @option{-ftrapping-math}.
3408 The following options control specific optimizations. The @option{-O2}
3409 option turns on all of these optimizations except @option{-funroll-loops}
3410 and @option{-funroll-all-loops}. On most machines, the @option{-O} option
3411 turns on the @option{-fthread-jumps} and @option{-fdelayed-branch} options,
3412 but specific machines may handle it differently.
3414 You can use the following flags in the rare cases when ``fine-tuning''
3415 of optimizations to be performed is desired.
3417 Not all of the optimizations performed by GCC have @option{-f} options
3421 @item -fstrength-reduce
3422 @opindex fstrength-reduce
3423 Perform the optimizations of loop strength reduction and
3424 elimination of iteration variables.
3426 @item -fthread-jumps
3427 @opindex fthread-jumps
3428 Perform optimizations where we check to see if a jump branches to a
3429 location where another comparison subsumed by the first is found. If
3430 so, the first branch is redirected to either the destination of the
3431 second branch or a point immediately following it, depending on whether
3432 the condition is known to be true or false.
3434 @item -fcse-follow-jumps
3435 @opindex fcse-follow-jumps
3436 In common subexpression elimination, scan through jump instructions
3437 when the target of the jump is not reached by any other path. For
3438 example, when CSE encounters an @code{if} statement with an
3439 @code{else} clause, CSE will follow the jump when the condition
3442 @item -fcse-skip-blocks
3443 @opindex fcse-skip-blocks
3444 This is similar to @option{-fcse-follow-jumps}, but causes CSE to
3445 follow jumps which conditionally skip over blocks. When CSE
3446 encounters a simple @code{if} statement with no else clause,
3447 @option{-fcse-skip-blocks} causes CSE to follow the jump around the
3448 body of the @code{if}.
3450 @item -frerun-cse-after-loop
3451 @opindex frerun-cse-after-loop
3452 Re-run common subexpression elimination after loop optimizations has been
3455 @item -frerun-loop-opt
3456 @opindex frerun-loop-opt
3457 Run the loop optimizer twice.
3461 Perform a global common subexpression elimination pass.
3462 This pass also performs global constant and copy propagation.
3464 @emph{Note:} When compiling a program using computed gotos, a GCC
3465 extension, you may get better runtime performance if you disable
3466 the global common subexpression elmination pass by adding
3467 @option{-fno-gcse} to the command line.
3471 When @option{-fgcse-lm} is enabled, global common subexpression elimination will
3472 attempt to move loads which are only killed by stores into themselves. This
3473 allows a loop containing a load/store sequence to be changed to a load outside
3474 the loop, and a copy/store within the loop.
3478 When @option{-fgcse-sm} is enabled, A store motion pass is run after global common
3479 subexpression elimination. This pass will attempt to move stores out of loops.
3480 When used in conjunction with @option{-fgcse-lm}, loops containing a load/store sequence
3481 can be changed to a load before the loop and a store after the loop.
3483 @item -fdelete-null-pointer-checks
3484 @opindex fdelete-null-pointer-checks
3485 Use global dataflow analysis to identify and eliminate useless checks
3486 for null pointers. The compiler assumes that dereferencing a null
3487 pointer would have halted the program. If a pointer is checked after
3488 it has already been dereferenced, it cannot be null.
3490 In some environments, this assumption is not true, and programs can
3491 safely dereference null pointers. Use
3492 @option{-fno-delete-null-pointer-checks} to disable this optimization
3493 for programs which depend on that behavior.
3495 @item -fexpensive-optimizations
3496 @opindex fexpensive-optimizations
3497 Perform a number of minor optimizations that are relatively expensive.
3499 @item -foptimize-register-move
3501 @opindex foptimize-register-move
3503 Attempt to reassign register numbers in move instructions and as
3504 operands of other simple instructions in order to maximize the amount of
3505 register tying. This is especially helpful on machines with two-operand
3506 instructions. GCC enables this optimization by default with @option{-O2}
3509 Note @option{-fregmove} and @option{-foptimize-register-move} are the same
3512 @item -fdelayed-branch
3513 @opindex fdelayed-branch
3514 If supported for the target machine, attempt to reorder instructions
3515 to exploit instruction slots available after delayed branch
3518 @item -fschedule-insns
3519 @opindex fschedule-insns
3520 If supported for the target machine, attempt to reorder instructions to
3521 eliminate execution stalls due to required data being unavailable. This
3522 helps machines that have slow floating point or memory load instructions
3523 by allowing other instructions to be issued until the result of the load
3524 or floating point instruction is required.
3526 @item -fschedule-insns2
3527 @opindex fschedule-insns2
3528 Similar to @option{-fschedule-insns}, but requests an additional pass of
3529 instruction scheduling after register allocation has been done. This is
3530 especially useful on machines with a relatively small number of
3531 registers and where memory load instructions take more than one cycle.
3533 @item -ffunction-sections
3534 @itemx -fdata-sections
3535 @opindex ffunction-sections
3536 @opindex fdata-sections
3537 Place each function or data item into its own section in the output
3538 file if the target supports arbitrary sections. The name of the
3539 function or the name of the data item determines the section's name
3542 Use these options on systems where the linker can perform optimizations
3543 to improve locality of reference in the instruction space. HPPA
3544 processors running HP-UX and Sparc processors running Solaris 2 have
3545 linkers with such optimizations. Other systems using the ELF object format
3546 as well as AIX may have these optimizations in the future.
3548 Only use these options when there are significant benefits from doing
3549 so. When you specify these options, the assembler and linker will
3550 create larger object and executable files and will also be slower.
3551 You will not be able to use @code{gprof} on all systems if you
3552 specify this option and you may have problems with debugging if
3553 you specify both this option and @option{-g}.
3555 @item -fcaller-saves
3556 @opindex fcaller-saves
3557 Enable values to be allocated in registers that will be clobbered by
3558 function calls, by emitting extra instructions to save and restore the
3559 registers around such calls. Such allocation is done only when it
3560 seems to result in better code than would otherwise be produced.
3562 This option is always enabled by default on certain machines, usually
3563 those which have no call-preserved registers to use instead.
3565 For all machines, optimization level 2 and higher enables this flag by
3568 @item -funroll-loops
3569 @opindex funroll-loops
3570 Unroll loops whose number of iterations can be determined at compile
3571 time or upon entry to the loop. @option{-funroll-loops} implies both
3572 @option{-fstrength-reduce} and @option{-frerun-cse-after-loop}. This
3573 option makes code larger, and may or may not make it run faster.
3575 @item -funroll-all-loops
3576 @opindex funroll-all-loops
3577 Unroll all loops, even if their number of iterations is uncertain when
3578 the loop is entered. This usually makes programs run more slowly.
3579 @option{-funroll-all-loops} implies the same options as
3580 @option{-funroll-loops},
3583 @item -fmove-all-movables
3584 @opindex fmove-all-movables
3585 Forces all invariant computations in loops to be moved
3588 @item -freduce-all-givs
3589 @opindex freduce-all-givs
3590 Forces all general-induction variables in loops to be
3593 @emph{Note:} When compiling programs written in Fortran,
3594 @option{-fmove-all-movables} and @option{-freduce-all-givs} are enabled
3595 by default when you use the optimizer.
3597 These options may generate better or worse code; results are highly
3598 dependent on the structure of loops within the source code.
3600 These two options are intended to be removed someday, once
3601 they have helped determine the efficacy of various
3602 approaches to improving loop optimizations.
3604 Please let us (@w{@email{gcc@@gcc.gnu.org}} and @w{@email{fortran@@gnu.org}})
3605 know how use of these options affects
3606 the performance of your production code.
3607 We're very interested in code that runs @emph{slower}
3608 when these options are @emph{enabled}.
3611 @itemx -fno-peephole2
3612 @opindex fno-peephole
3613 @opindex fno-peephole2
3614 Disable any machine-specific peephole optimizations. The difference
3615 between @option{-fno-peephole} and @option{-fno-peephole2} is in how they
3616 are implemented in the compiler; some targets use one, some use the
3617 other, a few use both.
3619 @item -fbranch-probabilities
3620 @opindex fbranch-probabilities
3621 After running a program compiled with @option{-fprofile-arcs}
3622 (@pxref{Debugging Options,, Options for Debugging Your Program or
3623 @command{gcc}}), you can compile it a second time using
3624 @option{-fbranch-probabilities}, to improve optimizations based on
3625 the number of times each branch was taken. When the program
3626 compiled with @option{-fprofile-arcs} exits it saves arc execution
3627 counts to a file called @file{@var{sourcename}.da} for each source
3628 file The information in this data file is very dependent on the
3629 structure of the generated code, so you must use the same source code
3630 and the same optimization options for both compilations.
3633 With @option{-fbranch-probabilities}, GCC puts a @samp{REG_EXEC_COUNT}
3634 note on the first instruction of each basic block, and a
3635 @samp{REG_BR_PROB} note on each @samp{JUMP_INSN} and @samp{CALL_INSN}.
3636 These can be used to improve optimization. Currently, they are only
3637 used in one place: in @file{reorg.c}, instead of guessing which path a
3638 branch is mostly to take, the @samp{REG_BR_PROB} values are used to
3639 exactly determine which path is taken more often.
3642 @item -fno-guess-branch-probability
3643 @opindex fno-guess-branch-probability
3644 Do not guess branch probabilities using a randomized model.
3646 Sometimes gcc will opt to use a randomized model to guess branch
3647 probabilities, when none are available from either profiling feedback
3648 (@option{-fprofile-arcs}) or @samp{__builtin_expect}. This means that
3649 different runs of the compiler on the same program may produce different
3652 In a hard real-time system, people don't want different runs of the
3653 compiler to produce code that has different behavior; minimizing
3654 non-determinism is of paramount import. This switch allows users to
3655 reduce non-determinism, possibly at the expense of inferior
3658 @item -fstrict-aliasing
3659 @opindex fstrict-aliasing
3660 Allows the compiler to assume the strictest aliasing rules applicable to
3661 the language being compiled. For C (and C++), this activates
3662 optimizations based on the type of expressions. In particular, an
3663 object of one type is assumed never to reside at the same address as an
3664 object of a different type, unless the types are almost the same. For
3665 example, an @code{unsigned int} can alias an @code{int}, but not a
3666 @code{void*} or a @code{double}. A character type may alias any other
3669 Pay special attention to code like this:
3682 The practice of reading from a different union member than the one most
3683 recently written to (called ``type-punning'') is common. Even with
3684 @option{-fstrict-aliasing}, type-punning is allowed, provided the memory
3685 is accessed through the union type. So, the code above will work as
3686 expected. However, this code might not:
3698 Every language that wishes to perform language-specific alias analysis
3699 should define a function that computes, given an @code{tree}
3700 node, an alias set for the node. Nodes in different alias sets are not
3701 allowed to alias. For an example, see the C front-end function
3702 @code{c_get_alias_set}.
3705 @item -falign-functions
3706 @itemx -falign-functions=@var{n}
3707 @opindex falign-functions
3708 Align the start of functions to the next power-of-two greater than
3709 @var{n}, skipping up to @var{n} bytes. For instance,
3710 @option{-falign-functions=32} aligns functions to the next 32-byte
3711 boundary, but @option{-falign-functions=24} would align to the next
3712 32-byte boundary only if this can be done by skipping 23 bytes or less.
3714 @option{-fno-align-functions} and @option{-falign-functions=1} are
3715 equivalent and mean that functions will not be aligned.
3717 Some assemblers only support this flag when @var{n} is a power of two;
3718 in that case, it is rounded up.
3720 If @var{n} is not specified, use a machine-dependent default.
3722 @item -falign-labels
3723 @itemx -falign-labels=@var{n}
3724 @opindex falign-labels
3725 Align all branch targets to a power-of-two boundary, skipping up to
3726 @var{n} bytes like @option{-falign-functions}. This option can easily
3727 make code slower, because it must insert dummy operations for when the
3728 branch target is reached in the usual flow of the code.
3730 If @option{-falign-loops} or @option{-falign-jumps} are applicable and
3731 are greater than this value, then their values are used instead.
3733 If @var{n} is not specified, use a machine-dependent default which is
3734 very likely to be @samp{1}, meaning no alignment.
3737 @itemx -falign-loops=@var{n}
3738 @opindex falign-loops
3739 Align loops to a power-of-two boundary, skipping up to @var{n} bytes
3740 like @option{-falign-functions}. The hope is that the loop will be
3741 executed many times, which will make up for any execution of the dummy
3744 If @var{n} is not specified, use a machine-dependent default.
3747 @itemx -falign-jumps=@var{n}
3748 @opindex falign-jumps
3749 Align branch targets to a power-of-two boundary, for branch targets
3750 where the targets can only be reached by jumping, skipping up to @var{n}
3751 bytes like @option{-falign-functions}. In this case, no dummy operations
3754 If @var{n} is not specified, use a machine-dependent default.
3758 Perform optimizations in static single assignment form. Each function's
3759 flow graph is translated into SSA form, optimizations are performed, and
3760 the flow graph is translated back from SSA form. Users should not
3761 specify this option, since it is not yet ready for production use.
3765 Perform Sparse Conditional Constant Propagation in SSA form. Requires
3766 @option{-fssa}. Like @option{-fssa}, this is an experimental feature.
3770 Perform aggressive dead-code elimination in SSA form. Requires @option{-fssa}.
3771 Like @option{-fssa}, this is an experimental feature.
3773 @item -fsingle-precision-constant
3774 @opindex fsingle-precision-constant
3775 Treat floating point constant as single precision constant instead of
3776 implicitly converting it to double precision constant.
3778 @item -frename-registers
3779 @opindex frename-registers
3780 Attempt to avoid false dependencies in scheduled code by making use
3781 of registers left over after register allocation. This optimization
3782 will most benefit processors with lots of registers. It can, however,
3783 make debugging impossible, since variables will no longer stay in
3784 a ``home register''.
3786 @item --param @var{name}=@var{value}
3788 In some places, GCC uses various constants to control the amount of
3789 optimization that is done. For example, GCC will not inline functions
3790 that contain more that a certain number of instructions. You can
3791 control some of these constants on the command-line using the
3792 @option{--param} option.
3794 In each case, the @var{value} is an integer. The allowable choices for
3795 @var{name} are given in the following table:
3798 @item max-delay-slot-insn-search
3799 The maximum number of instructions to consider when looking for an
3800 instruction to fill a delay slot. If more than this arbitrary number of
3801 instructions is searched, the time savings from filling the delay slot
3802 will be minimal so stop searching. Increasing values mean more
3803 aggressive optimization, making the compile time increase with probably
3804 small improvement in executable run time.
3806 @item max-delay-slot-live-search
3807 When trying to fill delay slots, the maximum number of instructions to
3808 consider when searching for a block with valid live register
3809 information. Increasing this arbitrarily chosen value means more
3810 aggressive optimization, increasing the compile time. This parameter
3811 should be removed when the delay slot code is rewritten to maintain the
3814 @item max-gcse-memory
3815 The approximate maximum amount of memory that will be allocated in
3816 order to perform the global common subexpression elimination
3817 optimization. If more memory than specified is required, the
3818 optimization will not be done.
3820 @item max-gcse-passes
3821 The maximum number of passes of GCSE to run.
3823 @item max-pending-list-length
3824 The maximum number of pending dependancies scheduling will allow
3825 before flushing the current state and starting over. Large functions
3826 with few branches or calls can create excessively large lists which
3827 needlessly consume memory and resources.
3829 @item max-inline-insns
3830 If an function contains more than this many instructions, it
3831 will not be inlined. This option is precisely equivalent to
3832 @option{-finline-limit}.
3837 @node Preprocessor Options
3838 @section Options Controlling the Preprocessor
3839 @cindex preprocessor options
3840 @cindex options, preprocessor
3842 These options control the C preprocessor, which is run on each C source
3843 file before actual compilation.
3845 If you use the @option{-E} option, nothing is done except preprocessing.
3846 Some of these options make sense only together with @option{-E} because
3847 they cause the preprocessor output to be unsuitable for actual
3851 @item -include @var{file}
3853 Process @var{file} as input before processing the regular input file.
3854 In effect, the contents of @var{file} are compiled first. Any @option{-D}
3855 and @option{-U} options on the command line are always processed before
3856 @option{-include @var{file}}, regardless of the order in which they are
3857 written. All the @option{-include} and @option{-imacros} options are
3858 processed in the order in which they are written.
3860 @item -imacros @var{file}
3862 Process @var{file} as input, discarding the resulting output, before
3863 processing the regular input file. Because the output generated from
3864 @var{file} is discarded, the only effect of @option{-imacros @var{file}}
3865 is to make the macros defined in @var{file} available for use in the
3866 main input. All the @option{-include} and @option{-imacros} options are
3867 processed in the order in which they are written.
3869 @item -idirafter @var{dir}
3871 @cindex second include path
3872 Add the directory @var{dir} to the second include path. The directories
3873 on the second include path are searched when a header file is not found
3874 in any of the directories in the main include path (the one that
3875 @option{-I} adds to).
3877 @item -iprefix @var{prefix}
3879 Specify @var{prefix} as the prefix for subsequent @option{-iwithprefix}
3882 @item -iwithprefix @var{dir}
3883 @opindex iwithprefix
3884 Add a directory to the second include path. The directory's name is
3885 made by concatenating @var{prefix} and @var{dir}, where @var{prefix} was
3886 specified previously with @option{-iprefix}. If you have not specified a
3887 prefix yet, the directory containing the installed passes of the
3888 compiler is used as the default.
3890 @item -iwithprefixbefore @var{dir}
3891 @opindex iwithprefixbefore
3892 Add a directory to the main include path. The directory's name is made
3893 by concatenating @var{prefix} and @var{dir}, as in the case of
3894 @option{-iwithprefix}.
3896 @item -isystem @var{dir}
3898 Add a directory to the beginning of the second include path, marking it
3899 as a system directory, so that it gets the same special treatment as
3900 is applied to the standard system directories.
3904 Do not search the standard system directories for header files. Only
3905 the directories you have specified with @option{-I} options (and the
3906 current directory, if appropriate) are searched. @xref{Directory
3907 Options}, for information on @option{-I}.
3909 By using both @option{-nostdinc} and @option{-I-}, you can limit the include-file
3910 search path to only those directories you specify explicitly.
3914 When searching for a header file in a directory, remap file names if a
3915 file named @file{header.gcc} exists in that directory. This can be used
3916 to work around limitations of file systems with file name restrictions.
3917 The @file{header.gcc} file should contain a series of lines with two
3918 tokens on each line: the first token is the name to map, and the second
3919 token is the actual name to use.
3923 Do not predefine any nonstandard macros. (Including architecture flags).
3927 Run only the C preprocessor. Preprocess all the C source files
3928 specified and output the results to standard output or to the
3929 specified output file.
3933 Tell the preprocessor not to discard comments. Used with the
3938 Tell the preprocessor not to generate @samp{#line} directives.
3939 Used with the @option{-E} option.
3942 @cindex dependencies, make
3945 Instead of outputting the result of preprocessing, output a rule
3946 suitable for @code{make} describing the dependencies of the main source
3947 file. The preprocessor outputs one @code{make} rule containing the
3948 object file name for that source file, a colon, and the names of all the
3949 included files. Unless overridden explicitly, the object file name
3950 consists of the basename of the source file with any suffix replaced with
3951 object file suffix. If there are many included files then the
3952 rule is split into several lines using @samp{\}-newline.
3954 @option{-M} implies @option{-E}.
3958 Like @option{-M}, but mention only the files included with @samp{#include
3959 "@var{file}"}. System header files included with @samp{#include
3960 <@var{file}>} are omitted.
3964 Like @option{-M} but the dependency information is written to a file
3965 rather than stdout. @code{gcc} will use the same file name and
3966 directory as the object file, but with the suffix @file{.d} instead.
3968 This is in addition to compiling the main file as specified---@option{-MD}
3969 does not inhibit ordinary compilation the way @option{-M} does,
3970 unless you also specify @option{-MG}.
3972 With Mach, you can use the utility @code{md} to merge multiple
3973 dependency files into a single dependency file suitable for using with
3974 the @samp{make} command.
3978 Like @option{-MD} except mention only user header files, not system
3981 @item -MF @var{file}
3983 When used with @option{-M} or @option{-MM}, specifies a file to write the
3984 dependencies to. This allows the preprocessor to write the preprocessed
3985 file to stdout normally. If no @option{-MF} switch is given, CPP sends
3986 the rules to stdout and suppresses normal preprocessed output.
3988 Another way to specify output of a @code{make} rule is by setting
3989 the environment variable @env{DEPENDENCIES_OUTPUT} (@pxref{Environment
3994 When used with @option{-M} or @option{-MM}, @option{-MG} says to treat missing
3995 header files as generated files and assume they live in the same
3996 directory as the source file. It suppresses preprocessed output, as a
3997 missing header file is ordinarily an error.
3999 This feature is used in automatic updating of makefiles.
4003 This option instructs CPP to add a phony target for each dependency
4004 other than the main file, causing each to depend on nothing. These
4005 dummy rules work around errors @code{make} gives if you remove header
4006 files without updating the @code{Makefile} to match.
4008 This is typical output:-
4011 /tmp/test.o: /tmp/test.c /tmp/test.h
4016 @item -MQ @var{target}
4017 @item -MT @var{target}
4020 By default CPP uses the main file name, including any path, and appends
4021 the object suffix, normally ``.o'', to it to obtain the name of the
4022 target for dependency generation. With @option{-MT} you can specify a
4023 target yourself, overriding the default one.
4025 If you want multiple targets, you can specify them as a single argument
4026 to @option{-MT}, or use multiple @option{-MT} options.
4028 The targets you specify are output in the order they appear on the
4029 command line. @option{-MQ} is identical to @option{-MT}, except that the
4030 target name is quoted for Make, but with @option{-MT} it isn't. For
4031 example, @option{-MT '$(objpfx)foo.o'} gives
4034 $(objpfx)foo.o: /tmp/foo.c
4037 but @option{-MQ '$(objpfx)foo.o'} gives
4040 $$(objpfx)foo.o: /tmp/foo.c
4043 The default target is automatically quoted, as if it were given with
4048 Print the name of each header file used, in addition to other normal
4051 @item -A@var{question}(@var{answer})
4053 Assert the answer @var{answer} for @var{question}, in case it is tested
4054 with a preprocessing conditional such as @samp{#if
4055 #@var{question}(@var{answer})}. @option{-A-} disables the standard
4056 assertions that normally describe the target machine.
4060 Define macro @var{macro} with the string @samp{1} as its definition.
4062 @item -D@var{macro}=@var{defn}
4063 Define macro @var{macro} as @var{defn}. All instances of @option{-D} on
4064 the command line are processed before any @option{-U} options.
4066 Any @option{-D} and @option{-U} options on the command line are processed in
4067 order, and always before @option{-imacros @var{file}}, regardless of the
4068 order in which they are written.
4072 Undefine macro @var{macro}. @option{-U} options are evaluated after all
4073 @option{-D} options, but before any @option{-include} and @option{-imacros}
4076 Any @option{-D} and @option{-U} options on the command line are processed in
4077 order, and always before @option{-imacros @var{file}}, regardless of the
4078 order in which they are written.
4082 Tell the preprocessor to output only a list of the macro definitions
4083 that are in effect at the end of preprocessing. Used with the @option{-E}
4088 Tell the preprocessing to pass all macro definitions into the output, in
4089 their proper sequence in the rest of the output.
4093 Like @option{-dD} except that the macro arguments and contents are omitted.
4094 Only @samp{#define @var{name}} is included in the output.
4098 Output @samp{#include} directives in addition to the result of
4101 @item -fpreprocessed
4102 @opindex fpreprocessed
4103 Indicate to the preprocessor that the input file has already been
4104 preprocessed. This suppresses things like macro expansion, trigraph
4105 conversion, escaped newline splicing, and processing of most directives.
4106 The preprocessor still recognizes and removes comments, so that you can
4107 pass a file preprocessed with @option{-C} to the compiler without
4108 problems. In this mode the integrated preprocessor is little more than
4109 a tokenizer for the front ends.
4111 @option{-fpreprocessed} is implicit if the input file has one of the
4112 extensions @samp{i}, @samp{ii} or @samp{mi}. These are the extensions
4113 that GCC uses for preprocessed files created by @option{-save-temps}.
4117 Process ISO standard trigraph sequences. These are three-character
4118 sequences, all starting with @samp{??}, that are defined by ISO C to
4119 stand for single characters. For example, @samp{??/} stands for
4120 @samp{\}, so @samp{'??/n'} is a character constant for a newline. By
4121 default, GCC ignores trigraphs, but in standard-conforming modes it
4122 converts them. See the @option{-std} and @option{-ansi} options.
4124 The nine trigraph sequences are
4127 @expansion{} @samp{[}
4130 @expansion{} @samp{]}
4133 @expansion{} @samp{@{}
4136 @expansion{} @samp{@}}
4139 @expansion{} @samp{#}
4142 @expansion{} @samp{\}
4145 @expansion{} @samp{^}
4148 @expansion{} @samp{|}
4151 @expansion{} @samp{~}
4155 Trigraph support is not popular, so many compilers do not implement it
4156 properly. Portable code should not rely on trigraphs being either
4157 converted or ignored.
4159 @item -Wp,@var{option}
4161 Pass @var{option} as an option to the preprocessor. If @var{option}
4162 contains commas, it is split into multiple options at the commas.
4165 @node Assembler Options
4166 @section Passing Options to the Assembler
4168 @c prevent bad page break with this line
4169 You can pass options to the assembler.
4172 @item -Wa,@var{option}
4174 Pass @var{option} as an option to the assembler. If @var{option}
4175 contains commas, it is split into multiple options at the commas.
4179 @section Options for Linking
4180 @cindex link options
4181 @cindex options, linking
4183 These options come into play when the compiler links object files into
4184 an executable output file. They are meaningless if the compiler is
4185 not doing a link step.
4189 @item @var{object-file-name}
4190 A file name that does not end in a special recognized suffix is
4191 considered to name an object file or library. (Object files are
4192 distinguished from libraries by the linker according to the file
4193 contents.) If linking is done, these object files are used as input
4202 If any of these options is used, then the linker is not run, and
4203 object file names should not be used as arguments. @xref{Overall
4207 @item -l@var{library}
4208 @itemx -l @var{library}
4210 Search the library named @var{library} when linking. (The second
4211 alternative with the library as a separate argument is only for
4212 POSIX compliance and is not recommended.)
4214 It makes a difference where in the command you write this option; the
4215 linker searches and processes libraries and object files in the order they
4216 are specified. Thus, @samp{foo.o -lz bar.o} searches library @samp{z}
4217 after file @file{foo.o} but before @file{bar.o}. If @file{bar.o} refers
4218 to functions in @samp{z}, those functions may not be loaded.
4220 The linker searches a standard list of directories for the library,
4221 which is actually a file named @file{lib@var{library}.a}. The linker
4222 then uses this file as if it had been specified precisely by name.
4224 The directories searched include several standard system directories
4225 plus any that you specify with @option{-L}.
4227 Normally the files found this way are library files---archive files
4228 whose members are object files. The linker handles an archive file by
4229 scanning through it for members which define symbols that have so far
4230 been referenced but not defined. But if the file that is found is an
4231 ordinary object file, it is linked in the usual fashion. The only
4232 difference between using an @option{-l} option and specifying a file name
4233 is that @option{-l} surrounds @var{library} with @samp{lib} and @samp{.a}
4234 and searches several directories.
4238 You need this special case of the @option{-l} option in order to
4239 link an Objective-C program.
4242 @opindex nostartfiles
4243 Do not use the standard system startup files when linking.
4244 The standard system libraries are used normally, unless @option{-nostdlib}
4245 or @option{-nodefaultlibs} is used.
4247 @item -nodefaultlibs
4248 @opindex nodefaultlibs
4249 Do not use the standard system libraries when linking.
4250 Only the libraries you specify will be passed to the linker.
4251 The standard startup files are used normally, unless @option{-nostartfiles}
4252 is used. The compiler may generate calls to memcmp, memset, and memcpy
4253 for System V (and ISO C) environments or to bcopy and bzero for
4254 BSD environments. These entries are usually resolved by entries in
4255 libc. These entry points should be supplied through some other
4256 mechanism when this option is specified.
4260 Do not use the standard system startup files or libraries when linking.
4261 No startup files and only the libraries you specify will be passed to
4262 the linker. The compiler may generate calls to memcmp, memset, and memcpy
4263 for System V (and ISO C) environments or to bcopy and bzero for
4264 BSD environments. These entries are usually resolved by entries in
4265 libc. These entry points should be supplied through some other
4266 mechanism when this option is specified.
4268 @cindex @option{-lgcc}, use with @option{-nostdlib}
4269 @cindex @option{-nostdlib} and unresolved references
4270 @cindex unresolved references and @option{-nostdlib}
4271 @cindex @option{-lgcc}, use with @option{-nodefaultlibs}
4272 @cindex @option{-nodefaultlibs} and unresolved references
4273 @cindex unresolved references and @option{-nodefaultlibs}
4274 One of the standard libraries bypassed by @option{-nostdlib} and
4275 @option{-nodefaultlibs} is @file{libgcc.a}, a library of internal subroutines
4276 that GCC uses to overcome shortcomings of particular machines, or special
4277 needs for some languages.
4279 (@xref{Interface,,Interfacing to GCC Output}, for more discussion of
4283 (@xref{Interface,,Interfacing to GCC Output,gcc.info,Porting GCC},
4284 for more discussion of @file{libgcc.a}.)
4286 In most cases, you need @file{libgcc.a} even when you want to avoid
4287 other standard libraries. In other words, when you specify @option{-nostdlib}
4288 or @option{-nodefaultlibs} you should usually specify @option{-lgcc} as well.
4289 This ensures that you have no unresolved references to internal GCC
4290 library subroutines. (For example, @samp{__main}, used to ensure C++
4291 constructors will be called; @pxref{Collect2,,@command{collect2}}.)
4295 Remove all symbol table and relocation information from the executable.
4299 On systems that support dynamic linking, this prevents linking with the shared
4300 libraries. On other systems, this option has no effect.
4304 Produce a shared object which can then be linked with other objects to
4305 form an executable. Not all systems support this option. For predictable
4306 results, you must also specify the same set of options that were used to
4307 generate code (@option{-fpic}, @option{-fPIC}, or model suboptions)
4308 when you specify this option.@footnote{On some systems, @samp{gcc -shared}
4309 needs to build supplementary stub code for constructors to work. On
4310 multi-libbed systems, @samp{gcc -shared} must select the correct support
4311 libraries to link against. Failing to supply the correct flags may lead
4312 to subtle defects. Supplying them in cases where they are not necessary
4315 @item -shared-libgcc
4316 @itemx -static-libgcc
4317 @opindex shared-libgcc
4318 @opindex static-libgcc
4319 On systems that provide @file{libgcc} as a shared library, these options
4320 force the use of either the shared or static version respectively.
4321 If no shared version of @file{libgcc} was built when the compiler was
4322 configured, these options have no effect.
4324 There are several situations in which an application should use the
4325 shared @file{libgcc} instead of the static version. The most common
4326 of these is when the application wishes to throw and catch exceptions
4327 across different shared libraries. In that case, each of the libraries
4328 as well as the application itself should use the shared @file{libgcc}.
4330 Therefore, whenever you specify the @option{-shared} option, the GCC
4331 driver automatically adds @option{-shared-libgcc}, unless you explicitly
4332 specify @option{-static-libgcc}. The G++ driver automatically adds
4333 @option{-shared-libgcc} when you build a main executable as well because
4334 for C++ programs that is typically the right thing to do.
4335 (Exception-handling will not work reliably otherwise.)
4337 However, when linking a main executable written in C, you must
4338 explicitly say @option{-shared-libgcc} if you want to use the shared
4343 Bind references to global symbols when building a shared object. Warn
4344 about any unresolved references (unless overridden by the link editor
4345 option @samp{-Xlinker -z -Xlinker defs}). Only a few systems support
4348 @item -Xlinker @var{option}
4350 Pass @var{option} as an option to the linker. You can use this to
4351 supply system-specific linker options which GCC does not know how to
4354 If you want to pass an option that takes an argument, you must use
4355 @option{-Xlinker} twice, once for the option and once for the argument.
4356 For example, to pass @option{-assert definitions}, you must write
4357 @samp{-Xlinker -assert -Xlinker definitions}. It does not work to write
4358 @option{-Xlinker "-assert definitions"}, because this passes the entire
4359 string as a single argument, which is not what the linker expects.
4361 @item -Wl,@var{option}
4363 Pass @var{option} as an option to the linker. If @var{option} contains
4364 commas, it is split into multiple options at the commas.
4366 @item -u @var{symbol}
4368 Pretend the symbol @var{symbol} is undefined, to force linking of
4369 library modules to define it. You can use @option{-u} multiple times with
4370 different symbols to force loading of additional library modules.
4373 @node Directory Options
4374 @section Options for Directory Search
4375 @cindex directory options
4376 @cindex options, directory search
4379 These options specify directories to search for header files, for
4380 libraries and for parts of the compiler:
4385 Add the directory @var{dir} to the head of the list of directories to be
4386 searched for header files. This can be used to override a system header
4387 file, substituting your own version, since these directories are
4388 searched before the system header file directories. However, you should
4389 not use this option to add directories that contain vendor-supplied
4390 system header files (use @option{-isystem} for that). If you use more than
4391 one @option{-I} option, the directories are scanned in left-to-right
4392 order; the standard system directories come after.
4394 If a standard system include directory, or a directory specified with
4395 @option{-isystem}, is also specified with @option{-I}, it will be
4396 searched only in the position requested by @option{-I}. Also, it will
4397 not be considered a system include directory. If that directory really
4398 does contain system headers, there is a good chance that they will
4399 break. For instance, if GCC's installation procedure edited the headers
4400 in @file{/usr/include} to fix bugs, @samp{-I/usr/include} will cause the
4401 original, buggy headers to be found instead of the corrected ones. GCC
4402 will issue a warning when a system include directory is hidden in this
4407 Any directories you specify with @option{-I} options before the @option{-I-}
4408 option are searched only for the case of @samp{#include "@var{file}"};
4409 they are not searched for @samp{#include <@var{file}>}.
4411 If additional directories are specified with @option{-I} options after
4412 the @option{-I-}, these directories are searched for all @samp{#include}
4413 directives. (Ordinarily @emph{all} @option{-I} directories are used
4416 In addition, the @option{-I-} option inhibits the use of the current
4417 directory (where the current input file came from) as the first search
4418 directory for @samp{#include "@var{file}"}. There is no way to
4419 override this effect of @option{-I-}. With @option{-I.} you can specify
4420 searching the directory which was current when the compiler was
4421 invoked. That is not exactly the same as what the preprocessor does
4422 by default, but it is often satisfactory.
4424 @option{-I-} does not inhibit the use of the standard system directories
4425 for header files. Thus, @option{-I-} and @option{-nostdinc} are
4430 Add directory @var{dir} to the list of directories to be searched
4433 @item -B@var{prefix}
4435 This option specifies where to find the executables, libraries,
4436 include files, and data files of the compiler itself.
4438 The compiler driver program runs one or more of the subprograms
4439 @file{cpp}, @file{cc1}, @file{as} and @file{ld}. It tries
4440 @var{prefix} as a prefix for each program it tries to run, both with and
4441 without @samp{@var{machine}/@var{version}/} (@pxref{Target Options}).
4443 For each subprogram to be run, the compiler driver first tries the
4444 @option{-B} prefix, if any. If that name is not found, or if @option{-B}
4445 was not specified, the driver tries two standard prefixes, which are
4446 @file{/usr/lib/gcc/} and @file{/usr/local/lib/gcc-lib/}. If neither of
4447 those results in a file name that is found, the unmodified program
4448 name is searched for using the directories specified in your
4449 @env{PATH} environment variable.
4451 The compiler will check to see if the path provided by the @option{-B}
4452 refers to a directory, and if necessary it will add a directory
4453 separator character at the end of the path.
4455 @option{-B} prefixes that effectively specify directory names also apply
4456 to libraries in the linker, because the compiler translates these
4457 options into @option{-L} options for the linker. They also apply to
4458 includes files in the preprocessor, because the compiler translates these
4459 options into @option{-isystem} options for the preprocessor. In this case,
4460 the compiler appends @samp{include} to the prefix.
4462 The run-time support file @file{libgcc.a} can also be searched for using
4463 the @option{-B} prefix, if needed. If it is not found there, the two
4464 standard prefixes above are tried, and that is all. The file is left
4465 out of the link if it is not found by those means.
4467 Another way to specify a prefix much like the @option{-B} prefix is to use
4468 the environment variable @env{GCC_EXEC_PREFIX}. @xref{Environment
4471 As a special kludge, if the path provided by @option{-B} is
4472 @file{[dir/]stage@var{N}/}, where @var{N} is a number in the range 0 to
4473 9, then it will be replaced by @file{[dir/]include}. This is to help
4474 with boot-strapping the compiler.
4476 @item -specs=@var{file}
4478 Process @var{file} after the compiler reads in the standard @file{specs}
4479 file, in order to override the defaults that the @file{gcc} driver
4480 program uses when determining what switches to pass to @file{cc1},
4481 @file{cc1plus}, @file{as}, @file{ld}, etc. More than one
4482 @option{-specs=@var{file}} can be specified on the command line, and they
4483 are processed in order, from left to right.
4489 @section Specifying subprocesses and the switches to pass to them
4491 @command{gcc} is a driver program. It performs its job by invoking a
4492 sequence of other programs to do the work of compiling, assembling and
4493 linking. GCC interprets its command-line parameters and uses these to
4494 deduce which programs it should invoke, and which command-line options
4495 it ought to place on their command lines. This behaviour is controlled
4496 by @dfn{spec strings}. In most cases there is one spec string for each
4497 program that GCC can invoke, but a few programs have multiple spec
4498 strings to control their behaviour. The spec strings built into GCC can
4499 be overridden by using the @option{-specs=} command-line switch to specify
4502 @dfn{Spec files} are plaintext files that are used to construct spec
4503 strings. They consist of a sequence of directives separated by blank
4504 lines. The type of directive is determined by the first non-whitespace
4505 character on the line and it can be one of the following:
4508 @item %@var{command}
4509 Issues a @var{command} to the spec file processor. The commands that can
4513 @item %include <@var{file}>
4515 Search for @var{file} and insert its text at the current point in the
4518 @item %include_noerr <@var{file}>
4519 @cindex %include_noerr
4520 Just like @samp{%include}, but do not generate an error message if the include
4521 file cannot be found.
4523 @item %rename @var{old_name} @var{new_name}
4525 Rename the spec string @var{old_name} to @var{new_name}.
4529 @item *[@var{spec_name}]:
4530 This tells the compiler to create, override or delete the named spec
4531 string. All lines after this directive up to the next directive or
4532 blank line are considered to be the text for the spec string. If this
4533 results in an empty string then the spec will be deleted. (Or, if the
4534 spec did not exist, then nothing will happened.) Otherwise, if the spec
4535 does not currently exist a new spec will be created. If the spec does
4536 exist then its contents will be overridden by the text of this
4537 directive, unless the first character of that text is the @samp{+}
4538 character, in which case the text will be appended to the spec.
4540 @item [@var{suffix}]:
4541 Creates a new @samp{[@var{suffix}] spec} pair. All lines after this directive
4542 and up to the next directive or blank line are considered to make up the
4543 spec string for the indicated suffix. When the compiler encounters an
4544 input file with the named suffix, it will processes the spec string in
4545 order to work out how to compile that file. For example:
4552 This says that any input file whose name ends in @samp{.ZZ} should be
4553 passed to the program @samp{z-compile}, which should be invoked with the
4554 command-line switch @option{-input} and with the result of performing the
4555 @samp{%i} substitution. (See below.)
4557 As an alternative to providing a spec string, the text that follows a
4558 suffix directive can be one of the following:
4561 @item @@@var{language}
4562 This says that the suffix is an alias for a known @var{language}. This is
4563 similar to using the @option{-x} command-line switch to GCC to specify a
4564 language explicitly. For example:
4571 Says that .ZZ files are, in fact, C++ source files.
4574 This causes an error messages saying:
4577 @var{name} compiler not installed on this system.
4581 GCC already has an extensive list of suffixes built into it.
4582 This directive will add an entry to the end of the list of suffixes, but
4583 since the list is searched from the end backwards, it is effectively
4584 possible to override earlier entries using this technique.
4588 GCC has the following spec strings built into it. Spec files can
4589 override these strings or create their own. Note that individual
4590 targets can also add their own spec strings to this list.
4593 asm Options to pass to the assembler
4594 asm_final Options to pass to the assembler post-processor
4595 cpp Options to pass to the C preprocessor
4596 cc1 Options to pass to the C compiler
4597 cc1plus Options to pass to the C++ compiler
4598 endfile Object files to include at the end of the link
4599 link Options to pass to the linker
4600 lib Libraries to include on the command line to the linker
4601 libgcc Decides which GCC support library to pass to the linker
4602 linker Sets the name of the linker
4603 predefines Defines to be passed to the C preprocessor
4604 signed_char Defines to pass to CPP to say whether @code{char} is signed
4606 startfile Object files to include at the start of the link
4609 Here is a small example of a spec file:
4615 --start-group -lgcc -lc -leval1 --end-group %(old_lib)
4618 This example renames the spec called @samp{lib} to @samp{old_lib} and
4619 then overrides the previous definition of @samp{lib} with a new one.
4620 The new definition adds in some extra command-line options before
4621 including the text of the old definition.
4623 @dfn{Spec strings} are a list of command-line options to be passed to their
4624 corresponding program. In addition, the spec strings can contain
4625 @samp{%}-prefixed sequences to substitute variable text or to
4626 conditionally insert text into the command line. Using these constructs
4627 it is possible to generate quite complex command lines.
4629 Here is a table of all defined @samp{%}-sequences for spec
4630 strings. Note that spaces are not generated automatically around the
4631 results of expanding these sequences. Therefore you can concatenate them
4632 together or combine them with constant text in a single argument.
4636 Substitute one @samp{%} into the program name or argument.
4639 Substitute the name of the input file being processed.
4642 Substitute the basename of the input file being processed.
4643 This is the substring up to (and not including) the last period
4644 and not including the directory.
4647 This is the same as @samp{%b}, but include the file suffix (text after
4651 Marks the argument containing or following the @samp{%d} as a
4652 temporary file name, so that that file will be deleted if GCC exits
4653 successfully. Unlike @samp{%g}, this contributes no text to the
4656 @item %g@var{suffix}
4657 Substitute a file name that has suffix @var{suffix} and is chosen
4658 once per compilation, and mark the argument in the same way as
4659 @samp{%d}. To reduce exposure to denial-of-service attacks, the file
4660 name is now chosen in a way that is hard to predict even when previously
4661 chosen file names are known. For example, @samp{%g.s @dots{} %g.o @dots{} %g.s}
4662 might turn into @samp{ccUVUUAU.s ccXYAXZ12.o ccUVUUAU.s}. @var{suffix} matches
4663 the regexp @samp{[.A-Za-z]*} or the special string @samp{%O}, which is
4664 treated exactly as if @samp{%O} had been preprocessed. Previously, @samp{%g}
4665 was simply substituted with a file name chosen once per compilation,
4666 without regard to any appended suffix (which was therefore treated
4667 just like ordinary text), making such attacks more likely to succeed.
4669 @item %u@var{suffix}
4670 Like @samp{%g}, but generates a new temporary file name even if
4671 @samp{%u@var{suffix}} was already seen.
4673 @item %U@var{suffix}
4674 Substitutes the last file name generated with @samp{%u@var{suffix}}, generating a
4675 new one if there is no such last file name. In the absence of any
4676 @samp{%u@var{suffix}}, this is just like @samp{%g@var{suffix}}, except they don't share
4677 the same suffix @emph{space}, so @samp{%g.s @dots{} %U.s @dots{} %g.s @dots{} %U.s}
4678 would involve the generation of two distinct file names, one
4679 for each @samp{%g.s} and another for each @samp{%U.s}. Previously, @samp{%U} was
4680 simply substituted with a file name chosen for the previous @samp{%u},
4681 without regard to any appended suffix.
4683 @item %j@var{SUFFIX}
4684 Substitutes the name of the @code{HOST_BIT_BUCKET}, if any, and if it is
4685 writable, and if save-temps is off; otherwise, substitute the name
4686 of a temporary file, just like @samp{%u}. This temporary file is not
4687 meant for communication between processes, but rather as a junk
4690 @item %.@var{SUFFIX}
4691 Substitutes @var{.SUFFIX} for the suffixes of a matched switch's args
4692 when it is subsequently output with @samp{%*}. @var{SUFFIX} is
4693 terminated by the next space or %.
4696 Marks the argument containing or following the @samp{%w} as the
4697 designated output file of this compilation. This puts the argument
4698 into the sequence of arguments that @samp{%o} will substitute later.
4701 Substitutes the names of all the output files, with spaces
4702 automatically placed around them. You should write spaces
4703 around the @samp{%o} as well or the results are undefined.
4704 @samp{%o} is for use in the specs for running the linker.
4705 Input files whose names have no recognized suffix are not compiled
4706 at all, but they are included among the output files, so they will
4710 Substitutes the suffix for object files. Note that this is
4711 handled specially when it immediately follows @samp{%g, %u, or %U},
4712 because of the need for those to form complete file names. The
4713 handling is such that @samp{%O} is treated exactly as if it had already
4714 been substituted, except that @samp{%g, %u, and %U} do not currently
4715 support additional @var{suffix} characters following @samp{%O} as they would
4716 following, for example, @samp{.o}.
4719 Substitutes the standard macro predefinitions for the
4720 current target machine. Use this when running @code{cpp}.
4723 Like @samp{%p}, but puts @samp{__} before and after the name of each
4724 predefined macro, except for macros that start with @samp{__} or with
4725 @samp{_@var{L}}, where @var{L} is an uppercase letter. This is for ISO
4729 Substitute a @option{-iprefix} option made from @env{GCC_EXEC_PREFIX}.
4732 Current argument is the name of a library or startup file of some sort.
4733 Search for that file in a standard list of directories and substitute
4734 the full name found.
4737 Print @var{str} as an error message. @var{str} is terminated by a newline.
4738 Use this when inconsistent options are detected.
4741 Output @samp{-} if the input for the current command is coming from a pipe.
4744 Substitute the contents of spec string @var{name} at this point.
4747 Like @samp{%(@dots{})} but put @samp{__} around @option{-D} arguments.
4749 @item %x@{@var{option}@}
4750 Accumulate an option for @samp{%X}.
4753 Output the accumulated linker options specified by @option{-Wl} or a @samp{%x}
4757 Output the accumulated assembler options specified by @option{-Wa}.
4760 Output the accumulated preprocessor options specified by @option{-Wp}.
4763 Substitute the major version number of GCC@.
4764 (For version 2.9.5, this is 2.)
4767 Substitute the minor version number of GCC@.
4768 (For version 2.9.5, this is 9.)
4771 Substitute the patch level number of GCC@.
4772 (For version 2.9.5, this is 5.)
4775 Process the @code{asm} spec. This is used to compute the
4776 switches to be passed to the assembler.
4779 Process the @code{asm_final} spec. This is a spec string for
4780 passing switches to an assembler post-processor, if such a program is
4784 Process the @code{link} spec. This is the spec for computing the
4785 command line passed to the linker. Typically it will make use of the
4786 @samp{%L %G %S %D and %E} sequences.
4789 Dump out a @option{-L} option for each directory that GCC believes might
4790 contain startup files. If the target supports multilibs then the
4791 current multilib directory will be prepended to each of these paths.
4794 Output the multilib directory with directory separators replaced with
4795 @samp{_}. If multilib directories are not set, or the multilib directory is
4796 @file{.} then this option emits nothing.
4799 Process the @code{lib} spec. This is a spec string for deciding which
4800 libraries should be included on the command line to the linker.
4803 Process the @code{libgcc} spec. This is a spec string for deciding
4804 which GCC support library should be included on the command line to the linker.
4807 Process the @code{startfile} spec. This is a spec for deciding which
4808 object files should be the first ones passed to the linker. Typically
4809 this might be a file named @file{crt0.o}.
4812 Process the @code{endfile} spec. This is a spec string that specifies
4813 the last object files that will be passed to the linker.
4816 Process the @code{cpp} spec. This is used to construct the arguments
4817 to be passed to the C preprocessor.
4820 Process the @code{signed_char} spec. This is intended to be used
4821 to tell cpp whether a char is signed. It typically has the definition:
4823 %@{funsigned-char:-D__CHAR_UNSIGNED__@}
4827 Process the @code{cc1} spec. This is used to construct the options to be
4828 passed to the actual C compiler (@samp{cc1}).
4831 Process the @code{cc1plus} spec. This is used to construct the options to be
4832 passed to the actual C++ compiler (@samp{cc1plus}).
4835 Substitute the variable part of a matched option. See below.
4836 Note that each comma in the substituted string is replaced by
4840 Substitutes the @code{-S} switch, if that switch was given to GCC@.
4841 If that switch was not specified, this substitutes nothing. Note that
4842 the leading dash is omitted when specifying this option, and it is
4843 automatically inserted if the substitution is performed. Thus the spec
4844 string @samp{%@{foo@}} would match the command-line option @option{-foo}
4845 and would output the command line option @option{-foo}.
4847 @item %W@{@code{S}@}
4848 Like %@{@code{S}@} but mark last argument supplied within as a file to be
4851 @item %@{@code{S}*@}
4852 Substitutes all the switches specified to GCC whose names start
4853 with @code{-S}, but which also take an argument. This is used for
4854 switches like @option{-o}, @option{-D}, @option{-I}, etc.
4855 GCC considers @option{-o foo} as being
4856 one switch whose names starts with @samp{o}. %@{o*@} would substitute this
4857 text, including the space. Thus two arguments would be generated.
4859 @item %@{^@code{S}*@}
4860 Like %@{@code{S}*@}, but don't put a blank between a switch and its
4861 argument. Thus %@{^o*@} would only generate one argument, not two.
4863 @item %@{@code{S}*&@code{T}*@}
4864 Like %@{@code{S}*@}, but preserve order of @code{S} and @code{T} options
4865 (the order of @code{S} and @code{T} in the spec is not significant).
4866 There can be any number of ampersand-separated variables; for each the
4867 wild card is optional. Useful for CPP as @samp{%@{D*&U*&A*@}}.
4869 @item %@{<@code{S}@}
4870 Remove all occurrences of @code{-S} from the command line. Note---this
4871 command is position dependent. @samp{%} commands in the spec string
4872 before this option will see @code{-S}, @samp{%} commands in the spec
4873 string after this option will not.
4875 @item %@{@code{S}*:@code{X}@}
4876 Substitutes @code{X} if one or more switches whose names start with
4877 @code{-S} are specified to GCC@. Note that the tail part of the
4878 @code{-S} option (i.e.@: the part matched by the @samp{*}) will be substituted
4879 for each occurrence of @samp{%*} within @code{X}.
4881 @item %@{@code{S}:@code{X}@}
4882 Substitutes @code{X}, but only if the @samp{-S} switch was given to GCC@.
4884 @item %@{!@code{S}:@code{X}@}
4885 Substitutes @code{X}, but only if the @samp{-S} switch was @emph{not} given to GCC@.
4887 @item %@{|@code{S}:@code{X}@}
4888 Like %@{@code{S}:@code{X}@}, but if no @code{S} switch, substitute @samp{-}.
4890 @item %@{|!@code{S}:@code{X}@}
4891 Like %@{!@code{S}:@code{X}@}, but if there is an @code{S} switch, substitute @samp{-}.
4893 @item %@{.@code{S}:@code{X}@}
4894 Substitutes @code{X}, but only if processing a file with suffix @code{S}.
4896 @item %@{!.@code{S}:@code{X}@}
4897 Substitutes @code{X}, but only if @emph{not} processing a file with suffix @code{S}.
4899 @item %@{@code{S}|@code{P}:@code{X}@}
4900 Substitutes @code{X} if either @code{-S} or @code{-P} was given to GCC@. This may be
4901 combined with @samp{!} and @samp{.} sequences as well, although they
4902 have a stronger binding than the @samp{|}. For example a spec string
4906 %@{.c:-foo@} %@{!.c:-bar@} %@{.c|d:-baz@} %@{!.c|d:-boggle@}
4909 will output the following command-line options from the following input
4910 command-line options:
4915 -d fred.c -foo -baz -boggle
4916 -d jim.d -bar -baz -boggle
4921 The conditional text @code{X} in a %@{@code{S}:@code{X}@} or
4922 %@{!@code{S}:@code{X}@} construct may contain other nested @samp{%} constructs
4923 or spaces, or even newlines. They are processed as usual, as described
4926 The @option{-O}, @option{-f}, @option{-m}, and @option{-W}
4927 switches are handled specifically in these
4928 constructs. If another value of @option{-O} or the negated form of a @option{-f}, @option{-m}, or
4929 @option{-W} switch is found later in the command line, the earlier switch
4930 value is ignored, except with @{@code{S}*@} where @code{S} is just one
4931 letter, which passes all matching options.
4933 The character @samp{|} at the beginning of the predicate text is used to indicate
4934 that a command should be piped to the following command, but only if @option{-pipe}
4937 It is built into GCC which switches take arguments and which do not.
4938 (You might think it would be useful to generalize this to allow each
4939 compiler's spec to say which switches take arguments. But this cannot
4940 be done in a consistent fashion. GCC cannot even decide which input
4941 files have been specified without knowing which switches take arguments,
4942 and it must know which input files to compile in order to tell which
4945 GCC also knows implicitly that arguments starting in @option{-l} are to be
4946 treated as compiler output files, and passed to the linker in their
4947 proper position among the other output files.
4949 @c man begin OPTIONS
4951 @node Target Options
4952 @section Specifying Target Machine and Compiler Version
4953 @cindex target options
4954 @cindex cross compiling
4955 @cindex specifying machine version
4956 @cindex specifying compiler version and target machine
4957 @cindex compiler version, specifying
4958 @cindex target machine, specifying
4960 By default, GCC compiles code for the same type of machine that you
4961 are using. However, it can also be installed as a cross-compiler, to
4962 compile for some other type of machine. In fact, several different
4963 configurations of GCC, for different target machines, can be
4964 installed side by side. Then you specify which one to use with the
4967 In addition, older and newer versions of GCC can be installed side
4968 by side. One of them (probably the newest) will be the default, but
4969 you may sometimes wish to use another.
4972 @item -b @var{machine}
4974 The argument @var{machine} specifies the target machine for compilation.
4975 This is useful when you have installed GCC as a cross-compiler.
4977 The value to use for @var{machine} is the same as was specified as the
4978 machine type when configuring GCC as a cross-compiler. For
4979 example, if a cross-compiler was configured with @samp{configure
4980 i386v}, meaning to compile for an 80386 running System V, then you
4981 would specify @option{-b i386v} to run that cross compiler.
4983 When you do not specify @option{-b}, it normally means to compile for
4984 the same type of machine that you are using.
4986 @item -V @var{version}
4988 The argument @var{version} specifies which version of GCC to run.
4989 This is useful when multiple versions are installed. For example,
4990 @var{version} might be @samp{2.0}, meaning to run GCC version 2.0.
4992 The default version, when you do not specify @option{-V}, is the last
4993 version of GCC that you installed.
4996 The @option{-b} and @option{-V} options actually work by controlling part of
4997 the file name used for the executable files and libraries used for
4998 compilation. A given version of GCC, for a given target machine, is
4999 normally kept in the directory @file{/usr/local/lib/gcc-lib/@var{machine}/@var{version}}.
5001 Thus, sites can customize the effect of @option{-b} or @option{-V} either by
5002 changing the names of these directories or adding alternate names (or
5003 symbolic links). If in directory @file{/usr/local/lib/gcc-lib/} the
5004 file @file{80386} is a link to the file @file{i386v}, then @option{-b
5005 80386} becomes an alias for @option{-b i386v}.
5007 In one respect, the @option{-b} or @option{-V} do not completely change
5008 to a different compiler: the top-level driver program @command{gcc}
5009 that you originally invoked continues to run and invoke the other
5010 executables (preprocessor, compiler per se, assembler and linker)
5011 that do the real work. However, since no real work is done in the
5012 driver program, it usually does not matter that the driver program
5013 in use is not the one for the specified target. It is common for the
5014 interface to the other executables to change incompatibly between
5015 compiler versions, so unless the version specified is very close to that
5016 of the driver (for example, @option{-V 3.0} with a driver program from GCC
5017 version 3.0.1), use of @option{-V} may not work; for example, using
5018 @option{-V 2.95.2} will not work with a driver program from GCC 3.0.
5020 The only way that the driver program depends on the target machine is
5021 in the parsing and handling of special machine-specific options.
5022 However, this is controlled by a file which is found, along with the
5023 other executables, in the directory for the specified version and
5024 target machine. As a result, a single installed driver program adapts
5025 to any specified target machine, and sufficiently similar compiler
5028 The driver program executable does control one significant thing,
5029 however: the default version and target machine. Therefore, you can
5030 install different instances of the driver program, compiled for
5031 different targets or versions, under different names.
5033 For example, if the driver for version 2.0 is installed as @command{ogcc}
5034 and that for version 2.1 is installed as @command{gcc}, then the command
5035 @command{gcc} will use version 2.1 by default, while @command{ogcc} will use
5036 2.0 by default. However, you can choose either version with either
5037 command with the @option{-V} option.
5039 @node Submodel Options
5040 @section Hardware Models and Configurations
5041 @cindex submodel options
5042 @cindex specifying hardware config
5043 @cindex hardware models and configurations, specifying
5044 @cindex machine dependent options
5046 Earlier we discussed the standard option @option{-b} which chooses among
5047 different installed compilers for completely different target
5048 machines, such as VAX vs.@: 68000 vs.@: 80386.
5050 In addition, each of these target machine types can have its own
5051 special options, starting with @samp{-m}, to choose among various
5052 hardware models or configurations---for example, 68010 vs 68020,
5053 floating coprocessor or none. A single installed version of the
5054 compiler can compile for any model or configuration, according to the
5057 Some configurations of the compiler also support additional special
5058 options, usually for compatibility with other compilers on the same
5062 These options are defined by the macro @code{TARGET_SWITCHES} in the
5063 machine description. The default for the options is also defined by
5064 that macro, which enables you to change the defaults.
5079 * RS/6000 and PowerPC Options::
5082 * i386 and x86-64 Options::
5084 * Intel 960 Options::
5085 * DEC Alpha Options::
5089 * System V Options::
5090 * TMS320C3x/C4x Options::
5098 * S/390 and zSeries Options::
5101 @node M680x0 Options
5102 @subsection M680x0 Options
5103 @cindex M680x0 options
5105 These are the @samp{-m} options defined for the 68000 series. The default
5106 values for these options depends on which style of 68000 was selected when
5107 the compiler was configured; the defaults for the most common choices are
5115 Generate output for a 68000. This is the default
5116 when the compiler is configured for 68000-based systems.
5118 Use this option for microcontrollers with a 68000 or EC000 core,
5119 including the 68008, 68302, 68306, 68307, 68322, 68328 and 68356.
5125 Generate output for a 68020. This is the default
5126 when the compiler is configured for 68020-based systems.
5130 Generate output containing 68881 instructions for floating point.
5131 This is the default for most 68020 systems unless @option{--nfp} was
5132 specified when the compiler was configured.
5136 Generate output for a 68030. This is the default when the compiler is
5137 configured for 68030-based systems.
5141 Generate output for a 68040. This is the default when the compiler is
5142 configured for 68040-based systems.
5144 This option inhibits the use of 68881/68882 instructions that have to be
5145 emulated by software on the 68040. Use this option if your 68040 does not
5146 have code to emulate those instructions.
5150 Generate output for a 68060. This is the default when the compiler is
5151 configured for 68060-based systems.
5153 This option inhibits the use of 68020 and 68881/68882 instructions that
5154 have to be emulated by software on the 68060. Use this option if your 68060
5155 does not have code to emulate those instructions.
5159 Generate output for a CPU32. This is the default
5160 when the compiler is configured for CPU32-based systems.
5162 Use this option for microcontrollers with a
5163 CPU32 or CPU32+ core, including the 68330, 68331, 68332, 68333, 68334,
5164 68336, 68340, 68341, 68349 and 68360.
5168 Generate output for a 520X ``coldfire'' family cpu. This is the default
5169 when the compiler is configured for 520X-based systems.
5171 Use this option for microcontroller with a 5200 core, including
5172 the MCF5202, MCF5203, MCF5204 and MCF5202.
5177 Generate output for a 68040, without using any of the new instructions.
5178 This results in code which can run relatively efficiently on either a
5179 68020/68881 or a 68030 or a 68040. The generated code does use the
5180 68881 instructions that are emulated on the 68040.
5184 Generate output for a 68060, without using any of the new instructions.
5185 This results in code which can run relatively efficiently on either a
5186 68020/68881 or a 68030 or a 68040. The generated code does use the
5187 68881 instructions that are emulated on the 68060.
5191 Generate output containing Sun FPA instructions for floating point.
5194 @opindex msoft-float
5195 Generate output containing library calls for floating point.
5196 @strong{Warning:} the requisite libraries are not available for all m68k
5197 targets. Normally the facilities of the machine's usual C compiler are
5198 used, but this can't be done directly in cross-compilation. You must
5199 make your own arrangements to provide suitable library functions for
5200 cross-compilation. The embedded targets @samp{m68k-*-aout} and
5201 @samp{m68k-*-coff} do provide software floating point support.
5205 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5208 @opindex mnobitfield
5209 Do not use the bit-field instructions. The @option{-m68000}, @option{-mcpu32}
5210 and @option{-m5200} options imply @w{@option{-mnobitfield}}.
5214 Do use the bit-field instructions. The @option{-m68020} option implies
5215 @option{-mbitfield}. This is the default if you use a configuration
5216 designed for a 68020.
5220 Use a different function-calling convention, in which functions
5221 that take a fixed number of arguments return with the @code{rtd}
5222 instruction, which pops their arguments while returning. This
5223 saves one instruction in the caller since there is no need to pop
5224 the arguments there.
5226 This calling convention is incompatible with the one normally
5227 used on Unix, so you cannot use it if you need to call libraries
5228 compiled with the Unix compiler.
5230 Also, you must provide function prototypes for all functions that
5231 take variable numbers of arguments (including @code{printf});
5232 otherwise incorrect code will be generated for calls to those
5235 In addition, seriously incorrect code will result if you call a
5236 function with too many arguments. (Normally, extra arguments are
5237 harmlessly ignored.)
5239 The @code{rtd} instruction is supported by the 68010, 68020, 68030,
5240 68040, 68060 and CPU32 processors, but not by the 68000 or 5200.
5243 @itemx -mno-align-int
5245 @opindex mno-align-int
5246 Control whether GCC aligns @code{int}, @code{long}, @code{long long},
5247 @code{float}, @code{double}, and @code{long double} variables on a 32-bit
5248 boundary (@option{-malign-int}) or a 16-bit boundary (@option{-mno-align-int}).
5249 Aligning variables on 32-bit boundaries produces code that runs somewhat
5250 faster on processors with 32-bit busses at the expense of more memory.
5252 @strong{Warning:} if you use the @option{-malign-int} switch, GCC will
5253 align structures containing the above types differently than
5254 most published application binary interface specifications for the m68k.
5258 Use the pc-relative addressing mode of the 68000 directly, instead of
5259 using a global offset table. At present, this option implies @option{-fpic},
5260 allowing at most a 16-bit offset for pc-relative addressing. @option{-fPIC} is
5261 not presently supported with @option{-mpcrel}, though this could be supported for
5262 68020 and higher processors.
5264 @item -mno-strict-align
5265 @itemx -mstrict-align
5266 @opindex mno-strict-align
5267 @opindex mstrict-align
5268 Do not (do) assume that unaligned memory references will be handled by
5273 @node M68hc1x Options
5274 @subsection M68hc1x Options
5275 @cindex M68hc1x options
5277 These are the @samp{-m} options defined for the 68hc11 and 68hc12
5278 microcontrollers. The default values for these options depends on
5279 which style of microcontroller was selected when the compiler was configured;
5280 the defaults for the most common choices are given below.
5287 Generate output for a 68HC11. This is the default
5288 when the compiler is configured for 68HC11-based systems.
5294 Generate output for a 68HC12. This is the default
5295 when the compiler is configured for 68HC12-based systems.
5298 @opindex mauto-incdec
5299 Enable the use of 68HC12 pre and post auto-increment and auto-decrement
5304 Consider type @code{int} to be 16 bits wide, like @code{short int}.
5306 @item -msoft-reg-count=@var{count}
5307 @opindex msoft-reg-count
5308 Specify the number of pseudo-soft registers which are used for the
5309 code generation. The maximum number is 32. Using more pseudo-soft
5310 register may or may not result in better code depending on the program.
5311 The default is 4 for 68HC11 and 2 for 68HC12.
5316 @subsection VAX Options
5319 These @samp{-m} options are defined for the VAX:
5324 Do not output certain jump instructions (@code{aobleq} and so on)
5325 that the Unix assembler for the VAX cannot handle across long
5330 Do output those jump instructions, on the assumption that you
5331 will assemble with the GNU assembler.
5335 Output code for g-format floating point numbers instead of d-format.
5339 @subsection SPARC Options
5340 @cindex SPARC options
5342 These @samp{-m} switches are supported on the SPARC:
5347 @opindex mno-app-regs
5349 Specify @option{-mapp-regs} to generate output using the global registers
5350 2 through 4, which the SPARC SVR4 ABI reserves for applications. This
5353 To be fully SVR4 ABI compliant at the cost of some performance loss,
5354 specify @option{-mno-app-regs}. You should compile libraries and system
5355 software with this option.
5360 @opindex mhard-float
5361 Generate output containing floating point instructions. This is the
5367 @opindex msoft-float
5368 Generate output containing library calls for floating point.
5369 @strong{Warning:} the requisite libraries are not available for all SPARC
5370 targets. Normally the facilities of the machine's usual C compiler are
5371 used, but this cannot be done directly in cross-compilation. You must make
5372 your own arrangements to provide suitable library functions for
5373 cross-compilation. The embedded targets @samp{sparc-*-aout} and
5374 @samp{sparclite-*-*} do provide software floating point support.
5376 @option{-msoft-float} changes the calling convention in the output file;
5377 therefore, it is only useful if you compile @emph{all} of a program with
5378 this option. In particular, you need to compile @file{libgcc.a}, the
5379 library that comes with GCC, with @option{-msoft-float} in order for
5382 @item -mhard-quad-float
5383 @opindex mhard-quad-float
5384 Generate output containing quad-word (long double) floating point
5387 @item -msoft-quad-float
5388 @opindex msoft-quad-float
5389 Generate output containing library calls for quad-word (long double)
5390 floating point instructions. The functions called are those specified
5391 in the SPARC ABI@. This is the default.
5393 As of this writing, there are no sparc implementations that have hardware
5394 support for the quad-word floating point instructions. They all invoke
5395 a trap handler for one of these instructions, and then the trap handler
5396 emulates the effect of the instruction. Because of the trap handler overhead,
5397 this is much slower than calling the ABI library routines. Thus the
5398 @option{-msoft-quad-float} option is the default.
5402 @opindex mno-epilogue
5404 With @option{-mepilogue} (the default), the compiler always emits code for
5405 function exit at the end of each function. Any function exit in
5406 the middle of the function (such as a return statement in C) will
5407 generate a jump to the exit code at the end of the function.
5409 With @option{-mno-epilogue}, the compiler tries to emit exit code inline
5410 at every function exit.
5416 With @option{-mflat}, the compiler does not generate save/restore instructions
5417 and will use a ``flat'' or single register window calling convention.
5418 This model uses %i7 as the frame pointer and is compatible with the normal
5419 register window model. Code from either may be intermixed.
5420 The local registers and the input registers (0--5) are still treated as
5421 ``call saved'' registers and will be saved on the stack as necessary.
5423 With @option{-mno-flat} (the default), the compiler emits save/restore
5424 instructions (except for leaf functions) and is the normal mode of operation.
5426 @item -mno-unaligned-doubles
5427 @itemx -munaligned-doubles
5428 @opindex mno-unaligned-doubles
5429 @opindex munaligned-doubles
5430 Assume that doubles have 8 byte alignment. This is the default.
5432 With @option{-munaligned-doubles}, GCC assumes that doubles have 8 byte
5433 alignment only if they are contained in another type, or if they have an
5434 absolute address. Otherwise, it assumes they have 4 byte alignment.
5435 Specifying this option avoids some rare compatibility problems with code
5436 generated by other compilers. It is not the default because it results
5437 in a performance loss, especially for floating point code.
5439 @item -mno-faster-structs
5440 @itemx -mfaster-structs
5441 @opindex mno-faster-structs
5442 @opindex mfaster-structs
5443 With @option{-mfaster-structs}, the compiler assumes that structures
5444 should have 8 byte alignment. This enables the use of pairs of
5445 @code{ldd} and @code{std} instructions for copies in structure
5446 assignment, in place of twice as many @code{ld} and @code{st} pairs.
5447 However, the use of this changed alignment directly violates the Sparc
5448 ABI@. Thus, it's intended only for use on targets where the developer
5449 acknowledges that their resulting code will not be directly in line with
5450 the rules of the ABI@.
5456 These two options select variations on the SPARC architecture.
5458 By default (unless specifically configured for the Fujitsu SPARClite),
5459 GCC generates code for the v7 variant of the SPARC architecture.
5461 @option{-mv8} will give you SPARC v8 code. The only difference from v7
5462 code is that the compiler emits the integer multiply and integer
5463 divide instructions which exist in SPARC v8 but not in SPARC v7.
5465 @option{-msparclite} will give you SPARClite code. This adds the integer
5466 multiply, integer divide step and scan (@code{ffs}) instructions which
5467 exist in SPARClite but not in SPARC v7.
5469 These options are deprecated and will be deleted in a future GCC release.
5470 They have been replaced with @option{-mcpu=xxx}.
5475 @opindex msupersparc
5476 These two options select the processor for which the code is optimised.
5478 With @option{-mcypress} (the default), the compiler optimizes code for the
5479 Cypress CY7C602 chip, as used in the SparcStation/SparcServer 3xx series.
5480 This is also appropriate for the older SparcStation 1, 2, IPX etc.
5482 With @option{-msupersparc} the compiler optimizes code for the SuperSparc cpu, as
5483 used in the SparcStation 10, 1000 and 2000 series. This flag also enables use
5484 of the full SPARC v8 instruction set.
5486 These options are deprecated and will be deleted in a future GCC release.
5487 They have been replaced with @option{-mcpu=xxx}.
5489 @item -mcpu=@var{cpu_type}
5491 Set the instruction set, register set, and instruction scheduling parameters
5492 for machine type @var{cpu_type}. Supported values for @var{cpu_type} are
5493 @samp{v7}, @samp{cypress}, @samp{v8}, @samp{supersparc}, @samp{sparclite},
5494 @samp{hypersparc}, @samp{sparclite86x}, @samp{f930}, @samp{f934},
5495 @samp{sparclet}, @samp{tsc701}, @samp{v9}, and @samp{ultrasparc}.
5497 Default instruction scheduling parameters are used for values that select
5498 an architecture and not an implementation. These are @samp{v7}, @samp{v8},
5499 @samp{sparclite}, @samp{sparclet}, @samp{v9}.
5501 Here is a list of each supported architecture and their supported
5506 v8: supersparc, hypersparc
5507 sparclite: f930, f934, sparclite86x
5512 @item -mtune=@var{cpu_type}
5514 Set the instruction scheduling parameters for machine type
5515 @var{cpu_type}, but do not set the instruction set or register set that the
5516 option @option{-mcpu=@var{cpu_type}} would.
5518 The same values for @option{-mcpu=@var{cpu_type}} can be used for
5519 @option{-mtune=@var{cpu_type}}, but the only useful values are those
5520 that select a particular cpu implementation. Those are @samp{cypress},
5521 @samp{supersparc}, @samp{hypersparc}, @samp{f930}, @samp{f934},
5522 @samp{sparclite86x}, @samp{tsc701}, and @samp{ultrasparc}.
5526 These @samp{-m} switches are supported in addition to the above
5527 on the SPARCLET processor.
5530 @item -mlittle-endian
5531 @opindex mlittle-endian
5532 Generate code for a processor running in little-endian mode.
5536 Treat register @code{%g0} as a normal register.
5537 GCC will continue to clobber it as necessary but will not assume
5538 it always reads as 0.
5540 @item -mbroken-saverestore
5541 @opindex mbroken-saverestore
5542 Generate code that does not use non-trivial forms of the @code{save} and
5543 @code{restore} instructions. Early versions of the SPARCLET processor do
5544 not correctly handle @code{save} and @code{restore} instructions used with
5545 arguments. They correctly handle them used without arguments. A @code{save}
5546 instruction used without arguments increments the current window pointer
5547 but does not allocate a new stack frame. It is assumed that the window
5548 overflow trap handler will properly handle this case as will interrupt
5552 These @samp{-m} switches are supported in addition to the above
5553 on SPARC V9 processors in 64-bit environments.
5556 @item -mlittle-endian
5557 @opindex mlittle-endian
5558 Generate code for a processor running in little-endian mode.
5564 Generate code for a 32-bit or 64-bit environment.
5565 The 32-bit environment sets int, long and pointer to 32 bits.
5566 The 64-bit environment sets int to 32 bits and long and pointer
5569 @item -mcmodel=medlow
5570 @opindex mcmodel=medlow
5571 Generate code for the Medium/Low code model: the program must be linked
5572 in the low 32 bits of the address space. Pointers are 64 bits.
5573 Programs can be statically or dynamically linked.
5575 @item -mcmodel=medmid
5576 @opindex mcmodel=medmid
5577 Generate code for the Medium/Middle code model: the program must be linked
5578 in the low 44 bits of the address space, the text segment must be less than
5579 2G bytes, and data segment must be within 2G of the text segment.
5580 Pointers are 64 bits.
5582 @item -mcmodel=medany
5583 @opindex mcmodel=medany
5584 Generate code for the Medium/Anywhere code model: the program may be linked
5585 anywhere in the address space, the text segment must be less than
5586 2G bytes, and data segment must be within 2G of the text segment.
5587 Pointers are 64 bits.
5589 @item -mcmodel=embmedany
5590 @opindex mcmodel=embmedany
5591 Generate code for the Medium/Anywhere code model for embedded systems:
5592 assume a 32-bit text and a 32-bit data segment, both starting anywhere
5593 (determined at link time). Register %g4 points to the base of the
5594 data segment. Pointers are still 64 bits.
5595 Programs are statically linked, PIC is not supported.
5598 @itemx -mno-stack-bias
5599 @opindex mstack-bias
5600 @opindex mno-stack-bias
5601 With @option{-mstack-bias}, GCC assumes that the stack pointer, and
5602 frame pointer if present, are offset by @minus{}2047 which must be added back
5603 when making stack frame references.
5604 Otherwise, assume no such offset is present.
5607 @node Convex Options
5608 @subsection Convex Options
5609 @cindex Convex options
5611 These @samp{-m} options are defined for Convex:
5616 Generate output for C1. The code will run on any Convex machine.
5617 The preprocessor symbol @code{__convex__c1__} is defined.
5621 Generate output for C2. Uses instructions not available on C1.
5622 Scheduling and other optimizations are chosen for max performance on C2.
5623 The preprocessor symbol @code{__convex_c2__} is defined.
5627 Generate output for C32xx. Uses instructions not available on C1.
5628 Scheduling and other optimizations are chosen for max performance on C32.
5629 The preprocessor symbol @code{__convex_c32__} is defined.
5633 Generate output for C34xx. Uses instructions not available on C1.
5634 Scheduling and other optimizations are chosen for max performance on C34.
5635 The preprocessor symbol @code{__convex_c34__} is defined.
5639 Generate output for C38xx. Uses instructions not available on C1.
5640 Scheduling and other optimizations are chosen for max performance on C38.
5641 The preprocessor symbol @code{__convex_c38__} is defined.
5645 Generate code which puts an argument count in the word preceding each
5646 argument list. This is compatible with regular CC, and a few programs
5647 may need the argument count word. GDB and other source-level debuggers
5648 do not need it; this info is in the symbol table.
5651 @opindex mnoargcount
5652 Omit the argument count word. This is the default.
5654 @item -mvolatile-cache
5655 @opindex mvolatile-cache
5656 Allow volatile references to be cached. This is the default.
5658 @item -mvolatile-nocache
5659 @opindex mvolatile-nocache
5660 Volatile references bypass the data cache, going all the way to memory.
5661 This is only needed for multi-processor code that does not use standard
5662 synchronization instructions. Making non-volatile references to volatile
5663 locations will not necessarily work.
5667 Type long is 32 bits, the same as type int. This is the default.
5671 Type long is 64 bits, the same as type long long. This option is useless,
5672 because no library support exists for it.
5675 @node AMD29K Options
5676 @subsection AMD29K Options
5677 @cindex AMD29K options
5679 These @samp{-m} options are defined for the AMD Am29000:
5684 @cindex DW bit (29k)
5685 Generate code that assumes the @code{DW} bit is set, i.e., that byte and
5686 halfword operations are directly supported by the hardware. This is the
5691 Generate code that assumes the @code{DW} bit is not set.
5695 @cindex byte writes (29k)
5696 Generate code that assumes the system supports byte and halfword write
5697 operations. This is the default.
5701 Generate code that assumes the systems does not support byte and
5702 halfword write operations. @option{-mnbw} implies @option{-mndw}.
5706 @cindex memory model (29k)
5707 Use a small memory model that assumes that all function addresses are
5708 either within a single 256 KB segment or at an absolute address of less
5709 than 256k. This allows the @code{call} instruction to be used instead
5710 of a @code{const}, @code{consth}, @code{calli} sequence.
5714 Use the normal memory model: Generate @code{call} instructions only when
5715 calling functions in the same file and @code{calli} instructions
5716 otherwise. This works if each file occupies less than 256 KB but allows
5717 the entire executable to be larger than 256 KB@. This is the default.
5721 Always use @code{calli} instructions. Specify this option if you expect
5722 a single file to compile into more than 256 KB of code.
5726 @cindex processor selection (29k)
5727 Generate code for the Am29050.
5731 Generate code for the Am29000. This is the default.
5733 @item -mkernel-registers
5734 @opindex mkernel-registers
5735 @cindex kernel and user registers (29k)
5736 Generate references to registers @code{gr64-gr95} instead of to
5737 registers @code{gr96-gr127}. This option can be used when compiling
5738 kernel code that wants a set of global registers disjoint from that used
5741 Note that when this option is used, register names in @samp{-f} flags
5742 must use the normal, user-mode, names.
5744 @item -muser-registers
5745 @opindex muser-registers
5746 Use the normal set of global registers, @code{gr96-gr127}. This is the
5750 @itemx -mno-stack-check
5751 @opindex mstack-check
5752 @opindex mno-stack-check
5753 @cindex stack checks (29k)
5754 Insert (or do not insert) a call to @code{__msp_check} after each stack
5755 adjustment. This is often used for kernel code.
5758 @itemx -mno-storem-bug
5759 @opindex mstorem-bug
5760 @opindex mno-storem-bug
5761 @cindex storem bug (29k)
5762 @option{-mstorem-bug} handles 29k processors which cannot handle the
5763 separation of a mtsrim insn and a storem instruction (most 29000 chips
5764 to date, but not the 29050).
5766 @item -mno-reuse-arg-regs
5767 @itemx -mreuse-arg-regs
5768 @opindex mno-reuse-arg-regs
5769 @opindex mreuse-arg-regs
5770 @option{-mno-reuse-arg-regs} tells the compiler to only use incoming argument
5771 registers for copying out arguments. This helps detect calling a function
5772 with fewer arguments than it was declared with.
5774 @item -mno-impure-text
5775 @itemx -mimpure-text
5776 @opindex mno-impure-text
5777 @opindex mimpure-text
5778 @option{-mimpure-text}, used in addition to @option{-shared}, tells the compiler to
5779 not pass @option{-assert pure-text} to the linker when linking a shared object.
5782 @opindex msoft-float
5783 Generate output containing library calls for floating point.
5784 @strong{Warning:} the requisite libraries are not part of GCC@.
5785 Normally the facilities of the machine's usual C compiler are used, but
5786 this can't be done directly in cross-compilation. You must make your
5787 own arrangements to provide suitable library functions for
5792 Do not generate multm or multmu instructions. This is useful for some embedded
5793 systems which do not have trap handlers for these instructions.
5797 @subsection ARM Options
5800 These @samp{-m} options are defined for Advanced RISC Machines (ARM)
5805 @opindex mapcs-frame
5806 Generate a stack frame that is compliant with the ARM Procedure Call
5807 Standard for all functions, even if this is not strictly necessary for
5808 correct execution of the code. Specifying @option{-fomit-frame-pointer}
5809 with this option will cause the stack frames not to be generated for
5810 leaf functions. The default is @option{-mno-apcs-frame}.
5814 This is a synonym for @option{-mapcs-frame}.
5818 Generate code for a processor running with a 26-bit program counter,
5819 and conforming to the function calling standards for the APCS 26-bit
5820 option. This option replaces the @option{-m2} and @option{-m3} options
5821 of previous releases of the compiler.
5825 Generate code for a processor running with a 32-bit program counter,
5826 and conforming to the function calling standards for the APCS 32-bit
5827 option. This option replaces the @option{-m6} option of previous releases
5831 @c not currently implemented
5832 @item -mapcs-stack-check
5833 @opindex mapcs-stack-check
5834 Generate code to check the amount of stack space available upon entry to
5835 every function (that actually uses some stack space). If there is
5836 insufficient space available then either the function
5837 @samp{__rt_stkovf_split_small} or @samp{__rt_stkovf_split_big} will be
5838 called, depending upon the amount of stack space required. The run time
5839 system is required to provide these functions. The default is
5840 @option{-mno-apcs-stack-check}, since this produces smaller code.
5842 @c not currently implemented
5844 @opindex mapcs-float
5845 Pass floating point arguments using the float point registers. This is
5846 one of the variants of the APCS@. This option is recommended if the
5847 target hardware has a floating point unit or if a lot of floating point
5848 arithmetic is going to be performed by the code. The default is
5849 @option{-mno-apcs-float}, since integer only code is slightly increased in
5850 size if @option{-mapcs-float} is used.
5852 @c not currently implemented
5853 @item -mapcs-reentrant
5854 @opindex mapcs-reentrant
5855 Generate reentrant, position independent code. The default is
5856 @option{-mno-apcs-reentrant}.
5859 @item -mthumb-interwork
5860 @opindex mthumb-interwork
5861 Generate code which supports calling between the ARM and Thumb
5862 instruction sets. Without this option the two instruction sets cannot
5863 be reliably used inside one program. The default is
5864 @option{-mno-thumb-interwork}, since slightly larger code is generated
5865 when @option{-mthumb-interwork} is specified.
5867 @item -mno-sched-prolog
5868 @opindex mno-sched-prolog
5869 Prevent the reordering of instructions in the function prolog, or the
5870 merging of those instruction with the instructions in the function's
5871 body. This means that all functions will start with a recognizable set
5872 of instructions (or in fact one of a choice from a small set of
5873 different function prologues), and this information can be used to
5874 locate the start if functions inside an executable piece of code. The
5875 default is @option{-msched-prolog}.
5878 @opindex mhard-float
5879 Generate output containing floating point instructions. This is the
5883 @opindex msoft-float
5884 Generate output containing library calls for floating point.
5885 @strong{Warning:} the requisite libraries are not available for all ARM
5886 targets. Normally the facilities of the machine's usual C compiler are
5887 used, but this cannot be done directly in cross-compilation. You must make
5888 your own arrangements to provide suitable library functions for
5891 @option{-msoft-float} changes the calling convention in the output file;
5892 therefore, it is only useful if you compile @emph{all} of a program with
5893 this option. In particular, you need to compile @file{libgcc.a}, the
5894 library that comes with GCC, with @option{-msoft-float} in order for
5897 @item -mlittle-endian
5898 @opindex mlittle-endian
5899 Generate code for a processor running in little-endian mode. This is
5900 the default for all standard configurations.
5903 @opindex mbig-endian
5904 Generate code for a processor running in big-endian mode; the default is
5905 to compile code for a little-endian processor.
5907 @item -mwords-little-endian
5908 @opindex mwords-little-endian
5909 This option only applies when generating code for big-endian processors.
5910 Generate code for a little-endian word order but a big-endian byte
5911 order. That is, a byte order of the form @samp{32107654}. Note: this
5912 option should only be used if you require compatibility with code for
5913 big-endian ARM processors generated by versions of the compiler prior to
5916 @item -malignment-traps
5917 @opindex malignment-traps
5918 Generate code that will not trap if the MMU has alignment traps enabled.
5919 On ARM architectures prior to ARMv4, there were no instructions to
5920 access half-word objects stored in memory. However, when reading from
5921 memory a feature of the ARM architecture allows a word load to be used,
5922 even if the address is unaligned, and the processor core will rotate the
5923 data as it is being loaded. This option tells the compiler that such
5924 misaligned accesses will cause a MMU trap and that it should instead
5925 synthesise the access as a series of byte accesses. The compiler can
5926 still use word accesses to load half-word data if it knows that the
5927 address is aligned to a word boundary.
5929 This option is ignored when compiling for ARM architecture 4 or later,
5930 since these processors have instructions to directly access half-word
5933 @item -mno-alignment-traps
5934 @opindex mno-alignment-traps
5935 Generate code that assumes that the MMU will not trap unaligned
5936 accesses. This produces better code when the target instruction set
5937 does not have half-word memory operations (i.e.@: implementations prior to
5940 Note that you cannot use this option to access unaligned word objects,
5941 since the processor will only fetch one 32-bit aligned object from
5944 The default setting for most targets is @option{-mno-alignment-traps}, since
5945 this produces better code when there are no half-word memory
5946 instructions available.
5948 @item -mshort-load-bytes
5949 @itemx -mno-short-load-words
5950 @opindex mshort-load-bytes
5951 @opindex mno-short-load-words
5952 These are deprecated aliases for @option{-malignment-traps}.
5954 @item -mno-short-load-bytes
5955 @itemx -mshort-load-words
5956 @opindex mno-short-load-bytes
5957 @opindex mshort-load-words
5958 This are deprecated aliases for @option{-mno-alignment-traps}.
5962 This option only applies to RISC iX@. Emulate the native BSD-mode
5963 compiler. This is the default if @option{-ansi} is not specified.
5967 This option only applies to RISC iX@. Emulate the native X/Open-mode
5970 @item -mno-symrename
5971 @opindex mno-symrename
5972 This option only applies to RISC iX@. Do not run the assembler
5973 post-processor, @samp{symrename}, after code has been assembled.
5974 Normally it is necessary to modify some of the standard symbols in
5975 preparation for linking with the RISC iX C library; this option
5976 suppresses this pass. The post-processor is never run when the
5977 compiler is built for cross-compilation.
5979 @item -mcpu=@var{name}
5981 This specifies the name of the target ARM processor. GCC uses this name
5982 to determine what kind of instructions it can emit when generating
5983 assembly code. Permissible names are: @samp{arm2}, @samp{arm250},
5984 @samp{arm3}, @samp{arm6}, @samp{arm60}, @samp{arm600}, @samp{arm610},
5985 @samp{arm620}, @samp{arm7}, @samp{arm7m}, @samp{arm7d}, @samp{arm7dm},
5986 @samp{arm7di}, @samp{arm7dmi}, @samp{arm70}, @samp{arm700},
5987 @samp{arm700i}, @samp{arm710}, @samp{arm710c}, @samp{arm7100},
5988 @samp{arm7500}, @samp{arm7500fe}, @samp{arm7tdmi}, @samp{arm8},
5989 @samp{strongarm}, @samp{strongarm110}, @samp{strongarm1100},
5990 @samp{arm8}, @samp{arm810}, @samp{arm9}, @samp{arm9e}, @samp{arm920},
5991 @samp{arm920t}, @samp{arm940t}, @samp{arm9tdmi}, @samp{arm10tdmi},
5992 @samp{arm1020t}, @samp{xscale}.
5994 @itemx -mtune=@var{name}
5996 This option is very similar to the @option{-mcpu=} option, except that
5997 instead of specifying the actual target processor type, and hence
5998 restricting which instructions can be used, it specifies that GCC should
5999 tune the performance of the code as if the target were of the type
6000 specified in this option, but still choosing the instructions that it
6001 will generate based on the cpu specified by a @option{-mcpu=} option.
6002 For some ARM implementations better performance can be obtained by using
6005 @item -march=@var{name}
6007 This specifies the name of the target ARM architecture. GCC uses this
6008 name to determine what kind of instructions it can emit when generating
6009 assembly code. This option can be used in conjunction with or instead
6010 of the @option{-mcpu=} option. Permissible names are: @samp{armv2},
6011 @samp{armv2a}, @samp{armv3}, @samp{armv3m}, @samp{armv4}, @samp{armv4t},
6012 @samp{armv5}, @samp{armv5t}, @samp{armv5te}.
6014 @item -mfpe=@var{number}
6015 @itemx -mfp=@var{number}
6018 This specifies the version of the floating point emulation available on
6019 the target. Permissible values are 2 and 3. @option{-mfp=} is a synonym
6020 for @option{-mfpe=}, for compatibility with older versions of GCC@.
6022 @item -mstructure-size-boundary=@var{n}
6023 @opindex mstructure-size-boundary
6024 The size of all structures and unions will be rounded up to a multiple
6025 of the number of bits set by this option. Permissible values are 8 and
6026 32. The default value varies for different toolchains. For the COFF
6027 targeted toolchain the default value is 8. Specifying the larger number
6028 can produce faster, more efficient code, but can also increase the size
6029 of the program. The two values are potentially incompatible. Code
6030 compiled with one value cannot necessarily expect to work with code or
6031 libraries compiled with the other value, if they exchange information
6032 using structures or unions.
6034 @item -mabort-on-noreturn
6035 @opindex mabort-on-noreturn
6036 Generate a call to the function @code{abort} at the end of a
6037 @code{noreturn} function. It will be executed if the function tries to
6041 @itemx -mno-long-calls
6042 @opindex mlong-calls
6043 @opindex mno-long-calls
6044 Tells the compiler to perform function calls by first loading the
6045 address of the function into a register and then performing a subroutine
6046 call on this register. This switch is needed if the target function
6047 will lie outside of the 64 megabyte addressing range of the offset based
6048 version of subroutine call instruction.
6050 Even if this switch is enabled, not all function calls will be turned
6051 into long calls. The heuristic is that static functions, functions
6052 which have the @samp{short-call} attribute, functions that are inside
6053 the scope of a @samp{#pragma no_long_calls} directive and functions whose
6054 definitions have already been compiled within the current compilation
6055 unit, will not be turned into long calls. The exception to this rule is
6056 that weak function definitions, functions with the @samp{long-call}
6057 attribute or the @samp{section} attribute, and functions that are within
6058 the scope of a @samp{#pragma long_calls} directive, will always be
6059 turned into long calls.
6061 This feature is not enabled by default. Specifying
6062 @option{-mno-long-calls} will restore the default behaviour, as will
6063 placing the function calls within the scope of a @samp{#pragma
6064 long_calls_off} directive. Note these switches have no effect on how
6065 the compiler generates code to handle function calls via function
6068 @item -mnop-fun-dllimport
6069 @opindex mnop-fun-dllimport
6070 Disable support for the @code{dllimport} attribute.
6072 @item -msingle-pic-base
6073 @opindex msingle-pic-base
6074 Treat the register used for PIC addressing as read-only, rather than
6075 loading it in the prologue for each function. The run-time system is
6076 responsible for initialising this register with an appropriate value
6077 before execution begins.
6079 @item -mpic-register=@var{reg}
6080 @opindex mpic-register
6081 Specify the register to be used for PIC addressing. The default is R10
6082 unless stack-checking is enabled, when R9 is used.
6084 @item -mpoke-function-name
6085 @opindex mpoke-function-name
6086 Write the name of each function into the text section, directly
6087 preceding the function prologue. The generated code is similar to this:
6091 .ascii "arm_poke_function_name", 0
6094 .word 0xff000000 + (t1 - t0)
6095 arm_poke_function_name
6097 stmfd sp!, @{fp, ip, lr, pc@}
6101 When performing a stack backtrace, code can inspect the value of
6102 @code{pc} stored at @code{fp + 0}. If the trace function then looks at
6103 location @code{pc - 12} and the top 8 bits are set, then we know that
6104 there is a function name embedded immediately preceding this location
6105 and has length @code{((pc[-3]) & 0xff000000)}.
6109 Generate code for the 16-bit Thumb instruction set. The default is to
6110 use the 32-bit ARM instruction set.
6113 @opindex mtpcs-frame
6114 Generate a stack frame that is compliant with the Thumb Procedure Call
6115 Standard for all non-leaf functions. (A leaf function is one that does
6116 not call any other functions.) The default is @option{-mno-tpcs-frame}.
6118 @item -mtpcs-leaf-frame
6119 @opindex mtpcs-leaf-frame
6120 Generate a stack frame that is compliant with the Thumb Procedure Call
6121 Standard for all leaf functions. (A leaf function is one that does
6122 not call any other functions.) The default is @option{-mno-apcs-leaf-frame}.
6124 @item -mcallee-super-interworking
6125 @opindex mcallee-super-interworking
6126 Gives all externally visible functions in the file being compiled an ARM
6127 instruction set header which switches to Thumb mode before executing the
6128 rest of the function. This allows these functions to be called from
6129 non-interworking code.
6131 @item -mcaller-super-interworking
6132 @opindex mcaller-super-interworking
6133 Allows calls via function pointers (including virtual functions) to
6134 execute correctly regardless of whether the target code has been
6135 compiled for interworking or not. There is a small overhead in the cost
6136 of executing a function pointer if this option is enabled.
6140 @node MN10200 Options
6141 @subsection MN10200 Options
6142 @cindex MN10200 options
6143 These @option{-m} options are defined for Matsushita MN10200 architectures:
6148 Indicate to the linker that it should perform a relaxation optimization pass
6149 to shorten branches, calls and absolute memory addresses. This option only
6150 has an effect when used on the command line for the final link step.
6152 This option makes symbolic debugging impossible.
6155 @node MN10300 Options
6156 @subsection MN10300 Options
6157 @cindex MN10300 options
6158 These @option{-m} options are defined for Matsushita MN10300 architectures:
6163 Generate code to avoid bugs in the multiply instructions for the MN10300
6164 processors. This is the default.
6167 @opindex mno-mult-bug
6168 Do not generate code to avoid bugs in the multiply instructions for the
6173 Generate code which uses features specific to the AM33 processor.
6177 Do not generate code which uses features specific to the AM33 processor. This
6182 Do not link in the C run-time initialization object file.
6186 Indicate to the linker that it should perform a relaxation optimization pass
6187 to shorten branches, calls and absolute memory addresses. This option only
6188 has an effect when used on the command line for the final link step.
6190 This option makes symbolic debugging impossible.
6194 @node M32R/D Options
6195 @subsection M32R/D Options
6196 @cindex M32R/D options
6198 These @option{-m} options are defined for Mitsubishi M32R/D architectures:
6201 @item -mcode-model=small
6202 @opindex mcode-model=small
6203 Assume all objects live in the lower 16MB of memory (so that their addresses
6204 can be loaded with the @code{ld24} instruction), and assume all subroutines
6205 are reachable with the @code{bl} instruction.
6206 This is the default.
6208 The addressability of a particular object can be set with the
6209 @code{model} attribute.
6211 @item -mcode-model=medium
6212 @opindex mcode-model=medium
6213 Assume objects may be anywhere in the 32-bit address space (the compiler
6214 will generate @code{seth/add3} instructions to load their addresses), and
6215 assume all subroutines are reachable with the @code{bl} instruction.
6217 @item -mcode-model=large
6218 @opindex mcode-model=large
6219 Assume objects may be anywhere in the 32-bit address space (the compiler
6220 will generate @code{seth/add3} instructions to load their addresses), and
6221 assume subroutines may not be reachable with the @code{bl} instruction
6222 (the compiler will generate the much slower @code{seth/add3/jl}
6223 instruction sequence).
6226 @opindex msdata=none
6227 Disable use of the small data area. Variables will be put into
6228 one of @samp{.data}, @samp{bss}, or @samp{.rodata} (unless the
6229 @code{section} attribute has been specified).
6230 This is the default.
6232 The small data area consists of sections @samp{.sdata} and @samp{.sbss}.
6233 Objects may be explicitly put in the small data area with the
6234 @code{section} attribute using one of these sections.
6237 @opindex msdata=sdata
6238 Put small global and static data in the small data area, but do not
6239 generate special code to reference them.
6243 Put small global and static data in the small data area, and generate
6244 special instructions to reference them.
6248 @cindex smaller data references
6249 Put global and static objects less than or equal to @var{num} bytes
6250 into the small data or bss sections instead of the normal data or bss
6251 sections. The default value of @var{num} is 8.
6252 The @option{-msdata} option must be set to one of @samp{sdata} or @samp{use}
6253 for this option to have any effect.
6255 All modules should be compiled with the same @option{-G @var{num}} value.
6256 Compiling with different values of @var{num} may or may not work; if it
6257 doesn't the linker will give an error message---incorrect code will not be
6263 @subsection M88K Options
6264 @cindex M88k options
6266 These @samp{-m} options are defined for Motorola 88k architectures:
6271 Generate code that works well on both the m88100 and the
6276 Generate code that works best for the m88100, but that also
6281 Generate code that works best for the m88110, and may not run
6286 Obsolete option to be removed from the next revision.
6289 @item -midentify-revision
6290 @opindex midentify-revision
6291 @cindex identifying source, compiler (88k)
6292 Include an @code{ident} directive in the assembler output recording the
6293 source file name, compiler name and version, timestamp, and compilation
6296 @item -mno-underscores
6297 @opindex mno-underscores
6298 @cindex underscores, avoiding (88k)
6299 In assembler output, emit symbol names without adding an underscore
6300 character at the beginning of each name. The default is to use an
6301 underscore as prefix on each name.
6303 @item -mocs-debug-info
6304 @itemx -mno-ocs-debug-info
6305 @opindex mocs-debug-info
6306 @opindex mno-ocs-debug-info
6308 @cindex debugging, 88k OCS
6309 Include (or omit) additional debugging information (about registers used
6310 in each stack frame) as specified in the 88open Object Compatibility
6311 Standard, ``OCS''@. This extra information allows debugging of code that
6312 has had the frame pointer eliminated. The default for DG/UX, SVr4, and
6313 Delta 88 SVr3.2 is to include this information; other 88k configurations
6314 omit this information by default.
6316 @item -mocs-frame-position
6317 @opindex mocs-frame-position
6318 @cindex register positions in frame (88k)
6319 When emitting COFF debugging information for automatic variables and
6320 parameters stored on the stack, use the offset from the canonical frame
6321 address, which is the stack pointer (register 31) on entry to the
6322 function. The DG/UX, SVr4, Delta88 SVr3.2, and BCS configurations use
6323 @option{-mocs-frame-position}; other 88k configurations have the default
6324 @option{-mno-ocs-frame-position}.
6326 @item -mno-ocs-frame-position
6327 @opindex mno-ocs-frame-position
6328 @cindex register positions in frame (88k)
6329 When emitting COFF debugging information for automatic variables and
6330 parameters stored on the stack, use the offset from the frame pointer
6331 register (register 30). When this option is in effect, the frame
6332 pointer is not eliminated when debugging information is selected by the
6335 @item -moptimize-arg-area
6336 @opindex moptimize-arg-area
6337 @cindex arguments in frame (88k)
6338 Save space by reorganizing the stack frame. This option generates code
6339 that does not agree with the 88open specifications, but uses less
6342 @itemx -mno-optimize-arg-area
6343 @opindex mno-optimize-arg-area
6344 Do not reorganize the stack frame to save space. This is the default.
6345 The generated conforms to the specification, but uses more memory.
6347 @item -mshort-data-@var{num}
6348 @opindex mshort-data
6349 @cindex smaller data references (88k)
6350 @cindex r0-relative references (88k)
6351 Generate smaller data references by making them relative to @code{r0},
6352 which allows loading a value using a single instruction (rather than the
6353 usual two). You control which data references are affected by
6354 specifying @var{num} with this option. For example, if you specify
6355 @option{-mshort-data-512}, then the data references affected are those
6356 involving displacements of less than 512 bytes.
6357 @option{-mshort-data-@var{num}} is not effective for @var{num} greater
6360 @item -mserialize-volatile
6361 @opindex mserialize-volatile
6362 @itemx -mno-serialize-volatile
6363 @opindex mno-serialize-volatile
6364 @cindex sequential consistency on 88k
6365 Do, or don't, generate code to guarantee sequential consistency
6366 of volatile memory references. By default, consistency is
6369 The order of memory references made by the MC88110 processor does
6370 not always match the order of the instructions requesting those
6371 references. In particular, a load instruction may execute before
6372 a preceding store instruction. Such reordering violates
6373 sequential consistency of volatile memory references, when there
6374 are multiple processors. When consistency must be guaranteed,
6375 GCC generates special instructions, as needed, to force
6376 execution in the proper order.
6378 The MC88100 processor does not reorder memory references and so
6379 always provides sequential consistency. However, by default, GCC
6380 generates the special instructions to guarantee consistency
6381 even when you use @option{-m88100}, so that the code may be run on an
6382 MC88110 processor. If you intend to run your code only on the
6383 MC88100 processor, you may use @option{-mno-serialize-volatile}.
6385 The extra code generated to guarantee consistency may affect the
6386 performance of your application. If you know that you can safely
6387 forgo this guarantee, you may use @option{-mno-serialize-volatile}.
6393 @cindex assembler syntax, 88k
6395 Turn on (@option{-msvr4}) or off (@option{-msvr3}) compiler extensions
6396 related to System V release 4 (SVr4). This controls the following:
6400 Which variant of the assembler syntax to emit.
6402 @option{-msvr4} makes the C preprocessor recognize @samp{#pragma weak}
6403 that is used on System V release 4.
6405 @option{-msvr4} makes GCC issue additional declaration directives used in
6409 @option{-msvr4} is the default for the m88k-motorola-sysv4 and
6410 m88k-dg-dgux m88k configurations. @option{-msvr3} is the default for all
6411 other m88k configurations.
6413 @item -mversion-03.00
6414 @opindex mversion-03.00
6415 This option is obsolete, and is ignored.
6416 @c ??? which asm syntax better for GAS? option there too?
6418 @item -mno-check-zero-division
6419 @itemx -mcheck-zero-division
6420 @opindex mno-check-zero-division
6421 @opindex mcheck-zero-division
6422 @cindex zero division on 88k
6423 Do, or don't, generate code to guarantee that integer division by
6424 zero will be detected. By default, detection is guaranteed.
6426 Some models of the MC88100 processor fail to trap upon integer
6427 division by zero under certain conditions. By default, when
6428 compiling code that might be run on such a processor, GCC
6429 generates code that explicitly checks for zero-valued divisors
6430 and traps with exception number 503 when one is detected. Use of
6431 @option{-mno-check-zero-division} suppresses such checking for code
6432 generated to run on an MC88100 processor.
6434 GCC assumes that the MC88110 processor correctly detects all instances
6435 of integer division by zero. When @option{-m88110} is specified, no
6436 explicit checks for zero-valued divisors are generated, and both
6437 @option{-mcheck-zero-division} and @option{-mno-check-zero-division} are
6440 @item -muse-div-instruction
6441 @opindex muse-div-instruction
6442 @cindex divide instruction, 88k
6443 Use the div instruction for signed integer division on the
6444 MC88100 processor. By default, the div instruction is not used.
6446 On the MC88100 processor the signed integer division instruction
6447 div) traps to the operating system on a negative operand. The
6448 operating system transparently completes the operation, but at a
6449 large cost in execution time. By default, when compiling code
6450 that might be run on an MC88100 processor, GCC emulates signed
6451 integer division using the unsigned integer division instruction
6452 divu), thereby avoiding the large penalty of a trap to the
6453 operating system. Such emulation has its own, smaller, execution
6454 cost in both time and space. To the extent that your code's
6455 important signed integer division operations are performed on two
6456 nonnegative operands, it may be desirable to use the div
6457 instruction directly.
6459 On the MC88110 processor the div instruction (also known as the
6460 divs instruction) processes negative operands without trapping to
6461 the operating system. When @option{-m88110} is specified,
6462 @option{-muse-div-instruction} is ignored, and the div instruction is used
6463 for signed integer division.
6465 Note that the result of dividing @code{INT_MIN} by @minus{}1 is undefined. In
6466 particular, the behavior of such a division with and without
6467 @option{-muse-div-instruction} may differ.
6469 @item -mtrap-large-shift
6470 @itemx -mhandle-large-shift
6471 @opindex mtrap-large-shift
6472 @opindex mhandle-large-shift
6473 @cindex bit shift overflow (88k)
6474 @cindex large bit shifts (88k)
6475 Include code to detect bit-shifts of more than 31 bits; respectively,
6476 trap such shifts or emit code to handle them properly. By default GCC
6477 makes no special provision for large bit shifts.
6479 @item -mwarn-passed-structs
6480 @opindex mwarn-passed-structs
6481 @cindex structure passing (88k)
6482 Warn when a function passes a struct as an argument or result.
6483 Structure-passing conventions have changed during the evolution of the C
6484 language, and are often the source of portability problems. By default,
6485 GCC issues no such warning.
6488 @c break page here to avoid unsightly interparagraph stretch.
6492 @node RS/6000 and PowerPC Options
6493 @subsection IBM RS/6000 and PowerPC Options
6494 @cindex RS/6000 and PowerPC Options
6495 @cindex IBM RS/6000 and PowerPC Options
6497 These @samp{-m} options are defined for the IBM RS/6000 and PowerPC:
6505 @itemx -mpowerpc-gpopt
6506 @itemx -mno-powerpc-gpopt
6507 @itemx -mpowerpc-gfxopt
6508 @itemx -mno-powerpc-gfxopt
6510 @itemx -mno-powerpc64
6516 @opindex mno-powerpc
6517 @opindex mpowerpc-gpopt
6518 @opindex mno-powerpc-gpopt
6519 @opindex mpowerpc-gfxopt
6520 @opindex mno-powerpc-gfxopt
6522 @opindex mno-powerpc64
6523 GCC supports two related instruction set architectures for the
6524 RS/6000 and PowerPC@. The @dfn{POWER} instruction set are those
6525 instructions supported by the @samp{rios} chip set used in the original
6526 RS/6000 systems and the @dfn{PowerPC} instruction set is the
6527 architecture of the Motorola MPC5xx, MPC6xx, MPC8xx microprocessors, and
6528 the IBM 4xx microprocessors.
6530 Neither architecture is a subset of the other. However there is a
6531 large common subset of instructions supported by both. An MQ
6532 register is included in processors supporting the POWER architecture.
6534 You use these options to specify which instructions are available on the
6535 processor you are using. The default value of these options is
6536 determined when configuring GCC@. Specifying the
6537 @option{-mcpu=@var{cpu_type}} overrides the specification of these
6538 options. We recommend you use the @option{-mcpu=@var{cpu_type}} option
6539 rather than the options listed above.
6541 The @option{-mpower} option allows GCC to generate instructions that
6542 are found only in the POWER architecture and to use the MQ register.
6543 Specifying @option{-mpower2} implies @option{-power} and also allows GCC
6544 to generate instructions that are present in the POWER2 architecture but
6545 not the original POWER architecture.
6547 The @option{-mpowerpc} option allows GCC to generate instructions that
6548 are found only in the 32-bit subset of the PowerPC architecture.
6549 Specifying @option{-mpowerpc-gpopt} implies @option{-mpowerpc} and also allows
6550 GCC to use the optional PowerPC architecture instructions in the
6551 General Purpose group, including floating-point square root. Specifying
6552 @option{-mpowerpc-gfxopt} implies @option{-mpowerpc} and also allows GCC to
6553 use the optional PowerPC architecture instructions in the Graphics
6554 group, including floating-point select.
6556 The @option{-mpowerpc64} option allows GCC to generate the additional
6557 64-bit instructions that are found in the full PowerPC64 architecture
6558 and to treat GPRs as 64-bit, doubleword quantities. GCC defaults to
6559 @option{-mno-powerpc64}.
6561 If you specify both @option{-mno-power} and @option{-mno-powerpc}, GCC
6562 will use only the instructions in the common subset of both
6563 architectures plus some special AIX common-mode calls, and will not use
6564 the MQ register. Specifying both @option{-mpower} and @option{-mpowerpc}
6565 permits GCC to use any instruction from either architecture and to
6566 allow use of the MQ register; specify this for the Motorola MPC601.
6568 @item -mnew-mnemonics
6569 @itemx -mold-mnemonics
6570 @opindex mnew-mnemonics
6571 @opindex mold-mnemonics
6572 Select which mnemonics to use in the generated assembler code. With
6573 @option{-mnew-mnemonics}, GCC uses the assembler mnemonics defined for
6574 the PowerPC architecture. With @option{-mold-mnemonics} it uses the
6575 assembler mnemonics defined for the POWER architecture. Instructions
6576 defined in only one architecture have only one mnemonic; GCC uses that
6577 mnemonic irrespective of which of these options is specified.
6579 GCC defaults to the mnemonics appropriate for the architecture in
6580 use. Specifying @option{-mcpu=@var{cpu_type}} sometimes overrides the
6581 value of these option. Unless you are building a cross-compiler, you
6582 should normally not specify either @option{-mnew-mnemonics} or
6583 @option{-mold-mnemonics}, but should instead accept the default.
6585 @item -mcpu=@var{cpu_type}
6587 Set architecture type, register usage, choice of mnemonics, and
6588 instruction scheduling parameters for machine type @var{cpu_type}.
6589 Supported values for @var{cpu_type} are @samp{rios}, @samp{rios1},
6590 @samp{rsc}, @samp{rios2}, @samp{rs64a}, @samp{601}, @samp{602},
6591 @samp{603}, @samp{603e}, @samp{604}, @samp{604e}, @samp{620},
6592 @samp{630}, @samp{740}, @samp{7400}, @samp{7450}, @samp{750},
6593 @samp{power}, @samp{power2}, @samp{powerpc}, @samp{403}, @samp{505},
6594 @samp{801}, @samp{821}, @samp{823}, and @samp{860} and @samp{common}.
6596 @option{-mcpu=common} selects a completely generic processor. Code
6597 generated under this option will run on any POWER or PowerPC processor.
6598 GCC will use only the instructions in the common subset of both
6599 architectures, and will not use the MQ register. GCC assumes a generic
6600 processor model for scheduling purposes.
6602 @option{-mcpu=power}, @option{-mcpu=power2}, @option{-mcpu=powerpc}, and
6603 @option{-mcpu=powerpc64} specify generic POWER, POWER2, pure 32-bit
6604 PowerPC (i.e., not MPC601), and 64-bit PowerPC architecture machine
6605 types, with an appropriate, generic processor model assumed for
6606 scheduling purposes.
6608 The other options specify a specific processor. Code generated under
6609 those options will run best on that processor, and may not run at all on
6612 The @option{-mcpu} options automatically enable or disable other
6613 @option{-m} options as follows:
6617 @option{-mno-power}, @option{-mno-powerc}
6624 @option{-mpower}, @option{-mno-powerpc}, @option{-mno-new-mnemonics}
6639 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6642 @option{-mpower}, @option{-mpowerpc}, @option{-mnew-mnemonics}
6647 @option{-mno-power}, @option{-mpowerpc}, @option{-mnew-mnemonics}, @option{-msoft-float}
6650 @item -mtune=@var{cpu_type}
6652 Set the instruction scheduling parameters for machine type
6653 @var{cpu_type}, but do not set the architecture type, register usage, or
6654 choice of mnemonics, as @option{-mcpu=@var{cpu_type}} would. The same
6655 values for @var{cpu_type} are used for @option{-mtune} as for
6656 @option{-mcpu}. If both are specified, the code generated will use the
6657 architecture, registers, and mnemonics set by @option{-mcpu}, but the
6658 scheduling parameters set by @option{-mtune}.
6661 @itemx -mno-fp-in-toc
6662 @itemx -mno-sum-in-toc
6663 @itemx -mminimal-toc
6665 @opindex mno-fp-in-toc
6666 @opindex mno-sum-in-toc
6667 @opindex mminimal-toc
6668 Modify generation of the TOC (Table Of Contents), which is created for
6669 every executable file. The @option{-mfull-toc} option is selected by
6670 default. In that case, GCC will allocate at least one TOC entry for
6671 each unique non-automatic variable reference in your program. GCC
6672 will also place floating-point constants in the TOC@. However, only
6673 16,384 entries are available in the TOC@.
6675 If you receive a linker error message that saying you have overflowed
6676 the available TOC space, you can reduce the amount of TOC space used
6677 with the @option{-mno-fp-in-toc} and @option{-mno-sum-in-toc} options.
6678 @option{-mno-fp-in-toc} prevents GCC from putting floating-point
6679 constants in the TOC and @option{-mno-sum-in-toc} forces GCC to
6680 generate code to calculate the sum of an address and a constant at
6681 run-time instead of putting that sum into the TOC@. You may specify one
6682 or both of these options. Each causes GCC to produce very slightly
6683 slower and larger code at the expense of conserving TOC space.
6685 If you still run out of space in the TOC even when you specify both of
6686 these options, specify @option{-mminimal-toc} instead. This option causes
6687 GCC to make only one TOC entry for every file. When you specify this
6688 option, GCC will produce code that is slower and larger but which
6689 uses extremely little TOC space. You may wish to use this option
6690 only on files that contain less frequently executed code.
6696 Enable 64-bit AIX ABI and calling convention: 64-bit pointers, 64-bit
6697 @code{long} type, and the infrastructure needed to support them.
6698 Specifying @option{-maix64} implies @option{-mpowerpc64} and
6699 @option{-mpowerpc}, while @option{-maix32} disables the 64-bit ABI and
6700 implies @option{-mno-powerpc64}. GCC defaults to @option{-maix32}.
6705 @opindex mno-xl-call
6706 On AIX, pass floating-point arguments to prototyped functions beyond the
6707 register save area (RSA) on the stack in addition to argument FPRs. The
6708 AIX calling convention was extended but not initially documented to
6709 handle an obscure K&R C case of calling a function that takes the
6710 address of its arguments with fewer arguments than declared. AIX XL
6711 compilers access floating point arguments which do not fit in the
6712 RSA from the stack when a subroutine is compiled without
6713 optimization. Because always storing floating-point arguments on the
6714 stack is inefficient and rarely needed, this option is not enabled by
6715 default and only is necessary when calling subroutines compiled by AIX
6716 XL compilers without optimization.
6720 Support @dfn{AIX Threads}. Link an application written to use
6721 @dfn{pthreads} with special libraries and startup code to enable the
6726 Support @dfn{IBM RS/6000 SP} @dfn{Parallel Environment} (PE)@. Link an
6727 application written to use message passing with special startup code to
6728 enable the application to run. The system must have PE installed in the
6729 standard location (@file{/usr/lpp/ppe.poe/}), or the @file{specs} file
6730 must be overridden with the @option{-specs=} option to specify the
6731 appropriate directory location. The Parallel Environment does not
6732 support threads, so the @option{-mpe} option and the @option{-mthreads}
6733 option are incompatible.
6737 @opindex msoft-float
6738 @opindex mhard-float
6739 Generate code that does not use (uses) the floating-point register set.
6740 Software floating point emulation is provided if you use the
6741 @option{-msoft-float} option, and pass the option to GCC when linking.
6744 @itemx -mno-multiple
6746 @opindex mno-multiple
6747 Generate code that uses (does not use) the load multiple word
6748 instructions and the store multiple word instructions. These
6749 instructions are generated by default on POWER systems, and not
6750 generated on PowerPC systems. Do not use @option{-mmultiple} on little
6751 endian PowerPC systems, since those instructions do not work when the
6752 processor is in little endian mode. The exceptions are PPC740 and
6753 PPC750 which permit the instructions usage in little endian mode.
6759 Generate code that uses (does not use) the load string instructions
6760 and the store string word instructions to save multiple registers and
6761 do small block moves. These instructions are generated by default on
6762 POWER systems, and not generated on PowerPC systems. Do not use
6763 @option{-mstring} on little endian PowerPC systems, since those
6764 instructions do not work when the processor is in little endian mode.
6765 The exceptions are PPC740 and PPC750 which permit the instructions
6766 usage in little endian mode.
6772 Generate code that uses (does not use) the load or store instructions
6773 that update the base register to the address of the calculated memory
6774 location. These instructions are generated by default. If you use
6775 @option{-mno-update}, there is a small window between the time that the
6776 stack pointer is updated and the address of the previous frame is
6777 stored, which means code that walks the stack frame across interrupts or
6778 signals may get corrupted data.
6781 @itemx -mno-fused-madd
6782 @opindex mfused-madd
6783 @opindex mno-fused-madd
6784 Generate code that uses (does not use) the floating point multiply and
6785 accumulate instructions. These instructions are generated by default if
6786 hardware floating is used.
6788 @item -mno-bit-align
6790 @opindex mno-bit-align
6792 On System V.4 and embedded PowerPC systems do not (do) force structures
6793 and unions that contain bit-fields to be aligned to the base type of the
6796 For example, by default a structure containing nothing but 8
6797 @code{unsigned} bit-fields of length 1 would be aligned to a 4 byte
6798 boundary and have a size of 4 bytes. By using @option{-mno-bit-align},
6799 the structure would be aligned to a 1 byte boundary and be one byte in
6802 @item -mno-strict-align
6803 @itemx -mstrict-align
6804 @opindex mno-strict-align
6805 @opindex mstrict-align
6806 On System V.4 and embedded PowerPC systems do not (do) assume that
6807 unaligned memory references will be handled by the system.
6810 @itemx -mno-relocatable
6811 @opindex mrelocatable
6812 @opindex mno-relocatable
6813 On embedded PowerPC systems generate code that allows (does not allow)
6814 the program to be relocated to a different address at runtime. If you
6815 use @option{-mrelocatable} on any module, all objects linked together must
6816 be compiled with @option{-mrelocatable} or @option{-mrelocatable-lib}.
6818 @item -mrelocatable-lib
6819 @itemx -mno-relocatable-lib
6820 @opindex mrelocatable-lib
6821 @opindex mno-relocatable-lib
6822 On embedded PowerPC systems generate code that allows (does not allow)
6823 the program to be relocated to a different address at runtime. Modules
6824 compiled with @option{-mrelocatable-lib} can be linked with either modules
6825 compiled without @option{-mrelocatable} and @option{-mrelocatable-lib} or
6826 with modules compiled with the @option{-mrelocatable} options.
6832 On System V.4 and embedded PowerPC systems do not (do) assume that
6833 register 2 contains a pointer to a global area pointing to the addresses
6834 used in the program.
6837 @itemx -mlittle-endian
6839 @opindex mlittle-endian
6840 On System V.4 and embedded PowerPC systems compile code for the
6841 processor in little endian mode. The @option{-mlittle-endian} option is
6842 the same as @option{-mlittle}.
6847 @opindex mbig-endian
6848 On System V.4 and embedded PowerPC systems compile code for the
6849 processor in big endian mode. The @option{-mbig-endian} option is
6850 the same as @option{-mbig}.
6854 On System V.4 and embedded PowerPC systems compile code using calling
6855 conventions that adheres to the March 1995 draft of the System V
6856 Application Binary Interface, PowerPC processor supplement. This is the
6857 default unless you configured GCC using @samp{powerpc-*-eabiaix}.
6859 @item -mcall-sysv-eabi
6860 @opindex mcall-sysv-eabi
6861 Specify both @option{-mcall-sysv} and @option{-meabi} options.
6863 @item -mcall-sysv-noeabi
6864 @opindex mcall-sysv-noeabi
6865 Specify both @option{-mcall-sysv} and @option{-mno-eabi} options.
6869 On System V.4 and embedded PowerPC systems compile code using calling
6870 conventions that are similar to those used on AIX@. This is the
6871 default if you configured GCC using @samp{powerpc-*-eabiaix}.
6873 @item -mcall-solaris
6874 @opindex mcall-solaris
6875 On System V.4 and embedded PowerPC systems compile code for the Solaris
6879 @opindex mcall-linux
6880 On System V.4 and embedded PowerPC systems compile code for the
6881 Linux-based GNU system.
6884 @opindex mcall-netbsd
6885 On System V.4 and embedded PowerPC systems compile code for the
6886 NetBSD operating system.
6889 @itemx -mno-prototype
6891 @opindex mno-prototype
6892 On System V.4 and embedded PowerPC systems assume that all calls to
6893 variable argument functions are properly prototyped. Otherwise, the
6894 compiler must insert an instruction before every non prototyped call to
6895 set or clear bit 6 of the condition code register (@var{CR}) to
6896 indicate whether floating point values were passed in the floating point
6897 registers in case the function takes a variable arguments. With
6898 @option{-mprototype}, only calls to prototyped variable argument functions
6899 will set or clear the bit.
6903 On embedded PowerPC systems, assume that the startup module is called
6904 @file{sim-crt0.o} and that the standard C libraries are @file{libsim.a} and
6905 @file{libc.a}. This is the default for @samp{powerpc-*-eabisim}.
6910 On embedded PowerPC systems, assume that the startup module is called
6911 @file{crt0.o} and the standard C libraries are @file{libmvme.a} and
6916 On embedded PowerPC systems, assume that the startup module is called
6917 @file{crt0.o} and the standard C libraries are @file{libads.a} and
6921 @opindex myellowknife
6922 On embedded PowerPC systems, assume that the startup module is called
6923 @file{crt0.o} and the standard C libraries are @file{libyk.a} and
6928 On System V.4 and embedded PowerPC systems, specify that you are
6929 compiling for a VxWorks system.
6933 On embedded PowerPC systems, set the @var{PPC_EMB} bit in the ELF flags
6934 header to indicate that @samp{eabi} extended relocations are used.
6940 On System V.4 and embedded PowerPC systems do (do not) adhere to the
6941 Embedded Applications Binary Interface (eabi) which is a set of
6942 modifications to the System V.4 specifications. Selecting @option{-meabi}
6943 means that the stack is aligned to an 8 byte boundary, a function
6944 @code{__eabi} is called to from @code{main} to set up the eabi
6945 environment, and the @option{-msdata} option can use both @code{r2} and
6946 @code{r13} to point to two separate small data areas. Selecting
6947 @option{-mno-eabi} means that the stack is aligned to a 16 byte boundary,
6948 do not call an initialization function from @code{main}, and the
6949 @option{-msdata} option will only use @code{r13} to point to a single
6950 small data area. The @option{-meabi} option is on by default if you
6951 configured GCC using one of the @samp{powerpc*-*-eabi*} options.
6954 @opindex msdata=eabi
6955 On System V.4 and embedded PowerPC systems, put small initialized
6956 @code{const} global and static data in the @samp{.sdata2} section, which
6957 is pointed to by register @code{r2}. Put small initialized
6958 non-@code{const} global and static data in the @samp{.sdata} section,
6959 which is pointed to by register @code{r13}. Put small uninitialized
6960 global and static data in the @samp{.sbss} section, which is adjacent to
6961 the @samp{.sdata} section. The @option{-msdata=eabi} option is
6962 incompatible with the @option{-mrelocatable} option. The
6963 @option{-msdata=eabi} option also sets the @option{-memb} option.
6966 @opindex msdata=sysv
6967 On System V.4 and embedded PowerPC systems, put small global and static
6968 data in the @samp{.sdata} section, which is pointed to by register
6969 @code{r13}. Put small uninitialized global and static data in the
6970 @samp{.sbss} section, which is adjacent to the @samp{.sdata} section.
6971 The @option{-msdata=sysv} option is incompatible with the
6972 @option{-mrelocatable} option.
6974 @item -msdata=default
6976 @opindex msdata=default
6978 On System V.4 and embedded PowerPC systems, if @option{-meabi} is used,
6979 compile code the same as @option{-msdata=eabi}, otherwise compile code the
6980 same as @option{-msdata=sysv}.
6983 @opindex msdata-data
6984 On System V.4 and embedded PowerPC systems, put small global and static
6985 data in the @samp{.sdata} section. Put small uninitialized global and
6986 static data in the @samp{.sbss} section. Do not use register @code{r13}
6987 to address small data however. This is the default behavior unless
6988 other @option{-msdata} options are used.
6992 @opindex msdata=none
6994 On embedded PowerPC systems, put all initialized global and static data
6995 in the @samp{.data} section, and all uninitialized data in the
6996 @samp{.bss} section.
7000 @cindex smaller data references (PowerPC)
7001 @cindex .sdata/.sdata2 references (PowerPC)
7002 On embedded PowerPC systems, put global and static items less than or
7003 equal to @var{num} bytes into the small data or bss sections instead of
7004 the normal data or bss section. By default, @var{num} is 8. The
7005 @option{-G @var{num}} switch is also passed to the linker.
7006 All modules should be compiled with the same @option{-G @var{num}} value.
7009 @itemx -mno-regnames
7011 @opindex mno-regnames
7012 On System V.4 and embedded PowerPC systems do (do not) emit register
7013 names in the assembly language output using symbolic forms.
7018 @subsection IBM RT Options
7020 @cindex IBM RT options
7022 These @samp{-m} options are defined for the IBM RT PC:
7026 @opindex min-line-mul
7027 Use an in-line code sequence for integer multiplies. This is the
7030 @item -mcall-lib-mul
7031 @opindex mcall-lib-mul
7032 Call @code{lmul$$} for integer multiples.
7034 @item -mfull-fp-blocks
7035 @opindex mfull-fp-blocks
7036 Generate full-size floating point data blocks, including the minimum
7037 amount of scratch space recommended by IBM@. This is the default.
7039 @item -mminimum-fp-blocks
7040 @opindex mminimum-fp-blocks
7041 Do not include extra scratch space in floating point data blocks. This
7042 results in smaller code, but slower execution, since scratch space must
7043 be allocated dynamically.
7045 @cindex @file{varargs.h} and RT PC
7046 @cindex @file{stdarg.h} and RT PC
7047 @item -mfp-arg-in-fpregs
7048 @opindex mfp-arg-in-fpregs
7049 Use a calling sequence incompatible with the IBM calling convention in
7050 which floating point arguments are passed in floating point registers.
7051 Note that @code{varargs.h} and @code{stdarg.h} will not work with
7052 floating point operands if this option is specified.
7054 @item -mfp-arg-in-gregs
7055 @opindex mfp-arg-in-gregs
7056 Use the normal calling convention for floating point arguments. This is
7059 @item -mhc-struct-return
7060 @opindex mhc-struct-return
7061 Return structures of more than one word in memory, rather than in a
7062 register. This provides compatibility with the MetaWare HighC (hc)
7063 compiler. Use the option @option{-fpcc-struct-return} for compatibility
7064 with the Portable C Compiler (pcc).
7066 @item -mnohc-struct-return
7067 @opindex mnohc-struct-return
7068 Return some structures of more than one word in registers, when
7069 convenient. This is the default. For compatibility with the
7070 IBM-supplied compilers, use the option @option{-fpcc-struct-return} or the
7071 option @option{-mhc-struct-return}.
7075 @subsection MIPS Options
7076 @cindex MIPS options
7078 These @samp{-m} options are defined for the MIPS family of computers:
7082 @item -march=@var{cpu-type}
7084 Assume the defaults for the machine type @var{cpu-type} when generating
7085 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7086 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7087 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7088 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7089 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7090 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc.
7092 @item -mtune=@var{cpu-type}
7094 Assume the defaults for the machine type @var{cpu-type} when scheduling
7095 instructions. The choices for @var{cpu-type} are @samp{r2000}, @samp{r3000},
7096 @samp{r3900}, @samp{r4000}, @samp{r4100}, @samp{r4300}, @samp{r4400},
7097 @samp{r4600}, @samp{r4650}, @samp{r5000}, @samp{r6000}, @samp{r8000},
7098 and @samp{orion}. Additionally, the @samp{r2000}, @samp{r3000},
7099 @samp{r4000}, @samp{r5000}, and @samp{r6000} can be abbreviated as
7100 @samp{r2k} (or @samp{r2K}), @samp{r3k}, etc. While picking a specific
7101 @var{cpu-type} will schedule things appropriately for that particular
7102 chip, the compiler will not generate any code that does not meet level 1
7103 of the MIPS ISA (instruction set architecture) without a @option{-mipsX}
7104 or @option{-mabi} switch being used.
7106 @item -mcpu=@var{cpu-type}
7108 This is identical to specifying both @option{-march} and @option{-mtune}.
7112 Issue instructions from level 1 of the MIPS ISA@. This is the default.
7113 @samp{r3000} is the default @var{cpu-type} at this ISA level.
7117 Issue instructions from level 2 of the MIPS ISA (branch likely, square
7118 root instructions). @samp{r6000} is the default @var{cpu-type} at this
7123 Issue instructions from level 3 of the MIPS ISA (64-bit instructions).
7124 @samp{r4000} is the default @var{cpu-type} at this ISA level.
7128 Issue instructions from level 4 of the MIPS ISA (conditional move,
7129 prefetch, enhanced FPU instructions). @samp{r8000} is the default
7130 @var{cpu-type} at this ISA level.
7134 Assume that 32 32-bit floating point registers are available. This is
7139 Assume that 32 64-bit floating point registers are available. This is
7140 the default when the @option{-mips3} option is used.
7143 @itemx -mno-fused-madd
7144 @opindex mfused-madd
7145 @opindex mno-fused-madd
7146 Generate code that uses (does not use) the floating point multiply and
7147 accumulate instructions, when they are available. These instructions
7148 are generated by default if they are available, but this may be
7149 undesirable if the extra precision causes problems or on certain chips
7150 in the mode where denormals are rounded to zero where denormals
7151 generated by multiply and accumulate instructions cause exceptions
7156 Assume that 32 32-bit general purpose registers are available. This is
7161 Assume that 32 64-bit general purpose registers are available. This is
7162 the default when the @option{-mips3} option is used.
7166 Force int and long types to be 64 bits wide. See @option{-mlong32} for an
7167 explanation of the default, and the width of pointers.
7171 Force long types to be 64 bits wide. See @option{-mlong32} for an
7172 explanation of the default, and the width of pointers.
7176 Force long, int, and pointer types to be 32 bits wide.
7178 If none of @option{-mlong32}, @option{-mlong64}, or @option{-mint64} are set,
7179 the size of ints, longs, and pointers depends on the ABI and ISA chosen.
7180 For @option{-mabi=32}, and @option{-mabi=n32}, ints and longs are 32 bits
7181 wide. For @option{-mabi=64}, ints are 32 bits, and longs are 64 bits wide.
7182 For @option{-mabi=eabi} and either @option{-mips1} or @option{-mips2}, ints
7183 and longs are 32 bits wide. For @option{-mabi=eabi} and higher ISAs, ints
7184 are 32 bits, and longs are 64 bits wide. The width of pointer types is
7185 the smaller of the width of longs or the width of general purpose
7186 registers (which in turn depends on the ISA)@.
7198 Generate code for the indicated ABI@. The default instruction level is
7199 @option{-mips1} for @samp{32}, @option{-mips3} for @samp{n32}, and
7200 @option{-mips4} otherwise. Conversely, with @option{-mips1} or
7201 @option{-mips2}, the default ABI is @samp{32}; otherwise, the default ABI
7206 Generate code for the MIPS assembler, and invoke @file{mips-tfile} to
7207 add normal debug information. This is the default for all
7208 platforms except for the OSF/1 reference platform, using the OSF/rose
7209 object format. If the either of the @option{-gstabs} or @option{-gstabs+}
7210 switches are used, the @file{mips-tfile} program will encapsulate the
7211 stabs within MIPS ECOFF@.
7215 Generate code for the GNU assembler. This is the default on the OSF/1
7216 reference platform, using the OSF/rose object format. Also, this is
7217 the default if the configure option @option{--with-gnu-as} is used.
7219 @item -msplit-addresses
7220 @itemx -mno-split-addresses
7221 @opindex msplit-addresses
7222 @opindex mno-split-addresses
7223 Generate code to load the high and low parts of address constants separately.
7224 This allows GCC to optimize away redundant loads of the high order
7225 bits of addresses. This optimization requires GNU as and GNU ld.
7226 This optimization is enabled by default for some embedded targets where
7227 GNU as and GNU ld are standard.
7233 The @option{-mrnames} switch says to output code using the MIPS software
7234 names for the registers, instead of the hardware names (ie, @var{a0}
7235 instead of @var{$4}). The only known assembler that supports this option
7236 is the Algorithmics assembler.
7242 The @option{-mgpopt} switch says to write all of the data declarations
7243 before the instructions in the text section, this allows the MIPS
7244 assembler to generate one word memory references instead of using two
7245 words for short global or static data items. This is on by default if
7246 optimization is selected.
7252 For each non-inline function processed, the @option{-mstats} switch
7253 causes the compiler to emit one line to the standard error file to
7254 print statistics about the program (number of registers saved, stack
7261 The @option{-mmemcpy} switch makes all block moves call the appropriate
7262 string function (@samp{memcpy} or @samp{bcopy}) instead of possibly
7263 generating inline code.
7266 @itemx -mno-mips-tfile
7267 @opindex mmips-tfile
7268 @opindex mno-mips-tfile
7269 The @option{-mno-mips-tfile} switch causes the compiler not
7270 postprocess the object file with the @file{mips-tfile} program,
7271 after the MIPS assembler has generated it to add debug support. If
7272 @file{mips-tfile} is not run, then no local variables will be
7273 available to the debugger. In addition, @file{stage2} and
7274 @file{stage3} objects will have the temporary file names passed to the
7275 assembler embedded in the object file, which means the objects will
7276 not compare the same. The @option{-mno-mips-tfile} switch should only
7277 be used when there are bugs in the @file{mips-tfile} program that
7278 prevents compilation.
7281 @opindex msoft-float
7282 Generate output containing library calls for floating point.
7283 @strong{Warning:} the requisite libraries are not part of GCC@.
7284 Normally the facilities of the machine's usual C compiler are used, but
7285 this can't be done directly in cross-compilation. You must make your
7286 own arrangements to provide suitable library functions for
7290 @opindex mhard-float
7291 Generate output containing floating point instructions. This is the
7292 default if you use the unmodified sources.
7295 @itemx -mno-abicalls
7297 @opindex mno-abicalls
7298 Emit (or do not emit) the pseudo operations @samp{.abicalls},
7299 @samp{.cpload}, and @samp{.cprestore} that some System V.4 ports use for
7300 position independent code.
7303 @itemx -mno-long-calls
7304 @opindex mlong-calls
7305 @opindex mno-long-calls
7306 Do all calls with the @samp{JALR} instruction, which requires
7307 loading up a function's address into a register before the call.
7308 You need to use this switch, if you call outside of the current
7309 512 megabyte segment to functions that are not through pointers.
7312 @itemx -mno-half-pic
7314 @opindex mno-half-pic
7315 Put pointers to extern references into the data section and load them
7316 up, rather than put the references in the text section.
7318 @item -membedded-pic
7319 @itemx -mno-embedded-pic
7320 @opindex membedded-pic
7321 @opindex mno-embedded-pic
7322 Generate PIC code suitable for some embedded systems. All calls are
7323 made using PC relative address, and all data is addressed using the $gp
7324 register. No more than 65536 bytes of global data may be used. This
7325 requires GNU as and GNU ld which do most of the work. This currently
7326 only works on targets which use ECOFF; it does not work with ELF@.
7328 @item -membedded-data
7329 @itemx -mno-embedded-data
7330 @opindex membedded-data
7331 @opindex mno-embedded-data
7332 Allocate variables to the read-only data section first if possible, then
7333 next in the small data section if possible, otherwise in data. This gives
7334 slightly slower code than the default, but reduces the amount of RAM required
7335 when executing, and thus may be preferred for some embedded systems.
7337 @item -muninit-const-in-rodata
7338 @itemx -mno-uninit-const-in-rodata
7339 @opindex muninit-const-in-rodata
7340 @opindex mno-uninit-const-in-rodata
7341 When used together with @option{-membedded-data}, it will always store uninitialized
7342 const variables in the read-only data section.
7344 @item -msingle-float
7345 @itemx -mdouble-float
7346 @opindex msingle-float
7347 @opindex mdouble-float
7348 The @option{-msingle-float} switch tells gcc to assume that the floating
7349 point coprocessor only supports single precision operations, as on the
7350 @samp{r4650} chip. The @option{-mdouble-float} switch permits gcc to use
7351 double precision operations. This is the default.
7357 Permit use of the @samp{mad}, @samp{madu} and @samp{mul} instructions,
7358 as on the @samp{r4650} chip.
7362 Turns on @option{-msingle-float}, @option{-mmad}, and, at least for now,
7363 @option{-mcpu=r4650}.
7369 Enable 16-bit instructions.
7373 Use the entry and exit pseudo ops. This option can only be used with
7378 Compile code for the processor in little endian mode.
7379 The requisite libraries are assumed to exist.
7383 Compile code for the processor in big endian mode.
7384 The requisite libraries are assumed to exist.
7388 @cindex smaller data references (MIPS)
7389 @cindex gp-relative references (MIPS)
7390 Put global and static items less than or equal to @var{num} bytes into
7391 the small data or bss sections instead of the normal data or bss
7392 section. This allows the assembler to emit one word memory reference
7393 instructions based on the global pointer (@var{gp} or @var{$28}),
7394 instead of the normal two words used. By default, @var{num} is 8 when
7395 the MIPS assembler is used, and 0 when the GNU assembler is used. The
7396 @option{-G @var{num}} switch is also passed to the assembler and linker.
7397 All modules should be compiled with the same @option{-G @var{num}}
7402 Tell the MIPS assembler to not run its preprocessor over user
7403 assembler files (with a @samp{.s} suffix) when assembling them.
7407 Pass an option to gas which will cause nops to be inserted if
7408 the read of the destination register of an mfhi or mflo instruction
7409 occurs in the following two instructions.
7413 Do not include the default crt0.
7417 These options are defined by the macro
7418 @code{TARGET_SWITCHES} in the machine description. The default for the
7419 options is also defined by that macro, which enables you to change the
7423 @node i386 and x86-64 Options
7424 @subsection Intel 386 and AMD x86-64 Options
7425 @cindex i386 Options
7426 @cindex x86-64 Options
7427 @cindex Intel 386 Options
7428 @cindex AMD x86-64 Options
7430 These @samp{-m} options are defined for the i386 and x86-64 family of
7434 @item -mcpu=@var{cpu-type}
7436 Assume the defaults for the machine type @var{cpu-type} when scheduling
7437 instructions. The choices for @var{cpu-type} are @samp{i386},
7438 @samp{i486}, @samp{i586}, @samp{i686}, @samp{pentium},
7439 @samp{pentiumpro}, @samp{pentium4}, @samp{k6}, and @samp{athlon}
7441 While picking a specific @var{cpu-type} will schedule things appropriately
7442 for that particular chip, the compiler will not generate any code that
7443 does not run on the i386 without the @option{-march=@var{cpu-type}} option
7444 being used. @samp{i586} is equivalent to @samp{pentium} and @samp{i686}
7445 is equivalent to @samp{pentiumpro}. @samp{k6} and @samp{athlon} are the
7446 AMD chips as opposed to the Intel ones.
7448 @item -march=@var{cpu-type}
7450 Generate instructions for the machine type @var{cpu-type}. The choices
7451 for @var{cpu-type} are the same as for @option{-mcpu}. Moreover,
7452 specifying @option{-march=@var{cpu-type}} implies @option{-mcpu=@var{cpu-type}}.
7461 @opindex mpentiumpro
7462 These options are synonyms for @option{-mcpu=i386}, @option{-mcpu=i486},
7463 @option{-mcpu=pentium}, and @option{-mcpu=pentiumpro} respectively.
7464 These synonyms are deprecated.
7466 @item -mintel-syntax
7467 @opindex mintel-syntax
7468 Emit assembly using Intel syntax opcodes instead of AT&T syntax.
7473 @opindex mno-ieee-fp
7474 Control whether or not the compiler uses IEEE floating point
7475 comparisons. These handle correctly the case where the result of a
7476 comparison is unordered.
7479 @opindex msoft-float
7480 Generate output containing library calls for floating point.
7481 @strong{Warning:} the requisite libraries are not part of GCC@.
7482 Normally the facilities of the machine's usual C compiler are used, but
7483 this can't be done directly in cross-compilation. You must make your
7484 own arrangements to provide suitable library functions for
7487 On machines where a function returns floating point results in the 80387
7488 register stack, some floating point opcodes may be emitted even if
7489 @option{-msoft-float} is used.
7491 @item -mno-fp-ret-in-387
7492 @opindex mno-fp-ret-in-387
7493 Do not use the FPU registers for return values of functions.
7495 The usual calling convention has functions return values of types
7496 @code{float} and @code{double} in an FPU register, even if there
7497 is no FPU@. The idea is that the operating system should emulate
7500 The option @option{-mno-fp-ret-in-387} causes such values to be returned
7501 in ordinary CPU registers instead.
7503 @item -mno-fancy-math-387
7504 @opindex mno-fancy-math-387
7505 Some 387 emulators do not support the @code{sin}, @code{cos} and
7506 @code{sqrt} instructions for the 387. Specify this option to avoid
7507 generating those instructions. This option is the default on FreeBSD@.
7508 As of revision 2.6.1, these instructions are not generated unless you
7509 also use the @option{-funsafe-math-optimizations} switch.
7511 @item -malign-double
7512 @itemx -mno-align-double
7513 @opindex malign-double
7514 @opindex mno-align-double
7515 Control whether GCC aligns @code{double}, @code{long double}, and
7516 @code{long long} variables on a two word boundary or a one word
7517 boundary. Aligning @code{double} variables on a two word boundary will
7518 produce code that runs somewhat faster on a @samp{Pentium} at the
7519 expense of more memory.
7521 @item -m128bit-long-double
7522 @opindex m128bit-long-double
7523 Control the size of @code{long double} type. i386 application binary interface
7524 specify the size to be 12 bytes, while modern architectures (Pentium and newer)
7525 prefer @code{long double} aligned to 8 or 16 byte boundary. This is
7526 impossible to reach with 12 byte long doubles in the array accesses.
7528 @strong{Warning:} if you use the @option{-m128bit-long-double} switch, the
7529 structures and arrays containing @code{long double} will change their size as
7530 well as function calling convention for function taking @code{long double}
7533 @item -m96bit-long-double
7534 @opindex m96bit-long-double
7535 Set the size of @code{long double} to 96 bits as required by the i386
7536 application binary interface. This is the default.
7539 @itemx -mno-svr3-shlib
7540 @opindex msvr3-shlib
7541 @opindex mno-svr3-shlib
7542 Control whether GCC places uninitialized local variables into the
7543 @code{bss} or @code{data} segments. @option{-msvr3-shlib} places them
7544 into @code{bss}. These options are meaningful only on System V Release 3.
7548 Use a different function-calling convention, in which functions that
7549 take a fixed number of arguments return with the @code{ret} @var{num}
7550 instruction, which pops their arguments while returning. This saves one
7551 instruction in the caller since there is no need to pop the arguments
7554 You can specify that an individual function is called with this calling
7555 sequence with the function attribute @samp{stdcall}. You can also
7556 override the @option{-mrtd} option by using the function attribute
7557 @samp{cdecl}. @xref{Function Attributes}.
7559 @strong{Warning:} this calling convention is incompatible with the one
7560 normally used on Unix, so you cannot use it if you need to call
7561 libraries compiled with the Unix compiler.
7563 Also, you must provide function prototypes for all functions that
7564 take variable numbers of arguments (including @code{printf});
7565 otherwise incorrect code will be generated for calls to those
7568 In addition, seriously incorrect code will result if you call a
7569 function with too many arguments. (Normally, extra arguments are
7570 harmlessly ignored.)
7572 @item -mregparm=@var{num}
7574 Control how many registers are used to pass integer arguments. By
7575 default, no registers are used to pass arguments, and at most 3
7576 registers can be used. You can control this behavior for a specific
7577 function by using the function attribute @samp{regparm}.
7578 @xref{Function Attributes}.
7580 @strong{Warning:} if you use this switch, and
7581 @var{num} is nonzero, then you must build all modules with the same
7582 value, including any libraries. This includes the system libraries and
7585 @item -mpreferred-stack-boundary=@var{num}
7586 @opindex mpreferred-stack-boundary
7587 Attempt to keep the stack boundary aligned to a 2 raised to @var{num}
7588 byte boundary. If @option{-mpreferred-stack-boundary} is not specified,
7589 the default is 4 (16 bytes or 128 bits).
7591 The stack is required to be aligned on a 4 byte boundary. On Pentium
7592 and PentiumPro, @code{double} and @code{long double} values should be
7593 aligned to an 8 byte boundary (see @option{-malign-double}) or suffer
7594 significant run time performance penalties. On Pentium III, the
7595 Streaming SIMD Extension (SSE) data type @code{__m128} suffers similar
7596 penalties if it is not 16 byte aligned.
7598 To ensure proper alignment of this values on the stack, the stack boundary
7599 must be as aligned as that required by any value stored on the stack.
7600 Further, every function must be generated such that it keeps the stack
7601 aligned. Thus calling a function compiled with a higher preferred
7602 stack boundary from a function compiled with a lower preferred stack
7603 boundary will most likely misalign the stack. It is recommended that
7604 libraries that use callbacks always use the default setting.
7606 This extra alignment does consume extra stack space. Code that is sensitive
7607 to stack space usage, such as embedded systems and operating system kernels,
7608 may want to reduce the preferred alignment to
7609 @option{-mpreferred-stack-boundary=2}.
7623 These switches enable or disable the use of built-in functions that allow
7624 direct access to the MMX, SSE and 3Dnow extensions of the instruction set.
7626 The following machine modes are available for use with MMX built-in functions
7627 (@pxref{Vector Extensions}): @code{V2SI} for a vector of two 32 bit integers,
7628 @code{V4HI} for a vector of four 16 bit integers, and @code{V8QI} for a
7629 vector of eight 8 bit integers. Some of the built-in functions operate on
7630 MMX registers as a whole 64 bit entity, these use @code{DI} as their mode.
7632 If 3Dnow extensions are enabled, @code{V2SF} is used as a mode for a vector
7633 of two 32 bit floating point values.
7635 If SSE extensions are enabled, @code{V4SF} is used for a vector of four 32 bit
7636 floating point values. Some instructions use a vector of four 32 bit
7637 integers, these use @code{V4SI}. Finally, some instructions operate on an
7638 entire vector register, interpreting it as a 128 bit integer, these use mode
7641 The following built-in functions are made available by @option{-mmmx}:
7643 @item v8qi __builtin_ia32_paddb (v8qi, v8qi)
7644 Generates the @code{paddb} machine instruction.
7645 @item v4hi __builtin_ia32_paddw (v4hi, v4hi)
7646 Generates the @code{paddw} machine instruction.
7647 @item v2si __builtin_ia32_paddd (v2si, v2si)
7648 Generates the @code{paddd} machine instruction.
7649 @item v8qi __builtin_ia32_psubb (v8qi, v8qi)
7650 Generates the @code{psubb} machine instruction.
7651 @item v4hi __builtin_ia32_psubw (v4hi, v4hi)
7652 Generates the @code{psubw} machine instruction.
7653 @item v2si __builtin_ia32_psubd (v2si, v2si)
7654 Generates the @code{psubd} machine instruction.
7656 @item v8qi __builtin_ia32_paddsb (v8qi, v8qi)
7657 Generates the @code{paddsb} machine instruction.
7658 @item v4hi __builtin_ia32_paddsw (v4hi, v4hi)
7659 Generates the @code{paddsw} machine instruction.
7660 @item v8qi __builtin_ia32_psubsb (v8qi, v8qi)
7661 Generates the @code{psubsb} machine instruction.
7662 @item v4hi __builtin_ia32_psubsw (v4hi, v4hi)
7663 Generates the @code{psubsw} machine instruction.
7665 @item v8qi __builtin_ia32_paddusb (v8qi, v8qi)
7666 Generates the @code{paddusb} machine instruction.
7667 @item v4hi __builtin_ia32_paddusw (v4hi, v4hi)
7668 Generates the @code{paddusw} machine instruction.
7669 @item v8qi __builtin_ia32_psubusb (v8qi, v8qi)
7670 Generates the @code{psubusb} machine instruction.
7671 @item v4hi __builtin_ia32_psubusw (v4hi, v4hi)
7672 Generates the @code{psubusw} machine instruction.
7674 @item v4hi __builtin_ia32_pmullw (v4hi, v4hi)
7675 Generates the @code{pmullw} machine instruction.
7676 @item v4hi __builtin_ia32_pmulhw (v4hi, v4hi)
7677 Generates the @code{pmulhw} machine instruction.
7679 @item di __builtin_ia32_pand (di, di)
7680 Generates the @code{pand} machine instruction.
7681 @item di __builtin_ia32_pandn (di,di)
7682 Generates the @code{pandn} machine instruction.
7683 @item di __builtin_ia32_por (di, di)
7684 Generates the @code{por} machine instruction.
7685 @item di __builtin_ia32_pxor (di, di)
7686 Generates the @code{pxor} machine instruction.
7688 @item v8qi __builtin_ia32_pcmpeqb (v8qi, v8qi)
7689 Generates the @code{pcmpeqb} machine instruction.
7690 @item v4hi __builtin_ia32_pcmpeqw (v4hi, v4hi)
7691 Generates the @code{pcmpeqw} machine instruction.
7692 @item v2si __builtin_ia32_pcmpeqd (v2si, v2si)
7693 Generates the @code{pcmpeqd} machine instruction.
7694 @item v8qi __builtin_ia32_pcmpgtb (v8qi, v8qi)
7695 Generates the @code{pcmpgtb} machine instruction.
7696 @item v4hi __builtin_ia32_pcmpgtw (v4hi, v4hi)
7697 Generates the @code{pcmpgtw} machine instruction.
7698 @item v2si __builtin_ia32_pcmpgtd (v2si, v2si)
7699 Generates the @code{pcmpgtd} machine instruction.
7701 @item v8qi __builtin_ia32_punpckhbw (v8qi, v8qi)
7702 Generates the @code{punpckhbw} machine instruction.
7703 @item v4hi __builtin_ia32_punpckhwd (v4hi, v4hi)
7704 Generates the @code{punpckhwd} machine instruction.
7705 @item v2si __builtin_ia32_punpckhdq (v2si, v2si)
7706 Generates the @code{punpckhdq} machine instruction.
7707 @item v8qi __builtin_ia32_punpcklbw (v8qi, v8qi)
7708 Generates the @code{punpcklbw} machine instruction.
7709 @item v4hi __builtin_ia32_punpcklwd (v4hi, v4hi)
7710 Generates the @code{punpcklwd} machine instruction.
7711 @item v2si __builtin_ia32_punpckldq (v2si, v2si)
7712 Generates the @code{punpckldq} machine instruction.
7714 @item v8qi __builtin_ia32_packsswb (v4hi, v4hi)
7715 Generates the @code{packsswb} machine instruction.
7716 @item v4hi __builtin_ia32_packssdw (v2si, v2si)
7717 Generates the @code{packssdw} machine instruction.
7718 @item v8qi __builtin_ia32_packuswb (v4hi, v4hi)
7719 Generates the @code{packuswb} machine instruction.
7723 The following built-in functions are made available either with @option{-msse}, or
7724 with a combination of @option{-m3dnow} and @option{-march=athlon}.
7727 @item v4hi __builtin_ia32_pmulhuw (v4hi, v4hi)
7728 Generates the @code{pmulhuw} machine instruction.
7730 @item v8qi __builtin_ia32_pavgb (v8qi, v8qi)
7731 Generates the @code{pavgb} machine instruction.
7732 @item v4hi __builtin_ia32_pavgw (v4hi, v4hi)
7733 Generates the @code{pavgw} machine instruction.
7734 @item v4hi __builtin_ia32_psadbw (v8qi, v8qi)
7735 Generates the @code{psadbw} machine instruction.
7737 @item v8qi __builtin_ia32_pmaxub (v8qi, v8qi)
7738 Generates the @code{pmaxub} machine instruction.
7739 @item v4hi __builtin_ia32_pmaxsw (v4hi, v4hi)
7740 Generates the @code{pmaxsw} machine instruction.
7741 @item v8qi __builtin_ia32_pminub (v8qi, v8qi)
7742 Generates the @code{pminub} machine instruction.
7743 @item v4hi __builtin_ia32_pminsw (v4hi, v4hi)
7744 Generates the @code{pminsw} machine instruction.
7746 @item int __builtin_ia32_pextrw (v4hi, int)
7747 Generates the @code{pextrw} machine instruction.
7748 @item v4hi __builtin_ia32_pinsrw (v4hi, int, int)
7749 Generates the @code{pinsrw} machine instruction.
7751 @item int __builtin_ia32_pmovmskb (v8qi)
7752 Generates the @code{pmovmskb} machine instruction.
7753 @item void __builtin_ia32_maskmovq (v8qi, v8qi, char *)
7754 Generates the @code{maskmovq} machine instruction.
7755 @item void __buitlin_ia32_movntq (di *, di)
7756 Generates the @code{movntq} machine instruction.
7757 @item void __buitlin_ia32_sfence (void)
7758 Generates the @code{sfence} machine instruction.
7759 @item void __builtin_ia32_prefetch (char *, int selector)
7760 Generates a prefetch machine instruction, depending on the value of
7761 selector. If @code{selector} is 0, it generates @code{prefetchnta}; for
7762 a value of 1, it generates @code{prefetcht0}; for a value of 2, it generates
7763 @code{prefetcht1}; and for a value of 3 it generates @code{prefetcht2}.
7767 The following built-in functions are available when @option{-msse} is used.
7770 @item int __buitlin_ia32_comieq (v4sf, v4sf)
7771 Generates the @code{comiss} machine instruction and performs an equality
7772 comparison. The return value is the truth value of that comparison.
7773 @item int __buitlin_ia32_comineq (v4sf, v4sf)
7774 Generates the @code{comiss} machine instruction and performs an inequality
7775 comparison. The return value is the truth value of that comparison.
7776 @item int __buitlin_ia32_comilt (v4sf, v4sf)
7777 Generates the @code{comiss} machine instruction and performs a ``less than''
7778 comparison. The return value is the truth value of that comparison.
7779 @item int __buitlin_ia32_comile (v4sf, v4sf)
7780 Generates the @code{comiss} machine instruction and performs a ``less or
7781 equal'' comparison. The return value is the truth value of that comparison.
7782 @item int __buitlin_ia32_comigt (v4sf, v4sf)
7783 Generates the @code{comiss} machine instruction and performs a ``greater than''
7784 comparison. The return value is the truth value of that comparison.
7785 @item int __buitlin_ia32_comige (v4sf, v4sf)
7786 Generates the @code{comiss} machine instruction and performs a ``greater or
7787 equal'' comparison. The return value is the truth value of that comparison.
7789 @item int __buitlin_ia32_ucomieq (v4sf, v4sf)
7790 Generates the @code{ucomiss} machine instruction and performs an equality
7791 comparison. The return value is the truth value of that comparison.
7792 @item int __buitlin_ia32_ucomineq (v4sf, v4sf)
7793 Generates the @code{ucomiss} machine instruction and performs an inequality
7794 comparison. The return value is the truth value of that comparison.
7795 @item int __buitlin_ia32_ucomilt (v4sf, v4sf)
7796 Generates the @code{ucomiss} machine instruction and performs a ``less than''
7797 comparison. The return value is the truth value of that comparison.
7798 @item int __buitlin_ia32_ucomile (v4sf, v4sf)
7799 Generates the @code{ucomiss} machine instruction and performs a ``less or
7800 equal'' comparison. The return value is the truth value of that comparison.
7801 @item int __buitlin_ia32_ucomigt (v4sf, v4sf)
7802 Generates the @code{ucomiss} machine instruction and performs a ``greater than''
7803 comparison. The return value is the truth value of that comparison.
7804 @item int __buitlin_ia32_ucomige (v4sf, v4sf)
7805 Generates the @code{ucomiss} machine instruction and performs a ``greater or
7806 equal'' comparison. The return value is the truth value of that comparison.
7808 @item v4sf __buitlin_ia32_addps (v4sf, v4sf)
7809 Generates the @code{addps} machine instruction.
7810 @item v4sf __buitlin_ia32_addss (v4sf, v4sf)
7811 Generates the @code{addss} machine instruction.
7812 @item v4sf __buitlin_ia32_subps (v4sf, v4sf)
7813 Generates the @code{subps} machine instruction.
7814 @item v4sf __buitlin_ia32_subss (v4sf, v4sf)
7815 Generates the @code{subss} machine instruction.
7816 @item v4sf __buitlin_ia32_mulps (v4sf, v4sf)
7817 Generates the @code{mulps} machine instruction.
7818 @item v4sf __buitlin_ia32_mulss (v4sf, v4sf)
7819 Generates the @code{mulss} machine instruction.
7820 @item v4sf __buitlin_ia32_divps (v4sf, v4sf)
7821 Generates the @code{divps} machine instruction.
7822 @item v4sf __buitlin_ia32_divss (v4sf, v4sf)
7823 Generates the @code{divss} machine instruction.
7825 @item v4si __buitlin_ia32_cmpeqps (v4sf, v4sf)
7826 Generates the @code{cmpeqps} machine instruction.
7827 @item v4si __buitlin_ia32_cmplts (v4sf, v4sf)
7828 Generates the @code{cmpltps} machine instruction.
7829 @item v4si __buitlin_ia32_cmpleps (v4sf, v4sf)
7830 Generates the @code{cmpleps} machine instruction.
7831 @item v4si __buitlin_ia32_cmpgtps (v4sf, v4sf)
7832 Generates the @code{cmpgtps} machine instruction.
7833 @item v4si __buitlin_ia32_cmpgeps (v4sf, v4sf)
7834 Generates the @code{cmpgeps} machine instruction.
7835 @item v4si __buitlin_ia32_cmpunordps (v4sf, v4sf)
7836 Generates the @code{cmpunodps} machine instruction.
7837 @item v4si __buitlin_ia32_cmpneqps (v4sf, v4sf)
7838 Generates the @code{cmpeqps} machine instruction.
7839 @item v4si __buitlin_ia32_cmpnltps (v4sf, v4sf)
7840 Generates the @code{cmpltps} machine instruction.
7841 @item v4si __buitlin_ia32_cmpnleps (v4sf, v4sf)
7842 Generates the @code{cmpleps} machine instruction.
7843 @item v4si __buitlin_ia32_cmpngtps (v4sf, v4sf)
7844 Generates the @code{cmpgtps} machine instruction.
7845 @item v4si __buitlin_ia32_cmpngeps (v4sf, v4sf)
7846 Generates the @code{cmpgeps} machine instruction.
7847 @item v4si __buitlin_ia32_cmpordps (v4sf, v4sf)
7848 Generates the @code{cmpunodps} machine instruction.
7850 @item v4si __buitlin_ia32_cmpeqss (v4sf, v4sf)
7851 Generates the @code{cmpeqss} machine instruction.
7852 @item v4si __buitlin_ia32_cmpltss (v4sf, v4sf)
7853 Generates the @code{cmpltss} machine instruction.
7854 @item v4si __buitlin_ia32_cmpless (v4sf, v4sf)
7855 Generates the @code{cmpless} machine instruction.
7856 @item v4si __buitlin_ia32_cmpgtss (v4sf, v4sf)
7857 Generates the @code{cmpgtss} machine instruction.
7858 @item v4si __buitlin_ia32_cmpgess (v4sf, v4sf)
7859 Generates the @code{cmpgess} machine instruction.
7860 @item v4si __buitlin_ia32_cmpunordss (v4sf, v4sf)
7861 Generates the @code{cmpunodss} machine instruction.
7862 @item v4si __buitlin_ia32_cmpneqss (v4sf, v4sf)
7863 Generates the @code{cmpeqss} machine instruction.
7864 @item v4si __buitlin_ia32_cmpnlts (v4sf, v4sf)
7865 Generates the @code{cmpltss} machine instruction.
7866 @item v4si __buitlin_ia32_cmpnless (v4sf, v4sf)
7867 Generates the @code{cmpless} machine instruction.
7868 @item v4si __buitlin_ia32_cmpngtss (v4sf, v4sf)
7869 Generates the @code{cmpgtss} machine instruction.
7870 @item v4si __buitlin_ia32_cmpngess (v4sf, v4sf)
7871 Generates the @code{cmpgess} machine instruction.
7872 @item v4si __buitlin_ia32_cmpordss (v4sf, v4sf)
7873 Generates the @code{cmpunodss} machine instruction.
7875 @item v4sf __buitlin_ia32_maxps (v4sf, v4sf)
7876 Generates the @code{maxps} machine instruction.
7877 @item v4sf __buitlin_ia32_maxsss (v4sf, v4sf)
7878 Generates the @code{maxss} machine instruction.
7879 @item v4sf __buitlin_ia32_minps (v4sf, v4sf)
7880 Generates the @code{minps} machine instruction.
7881 @item v4sf __buitlin_ia32_minsss (v4sf, v4sf)
7882 Generates the @code{minss} machine instruction.
7884 @item ti __buitlin_ia32_andps (ti, ti)
7885 Generates the @code{andps} machine instruction.
7886 @item ti __buitlin_ia32_andnps (ti, ti)
7887 Generates the @code{andnps} machine instruction.
7888 @item ti __buitlin_ia32_orps (ti, ti)
7889 Generates the @code{orps} machine instruction.
7890 @item ti __buitlin_ia32_xorps (ti, ti)
7891 Generates the @code{xorps} machine instruction.
7893 @item v4sf __buitlin_ia32_movps (v4sf, v4sf)
7894 Generates the @code{movps} machine instruction.
7895 @item v4sf __buitlin_ia32_movhlps (v4sf, v4sf)
7896 Generates the @code{movhlps} machine instruction.
7897 @item v4sf __buitlin_ia32_movlhps (v4sf, v4sf)
7898 Generates the @code{movlhps} machine instruction.
7899 @item v4sf __buitlin_ia32_unpckhps (v4sf, v4sf)
7900 Generates the @code{unpckhps} machine instruction.
7901 @item v4sf __buitlin_ia32_unpcklps (v4sf, v4sf)
7902 Generates the @code{unpcklps} machine instruction.
7904 @item v4sf __buitlin_ia32_cvtpi2ps (v4sf, v2si)
7905 Generates the @code{cvtpi2ps} machine instruction.
7906 @item v2si __buitlin_ia32_cvtps2pi (v4sf)
7907 Generates the @code{cvtps2pi} machine instruction.
7908 @item v4sf __buitlin_ia32_cvtsi2ss (v4sf, int)
7909 Generates the @code{cvtsi2ss} machine instruction.
7910 @item int __buitlin_ia32_cvtss2si (v4sf)
7911 Generates the @code{cvtsi2ss} machine instruction.
7912 @item v2si __buitlin_ia32_cvttps2pi (v4sf)
7913 Generates the @code{cvttps2pi} machine instruction.
7914 @item int __buitlin_ia32_cvttss2si (v4sf)
7915 Generates the @code{cvttsi2ss} machine instruction.
7917 @item v4sf __buitlin_ia32_rcpps (v4sf)
7918 Generates the @code{rcpps} machine instruction.
7919 @item v4sf __buitlin_ia32_rsqrtps (v4sf)
7920 Generates the @code{rsqrtps} machine instruction.
7921 @item v4sf __buitlin_ia32_sqrtps (v4sf)
7922 Generates the @code{sqrtps} machine instruction.
7923 @item v4sf __buitlin_ia32_rcpss (v4sf)
7924 Generates the @code{rcpss} machine instruction.
7925 @item v4sf __buitlin_ia32_rsqrtss (v4sf)
7926 Generates the @code{rsqrtss} machine instruction.
7927 @item v4sf __buitlin_ia32_sqrtss (v4sf)
7928 Generates the @code{sqrtss} machine instruction.
7930 @item v4sf __buitlin_ia32_shufps (v4sf, v4sf, int)
7931 Generates the @code{shufps} machine instruction.
7933 @item v4sf __buitlin_ia32_loadaps (float *)
7934 Generates the @code{movaps} machine instruction as a load from memory.
7935 @item void __buitlin_ia32_storeaps (float *, v4sf)
7936 Generates the @code{movaps} machine instruction as a store to memory.
7937 @item v4sf __buitlin_ia32_loadups (float *)
7938 Generates the @code{movups} machine instruction as a load from memory.
7939 @item void __buitlin_ia32_storeups (float *, v4sf)
7940 Generates the @code{movups} machine instruction as a store to memory.
7941 @item v4sf __buitlin_ia32_loadsss (float *)
7942 Generates the @code{movss} machine instruction as a load from memory.
7943 @item void __buitlin_ia32_storess (float *, v4sf)
7944 Generates the @code{movss} machine instruction as a store to memory.
7946 @item v4sf __buitlin_ia32_loadhps (v4sf, v2si *)
7947 Generates the @code{movhps} machine instruction as a load from memory.
7948 @item v4sf __buitlin_ia32_loadlps (v4sf, v2si *)
7949 Generates the @code{movlps} machine instruction as a load from memory
7950 @item void __buitlin_ia32_storehps (v4sf, v2si *)
7951 Generates the @code{movhps} machine instruction as a store to memory.
7952 @item void __buitlin_ia32_storelps (v4sf, v2si *)
7953 Generates the @code{movlps} machine instruction as a store to memory.
7955 @item void __buitlin_ia32_movntps (float *, v4sf)
7956 Generates the @code{movntps} machine instruction.
7957 @item int __buitlin_ia32_movmskps (v4sf)
7958 Generates the @code{movntps} machine instruction.
7960 @item void __buitlin_ia32_storeps1 (float *, v4sf)
7961 Generates the @code{movaps} machine instruction as a store to memory.
7962 Before storing, the value is modified with a @code{shufps} instruction
7963 so that the lowest of the four floating point elements is replicated
7964 across the entire vector that is stored.
7965 @item void __buitlin_ia32_storerps (float *, v4sf)
7966 Generates the @code{movaps} machine instruction as a store to memory.
7967 Before storing, the value is modified with a @code{shufps} instruction
7968 so that the order of the four floating point elements in the vector is
7970 @item v4sf __buitlin_ia32_loadps1 (float *)
7971 Generates a @code{movss} machine instruction to load a floating point
7972 value from memory, and a @code{shufps} instruction to replicate the
7973 loaded value across all four elements of the result vector.
7974 @item v4sf __buitlin_ia32_loadrps (float *)
7975 Generates a @code{movaps} machine instruction to load a vector from
7976 memory, and a @code{shufps} instruction to reverse the order of the
7977 four floating point elements in the result vector.
7978 @item v4sf __builtin_ia32_setps (float, float, float, float)
7979 Constructs a vector from four single floating point values. The return
7980 value is equal to the value that would result from storing the four
7981 arguments into consecutive memory locations and then executing a
7982 @code{movaps} to load the vector from memory.
7983 @item v4sf __builtin_ia32_setps1 (float)
7984 Constructs a vector from a single floating point value by replicating
7985 it across all four elements of the result vector.
7989 @itemx -mno-push-args
7991 @opindex mno-push-args
7992 Use PUSH operations to store outgoing parameters. This method is shorter
7993 and usually equally fast as method using SUB/MOV operations and is enabled
7994 by default. In some cases disabling it may improve performance because of
7995 improved scheduling and reduced dependencies.
7997 @item -maccumulate-outgoing-args
7998 @opindex maccumulate-outgoing-args
7999 If enabled, the maximum amount of space required for outgoing arguments will be
8000 computed in the function prologue. This is faster on most modern CPUs
8001 because of reduced dependencies, improved scheduling and reduced stack usage
8002 when preferred stack boundary is not equal to 2. The drawback is a notable
8003 increase in code size. This switch implies @option{-mno-push-args}.
8007 Support thread-safe exception handling on @samp{Mingw32}. Code that relies
8008 on thread-safe exception handling must compile and link all code with the
8009 @option{-mthreads} option. When compiling, @option{-mthreads} defines
8010 @option{-D_MT}; when linking, it links in a special thread helper library
8011 @option{-lmingwthrd} which cleans up per thread exception handling data.
8013 @item -mno-align-stringops
8014 @opindex mno-align-stringops
8015 Do not align destination of inlined string operations. This switch reduces
8016 code size and improves performance in case the destination is already aligned,
8017 but gcc don't know about it.
8019 @item -minline-all-stringops
8020 @opindex minline-all-stringops
8021 By default GCC inlines string operations only when destination is known to be
8022 aligned at least to 4 byte boundary. This enables more inlining, increase code
8023 size, but may improve performance of code that depends on fast memcpy, strlen
8024 and memset for short lengths.
8026 @item -momit-leaf-frame-pointer
8027 @opindex momit-leaf-frame-pointer
8028 Don't keep the frame pointer in a register for leaf functions. This
8029 avoids the instructions to save, set up and restore frame pointers and
8030 makes an extra register available in leaf functions. The option
8031 @option{-fomit-frame-pointer} removes the frame pointer for all functions
8032 which might make debugging harder.
8035 These @samp{-m} switches are supported in addition to the above
8036 on AMD x86-64 processors in 64-bit environments.
8043 Generate code for a 32-bit or 64-bit environment.
8044 The 32-bit environment sets int, long and pointer to 32 bits and
8045 generates code that runs on any i386 system.
8046 The 64-bit environment sets int to 32 bits and long and pointer
8047 to 64 bits and generates code for AMD's x86-64 architecture.
8050 @opindex no-red-zone
8051 Do not use a so called red zone for x86-64 code. The red zone is mandated
8052 by the x86-64 ABI, it is a 128-byte area beyond the location of the
8053 stack pointer that will not be modified by signal or interrupt handlers
8054 and therefore can be used for temporary data without adjusting the stack
8055 pointer. The flag @option{-mno-red-zone} disables this red zone.
8059 @subsection HPPA Options
8060 @cindex HPPA Options
8062 These @samp{-m} options are defined for the HPPA family of computers:
8065 @item -march=@var{architecture-type}
8067 Generate code for the specified architecture. The choices for
8068 @var{architecture-type} are @samp{1.0} for PA 1.0, @samp{1.1} for PA
8069 1.1, and @samp{2.0} for PA 2.0 processors. Refer to
8070 @file{/usr/lib/sched.models} on an HP-UX system to determine the proper
8071 architecture option for your machine. Code compiled for lower numbered
8072 architectures will run on higher numbered architectures, but not the
8075 PA 2.0 support currently requires gas snapshot 19990413 or later. The
8076 next release of binutils (current is 2.9.1) will probably contain PA 2.0
8080 @itemx -mpa-risc-1-1
8081 @itemx -mpa-risc-2-0
8082 @opindex mpa-risc-1-0
8083 @opindex mpa-risc-1-1
8084 @opindex mpa-risc-2-0
8085 Synonyms for @option{-march=1.0}, @option{-march=1.1}, and @option{-march=2.0} respectively.
8088 @opindex mbig-switch
8089 Generate code suitable for big switch tables. Use this option only if
8090 the assembler/linker complain about out of range branches within a switch
8093 @item -mjump-in-delay
8094 @opindex mjump-in-delay
8095 Fill delay slots of function calls with unconditional jump instructions
8096 by modifying the return pointer for the function call to be the target
8097 of the conditional jump.
8099 @item -mdisable-fpregs
8100 @opindex mdisable-fpregs
8101 Prevent floating point registers from being used in any manner. This is
8102 necessary for compiling kernels which perform lazy context switching of
8103 floating point registers. If you use this option and attempt to perform
8104 floating point operations, the compiler will abort.
8106 @item -mdisable-indexing
8107 @opindex mdisable-indexing
8108 Prevent the compiler from using indexing address modes. This avoids some
8109 rather obscure problems when compiling MIG generated code under MACH@.
8111 @item -mno-space-regs
8112 @opindex mno-space-regs
8113 Generate code that assumes the target has no space registers. This allows
8114 GCC to generate faster indirect calls and use unscaled index address modes.
8116 Such code is suitable for level 0 PA systems and kernels.
8118 @item -mfast-indirect-calls
8119 @opindex mfast-indirect-calls
8120 Generate code that assumes calls never cross space boundaries. This
8121 allows GCC to emit code which performs faster indirect calls.
8123 This option will not work in the presence of shared libraries or nested
8126 @item -mlong-load-store
8127 @opindex mlong-load-store
8128 Generate 3-instruction load and store sequences as sometimes required by
8129 the HP-UX 10 linker. This is equivalent to the @samp{+k} option to
8132 @item -mportable-runtime
8133 @opindex mportable-runtime
8134 Use the portable calling conventions proposed by HP for ELF systems.
8138 Enable the use of assembler directives only GAS understands.
8140 @item -mschedule=@var{cpu-type}
8142 Schedule code according to the constraints for the machine type
8143 @var{cpu-type}. The choices for @var{cpu-type} are @samp{700}
8144 @samp{7100}, @samp{7100LC}, @samp{7200}, and @samp{8000}. Refer to
8145 @file{/usr/lib/sched.models} on an HP-UX system to determine the
8146 proper scheduling option for your machine.
8149 @opindex mlinker-opt
8150 Enable the optimization pass in the HPUX linker. Note this makes symbolic
8151 debugging impossible. It also triggers a bug in the HPUX 8 and HPUX 9 linkers
8152 in which they give bogus error messages when linking some programs.
8155 @opindex msoft-float
8156 Generate output containing library calls for floating point.
8157 @strong{Warning:} the requisite libraries are not available for all HPPA
8158 targets. Normally the facilities of the machine's usual C compiler are
8159 used, but this cannot be done directly in cross-compilation. You must make
8160 your own arrangements to provide suitable library functions for
8161 cross-compilation. The embedded target @samp{hppa1.1-*-pro}
8162 does provide software floating point support.
8164 @option{-msoft-float} changes the calling convention in the output file;
8165 therefore, it is only useful if you compile @emph{all} of a program with
8166 this option. In particular, you need to compile @file{libgcc.a}, the
8167 library that comes with GCC, with @option{-msoft-float} in order for
8171 @node Intel 960 Options
8172 @subsection Intel 960 Options
8174 These @samp{-m} options are defined for the Intel 960 implementations:
8177 @item -m@var{cpu-type}
8185 Assume the defaults for the machine type @var{cpu-type} for some of
8186 the other options, including instruction scheduling, floating point
8187 support, and addressing modes. The choices for @var{cpu-type} are
8188 @samp{ka}, @samp{kb}, @samp{mc}, @samp{ca}, @samp{cf},
8189 @samp{sa}, and @samp{sb}.
8196 @opindex msoft-float
8197 The @option{-mnumerics} option indicates that the processor does support
8198 floating-point instructions. The @option{-msoft-float} option indicates
8199 that floating-point support should not be assumed.
8201 @item -mleaf-procedures
8202 @itemx -mno-leaf-procedures
8203 @opindex mleaf-procedures
8204 @opindex mno-leaf-procedures
8205 Do (or do not) attempt to alter leaf procedures to be callable with the
8206 @code{bal} instruction as well as @code{call}. This will result in more
8207 efficient code for explicit calls when the @code{bal} instruction can be
8208 substituted by the assembler or linker, but less efficient code in other
8209 cases, such as calls via function pointers, or using a linker that doesn't
8210 support this optimization.
8213 @itemx -mno-tail-call
8215 @opindex mno-tail-call
8216 Do (or do not) make additional attempts (beyond those of the
8217 machine-independent portions of the compiler) to optimize tail-recursive
8218 calls into branches. You may not want to do this because the detection of
8219 cases where this is not valid is not totally complete. The default is
8220 @option{-mno-tail-call}.
8222 @item -mcomplex-addr
8223 @itemx -mno-complex-addr
8224 @opindex mcomplex-addr
8225 @opindex mno-complex-addr
8226 Assume (or do not assume) that the use of a complex addressing mode is a
8227 win on this implementation of the i960. Complex addressing modes may not
8228 be worthwhile on the K-series, but they definitely are on the C-series.
8229 The default is currently @option{-mcomplex-addr} for all processors except
8233 @itemx -mno-code-align
8234 @opindex mcode-align
8235 @opindex mno-code-align
8236 Align code to 8-byte boundaries for faster fetching (or don't bother).
8237 Currently turned on by default for C-series implementations only.
8240 @item -mclean-linkage
8241 @itemx -mno-clean-linkage
8242 @opindex mclean-linkage
8243 @opindex mno-clean-linkage
8244 These options are not fully implemented.
8248 @itemx -mic2.0-compat
8249 @itemx -mic3.0-compat
8251 @opindex mic2.0-compat
8252 @opindex mic3.0-compat
8253 Enable compatibility with iC960 v2.0 or v3.0.
8257 @opindex masm-compat
8259 Enable compatibility with the iC960 assembler.
8261 @item -mstrict-align
8262 @itemx -mno-strict-align
8263 @opindex mstrict-align
8264 @opindex mno-strict-align
8265 Do not permit (do permit) unaligned accesses.
8269 Enable structure-alignment compatibility with Intel's gcc release version
8270 1.3 (based on gcc 1.37). This option implies @option{-mstrict-align}.
8272 @item -mlong-double-64
8273 @opindex mlong-double-64
8274 Implement type @samp{long double} as 64-bit floating point numbers.
8275 Without the option @samp{long double} is implemented by 80-bit
8276 floating point numbers. The only reason we have it because there is
8277 no 128-bit @samp{long double} support in @samp{fp-bit.c} yet. So it
8278 is only useful for people using soft-float targets. Otherwise, we
8279 should recommend against use of it.
8283 @node DEC Alpha Options
8284 @subsection DEC Alpha Options
8286 These @samp{-m} options are defined for the DEC Alpha implementations:
8289 @item -mno-soft-float
8291 @opindex mno-soft-float
8292 @opindex msoft-float
8293 Use (do not use) the hardware floating-point instructions for
8294 floating-point operations. When @option{-msoft-float} is specified,
8295 functions in @file{libgcc.a} will be used to perform floating-point
8296 operations. Unless they are replaced by routines that emulate the
8297 floating-point operations, or compiled in such a way as to call such
8298 emulations routines, these routines will issue floating-point
8299 operations. If you are compiling for an Alpha without floating-point
8300 operations, you must ensure that the library is built so as not to call
8303 Note that Alpha implementations without floating-point operations are
8304 required to have floating-point registers.
8309 @opindex mno-fp-regs
8310 Generate code that uses (does not use) the floating-point register set.
8311 @option{-mno-fp-regs} implies @option{-msoft-float}. If the floating-point
8312 register set is not used, floating point operands are passed in integer
8313 registers as if they were integers and floating-point results are passed
8314 in $0 instead of $f0. This is a non-standard calling sequence, so any
8315 function with a floating-point argument or return value called by code
8316 compiled with @option{-mno-fp-regs} must also be compiled with that
8319 A typical use of this option is building a kernel that does not use,
8320 and hence need not save and restore, any floating-point registers.
8324 The Alpha architecture implements floating-point hardware optimized for
8325 maximum performance. It is mostly compliant with the IEEE floating
8326 point standard. However, for full compliance, software assistance is
8327 required. This option generates code fully IEEE compliant code
8328 @emph{except} that the @var{inexact-flag} is not maintained (see below).
8329 If this option is turned on, the preprocessor macro @code{_IEEE_FP} is
8330 defined during compilation. The resulting code is less efficient but is
8331 able to correctly support denormalized numbers and exceptional IEEE
8332 values such as not-a-number and plus/minus infinity. Other Alpha
8333 compilers call this option @option{-ieee_with_no_inexact}.
8335 @item -mieee-with-inexact
8336 @opindex mieee-with-inexact
8337 This is like @option{-mieee} except the generated code also maintains
8338 the IEEE @var{inexact-flag}. Turning on this option causes the
8339 generated code to implement fully-compliant IEEE math. In addition to
8340 @code{_IEEE_FP}, @code{_IEEE_FP_EXACT} is defined as a preprocessor
8341 macro. On some Alpha implementations the resulting code may execute
8342 significantly slower than the code generated by default. Since there is
8343 very little code that depends on the @var{inexact-flag}, you should
8344 normally not specify this option. Other Alpha compilers call this
8345 option @option{-ieee_with_inexact}.
8347 @item -mfp-trap-mode=@var{trap-mode}
8348 @opindex mfp-trap-mode
8349 This option controls what floating-point related traps are enabled.
8350 Other Alpha compilers call this option @option{-fptm @var{trap-mode}}.
8351 The trap mode can be set to one of four values:
8355 This is the default (normal) setting. The only traps that are enabled
8356 are the ones that cannot be disabled in software (e.g., division by zero
8360 In addition to the traps enabled by @samp{n}, underflow traps are enabled
8364 Like @samp{su}, but the instructions are marked to be safe for software
8365 completion (see Alpha architecture manual for details).
8368 Like @samp{su}, but inexact traps are enabled as well.
8371 @item -mfp-rounding-mode=@var{rounding-mode}
8372 @opindex mfp-rounding-mode
8373 Selects the IEEE rounding mode. Other Alpha compilers call this option
8374 @option{-fprm @var{rounding-mode}}. The @var{rounding-mode} can be one
8379 Normal IEEE rounding mode. Floating point numbers are rounded towards
8380 the nearest machine number or towards the even machine number in case
8384 Round towards minus infinity.
8387 Chopped rounding mode. Floating point numbers are rounded towards zero.
8390 Dynamic rounding mode. A field in the floating point control register
8391 (@var{fpcr}, see Alpha architecture reference manual) controls the
8392 rounding mode in effect. The C library initializes this register for
8393 rounding towards plus infinity. Thus, unless your program modifies the
8394 @var{fpcr}, @samp{d} corresponds to round towards plus infinity.
8397 @item -mtrap-precision=@var{trap-precision}
8398 @opindex mtrap-precision
8399 In the Alpha architecture, floating point traps are imprecise. This
8400 means without software assistance it is impossible to recover from a
8401 floating trap and program execution normally needs to be terminated.
8402 GCC can generate code that can assist operating system trap handlers
8403 in determining the exact location that caused a floating point trap.
8404 Depending on the requirements of an application, different levels of
8405 precisions can be selected:
8409 Program precision. This option is the default and means a trap handler
8410 can only identify which program caused a floating point exception.
8413 Function precision. The trap handler can determine the function that
8414 caused a floating point exception.
8417 Instruction precision. The trap handler can determine the exact
8418 instruction that caused a floating point exception.
8421 Other Alpha compilers provide the equivalent options called
8422 @option{-scope_safe} and @option{-resumption_safe}.
8424 @item -mieee-conformant
8425 @opindex mieee-conformant
8426 This option marks the generated code as IEEE conformant. You must not
8427 use this option unless you also specify @option{-mtrap-precision=i} and either
8428 @option{-mfp-trap-mode=su} or @option{-mfp-trap-mode=sui}. Its only effect
8429 is to emit the line @samp{.eflag 48} in the function prologue of the
8430 generated assembly file. Under DEC Unix, this has the effect that
8431 IEEE-conformant math library routines will be linked in.
8433 @item -mbuild-constants
8434 @opindex mbuild-constants
8435 Normally GCC examines a 32- or 64-bit integer constant to
8436 see if it can construct it from smaller constants in two or three
8437 instructions. If it cannot, it will output the constant as a literal and
8438 generate code to load it from the data segment at runtime.
8440 Use this option to require GCC to construct @emph{all} integer constants
8441 using code, even if it takes more instructions (the maximum is six).
8443 You would typically use this option to build a shared library dynamic
8444 loader. Itself a shared library, it must relocate itself in memory
8445 before it can find the variables and constants in its own data segment.
8451 Select whether to generate code to be assembled by the vendor-supplied
8452 assembler (@option{-malpha-as}) or by the GNU assembler @option{-mgas}.
8466 Indicate whether GCC should generate code to use the optional BWX,
8467 CIX, and MAX instruction sets. The default is to use the instruction sets
8468 supported by the CPU type specified via @option{-mcpu=} option or that
8469 of the CPU on which GCC was built if none was specified.
8471 @item -mcpu=@var{cpu_type}
8473 Set the instruction set, register set, and instruction scheduling
8474 parameters for machine type @var{cpu_type}. You can specify either the
8475 @samp{EV} style name or the corresponding chip number. GCC
8476 supports scheduling parameters for the EV4 and EV5 family of processors
8477 and will choose the default values for the instruction set from
8478 the processor you specify. If you do not specify a processor type,
8479 GCC will default to the processor on which the compiler was built.
8481 Supported values for @var{cpu_type} are
8486 Schedules as an EV4 and has no instruction set extensions.
8490 Schedules as an EV5 and has no instruction set extensions.
8494 Schedules as an EV5 and supports the BWX extension.
8499 Schedules as an EV5 and supports the BWX and MAX extensions.
8503 Schedules as an EV5 (until Digital releases the scheduling parameters
8504 for the EV6) and supports the BWX, CIX, and MAX extensions.
8507 @item -mmemory-latency=@var{time}
8508 @opindex mmemory-latency
8509 Sets the latency the scheduler should assume for typical memory
8510 references as seen by the application. This number is highly
8511 dependent on the memory access patterns used by the application
8512 and the size of the external cache on the machine.
8514 Valid options for @var{time} are
8518 A decimal number representing clock cycles.
8524 The compiler contains estimates of the number of clock cycles for
8525 ``typical'' EV4 & EV5 hardware for the Level 1, 2 & 3 caches
8526 (also called Dcache, Scache, and Bcache), as well as to main memory.
8527 Note that L3 is only valid for EV5.
8532 @node Clipper Options
8533 @subsection Clipper Options
8535 These @samp{-m} options are defined for the Clipper implementations:
8540 Produce code for a C300 Clipper processor. This is the default.
8544 Produce code for a C400 Clipper processor, i.e.@: use floating point
8548 @node H8/300 Options
8549 @subsection H8/300 Options
8551 These @samp{-m} options are defined for the H8/300 implementations:
8556 Shorten some address references at link time, when possible; uses the
8557 linker option @option{-relax}. @xref{H8/300,, @code{ld} and the H8/300,
8558 ld.info, Using ld}, for a fuller description.
8562 Generate code for the H8/300H@.
8566 Generate code for the H8/S@.
8570 Generate code for the H8/S2600. This switch must be used with @option{-ms}.
8574 Make @code{int} data 32 bits by default.
8578 On the H8/300H and H8/S, use the same alignment rules as for the H8/300.
8579 The default for the H8/300H and H8/S is to align longs and floats on 4
8581 @option{-malign-300} causes them to be aligned on 2 byte boundaries.
8582 This option has no effect on the H8/300.
8586 @subsection SH Options
8588 These @samp{-m} options are defined for the SH implementations:
8593 Generate code for the SH1.
8597 Generate code for the SH2.
8601 Generate code for the SH3.
8605 Generate code for the SH3e.
8609 Generate code for the SH4 without a floating-point unit.
8611 @item -m4-single-only
8612 @opindex m4-single-only
8613 Generate code for the SH4 with a floating-point unit that only
8614 supports single-precision arithmetic.
8618 Generate code for the SH4 assuming the floating-point unit is in
8619 single-precision mode by default.
8623 Generate code for the SH4.
8627 Compile code for the processor in big endian mode.
8631 Compile code for the processor in little endian mode.
8635 Align doubles at 64-bit boundaries. Note that this changes the calling
8636 conventions, and thus some functions from the standard C library will
8637 not work unless you recompile it first with @option{-mdalign}.
8641 Shorten some address references at link time, when possible; uses the
8642 linker option @option{-relax}.
8646 Use 32-bit offsets in @code{switch} tables. The default is to use
8651 Enable the use of the instruction @code{fmovd}.
8655 Comply with the calling conventions defined by Hitachi.
8659 Mark the @code{MAC} register as call-clobbered, even if
8660 @option{-mhitachi} is given.
8664 Increase IEEE-compliance of floating-point code.
8668 Dump instruction size and location in the assembly code.
8672 This option is deprecated. It pads structures to multiple of 4 bytes,
8673 which is incompatible with the SH ABI@.
8677 Optimize for space instead of speed. Implied by @option{-Os}.
8681 When generating position-independent code, emit function calls using
8682 the Global Offset Table instead of the Procedure Linkage Table.
8686 Generate a library function call to invalidate instruction cache
8687 entries, after fixing up a trampoline. This library function call
8688 doesn't assume it can write to the whole memory address space. This
8689 is the default when the target is @code{sh-*-linux*}.
8692 @node System V Options
8693 @subsection Options for System V
8695 These additional options are available on System V Release 4 for
8696 compatibility with other compilers on those systems:
8701 Create a shared object.
8702 It is recommended that @option{-symbolic} or @option{-shared} be used instead.
8706 Identify the versions of each tool used by the compiler, in a
8707 @code{.ident} assembler directive in the output.
8711 Refrain from adding @code{.ident} directives to the output file (this is
8714 @item -YP,@var{dirs}
8716 Search the directories @var{dirs}, and no others, for libraries
8717 specified with @option{-l}.
8721 Look in the directory @var{dir} to find the M4 preprocessor.
8722 The assembler uses this option.
8723 @c This is supposed to go with a -Yd for predefined M4 macro files, but
8724 @c the generic assembler that comes with Solaris takes just -Ym.
8727 @node TMS320C3x/C4x Options
8728 @subsection TMS320C3x/C4x Options
8729 @cindex TMS320C3x/C4x Options
8731 These @samp{-m} options are defined for TMS320C3x/C4x implementations:
8735 @item -mcpu=@var{cpu_type}
8737 Set the instruction set, register set, and instruction scheduling
8738 parameters for machine type @var{cpu_type}. Supported values for
8739 @var{cpu_type} are @samp{c30}, @samp{c31}, @samp{c32}, @samp{c40}, and
8740 @samp{c44}. The default is @samp{c40} to generate code for the
8745 @itemx -msmall-memory
8747 @opindex mbig-memory
8749 @opindex msmall-memory
8751 Generates code for the big or small memory model. The small memory
8752 model assumed that all data fits into one 64K word page. At run-time
8753 the data page (DP) register must be set to point to the 64K page
8754 containing the .bss and .data program sections. The big memory model is
8755 the default and requires reloading of the DP register for every direct
8762 Allow (disallow) allocation of general integer operands into the block
8769 Enable (disable) generation of code using decrement and branch,
8770 DBcond(D), instructions. This is enabled by default for the C4x. To be
8771 on the safe side, this is disabled for the C3x, since the maximum
8772 iteration count on the C3x is @math{2^23 + 1} (but who iterates loops more than
8773 @math{2^23} times on the C3x?). Note that GCC will try to reverse a loop so
8774 that it can utilise the decrement and branch instruction, but will give
8775 up if there is more than one memory reference in the loop. Thus a loop
8776 where the loop counter is decremented can generate slightly more
8777 efficient code, in cases where the RPTB instruction cannot be utilised.
8779 @item -mdp-isr-reload
8781 @opindex mdp-isr-reload
8783 Force the DP register to be saved on entry to an interrupt service
8784 routine (ISR), reloaded to point to the data section, and restored on
8785 exit from the ISR@. This should not be required unless someone has
8786 violated the small memory model by modifying the DP register, say within
8793 For the C3x use the 24-bit MPYI instruction for integer multiplies
8794 instead of a library call to guarantee 32-bit results. Note that if one
8795 of the operands is a constant, then the multiplication will be performed
8796 using shifts and adds. If the @option{-mmpyi} option is not specified for the C3x,
8797 then squaring operations are performed inline instead of a library call.
8800 @itemx -mno-fast-fix
8802 @opindex mno-fast-fix
8803 The C3x/C4x FIX instruction to convert a floating point value to an
8804 integer value chooses the nearest integer less than or equal to the
8805 floating point value rather than to the nearest integer. Thus if the
8806 floating point number is negative, the result will be incorrectly
8807 truncated an additional code is necessary to detect and correct this
8808 case. This option can be used to disable generation of the additional
8809 code required to correct the result.
8815 Enable (disable) generation of repeat block sequences using the RPTB
8816 instruction for zero overhead looping. The RPTB construct is only used
8817 for innermost loops that do not call functions or jump across the loop
8818 boundaries. There is no advantage having nested RPTB loops due to the
8819 overhead required to save and restore the RC, RS, and RE registers.
8820 This is enabled by default with @option{-O2}.
8822 @item -mrpts=@var{count}
8826 Enable (disable) the use of the single instruction repeat instruction
8827 RPTS@. If a repeat block contains a single instruction, and the loop
8828 count can be guaranteed to be less than the value @var{count}, GCC will
8829 emit a RPTS instruction instead of a RPTB@. If no value is specified,
8830 then a RPTS will be emitted even if the loop count cannot be determined
8831 at compile time. Note that the repeated instruction following RPTS does
8832 not have to be reloaded from memory each iteration, thus freeing up the
8833 CPU buses for operands. However, since interrupts are blocked by this
8834 instruction, it is disabled by default.
8836 @item -mloop-unsigned
8837 @itemx -mno-loop-unsigned
8838 @opindex mloop-unsigned
8839 @opindex mno-loop-unsigned
8840 The maximum iteration count when using RPTS and RPTB (and DB on the C40)
8841 is @math{2^31 + 1} since these instructions test if the iteration count is
8842 negative to terminate the loop. If the iteration count is unsigned
8843 there is a possibility than the @math{2^31 + 1} maximum iteration count may be
8844 exceeded. This switch allows an unsigned iteration count.
8848 Try to emit an assembler syntax that the TI assembler (asm30) is happy
8849 with. This also enforces compatibility with the API employed by the TI
8850 C3x C compiler. For example, long doubles are passed as structures
8851 rather than in floating point registers.
8857 Generate code that uses registers (stack) for passing arguments to functions.
8858 By default, arguments are passed in registers where possible rather
8859 than by pushing arguments on to the stack.
8861 @item -mparallel-insns
8862 @itemx -mno-parallel-insns
8863 @opindex mparallel-insns
8864 @opindex mno-parallel-insns
8865 Allow the generation of parallel instructions. This is enabled by
8866 default with @option{-O2}.
8868 @item -mparallel-mpy
8869 @itemx -mno-parallel-mpy
8870 @opindex mparallel-mpy
8871 @opindex mno-parallel-mpy
8872 Allow the generation of MPY||ADD and MPY||SUB parallel instructions,
8873 provided @option{-mparallel-insns} is also specified. These instructions have
8874 tight register constraints which can pessimize the code generation
8880 @subsection V850 Options
8881 @cindex V850 Options
8883 These @samp{-m} options are defined for V850 implementations:
8887 @itemx -mno-long-calls
8888 @opindex mlong-calls
8889 @opindex mno-long-calls
8890 Treat all calls as being far away (near). If calls are assumed to be
8891 far away, the compiler will always load the functions address up into a
8892 register, and call indirect through the pointer.
8898 Do not optimize (do optimize) basic blocks that use the same index
8899 pointer 4 or more times to copy pointer into the @code{ep} register, and
8900 use the shorter @code{sld} and @code{sst} instructions. The @option{-mep}
8901 option is on by default if you optimize.
8903 @item -mno-prolog-function
8904 @itemx -mprolog-function
8905 @opindex mno-prolog-function
8906 @opindex mprolog-function
8907 Do not use (do use) external functions to save and restore registers at
8908 the prolog and epilog of a function. The external functions are slower,
8909 but use less code space if more than one function saves the same number
8910 of registers. The @option{-mprolog-function} option is on by default if
8915 Try to make the code as small as possible. At present, this just turns
8916 on the @option{-mep} and @option{-mprolog-function} options.
8920 Put static or global variables whose size is @var{n} bytes or less into
8921 the tiny data area that register @code{ep} points to. The tiny data
8922 area can hold up to 256 bytes in total (128 bytes for byte references).
8926 Put static or global variables whose size is @var{n} bytes or less into
8927 the small data area that register @code{gp} points to. The small data
8928 area can hold up to 64 kilobytes.
8932 Put static or global variables whose size is @var{n} bytes or less into
8933 the first 32 kilobytes of memory.
8937 Specify that the target processor is the V850.
8940 @opindex mbig-switch
8941 Generate code suitable for big switch tables. Use this option only if
8942 the assembler/linker complain about out of range branches within a switch
8947 @subsection ARC Options
8950 These options are defined for ARC implementations:
8955 Compile code for little endian mode. This is the default.
8959 Compile code for big endian mode.
8962 @opindex mmangle-cpu
8963 Prepend the name of the cpu to all public symbol names.
8964 In multiple-processor systems, there are many ARC variants with different
8965 instruction and register set characteristics. This flag prevents code
8966 compiled for one cpu to be linked with code compiled for another.
8967 No facility exists for handling variants that are ``almost identical''.
8968 This is an all or nothing option.
8970 @item -mcpu=@var{cpu}
8972 Compile code for ARC variant @var{cpu}.
8973 Which variants are supported depend on the configuration.
8974 All variants support @option{-mcpu=base}, this is the default.
8976 @item -mtext=@var{text-section}
8977 @itemx -mdata=@var{data-section}
8978 @itemx -mrodata=@var{readonly-data-section}
8982 Put functions, data, and readonly data in @var{text-section},
8983 @var{data-section}, and @var{readonly-data-section} respectively
8984 by default. This can be overridden with the @code{section} attribute.
8985 @xref{Variable Attributes}.
8990 @subsection NS32K Options
8991 @cindex NS32K options
8993 These are the @samp{-m} options defined for the 32000 series. The default
8994 values for these options depends on which style of 32000 was selected when
8995 the compiler was configured; the defaults for the most common choices are
9003 Generate output for a 32032. This is the default
9004 when the compiler is configured for 32032 and 32016 based systems.
9010 Generate output for a 32332. This is the default
9011 when the compiler is configured for 32332-based systems.
9017 Generate output for a 32532. This is the default
9018 when the compiler is configured for 32532-based systems.
9022 Generate output containing 32081 instructions for floating point.
9023 This is the default for all systems.
9027 Generate output containing 32381 instructions for floating point. This
9028 also implies @option{-m32081}. The 32381 is only compatible with the 32332
9029 and 32532 cpus. This is the default for the pc532-netbsd configuration.
9033 Try and generate multiply-add floating point instructions @code{polyF}
9034 and @code{dotF}. This option is only available if the @option{-m32381}
9035 option is in effect. Using these instructions requires changes to to
9036 register allocation which generally has a negative impact on
9037 performance. This option should only be enabled when compiling code
9038 particularly likely to make heavy use of multiply-add instructions.
9041 @opindex mnomulti-add
9042 Do not try and generate multiply-add floating point instructions
9043 @code{polyF} and @code{dotF}. This is the default on all platforms.
9046 @opindex msoft-float
9047 Generate output containing library calls for floating point.
9048 @strong{Warning:} the requisite libraries may not be available.
9051 @opindex mnobitfield
9052 Do not use the bit-field instructions. On some machines it is faster to
9053 use shifting and masking operations. This is the default for the pc532.
9057 Do use the bit-field instructions. This is the default for all platforms
9062 Use a different function-calling convention, in which functions
9063 that take a fixed number of arguments return pop their
9064 arguments on return with the @code{ret} instruction.
9066 This calling convention is incompatible with the one normally
9067 used on Unix, so you cannot use it if you need to call libraries
9068 compiled with the Unix compiler.
9070 Also, you must provide function prototypes for all functions that
9071 take variable numbers of arguments (including @code{printf});
9072 otherwise incorrect code will be generated for calls to those
9075 In addition, seriously incorrect code will result if you call a
9076 function with too many arguments. (Normally, extra arguments are
9077 harmlessly ignored.)
9079 This option takes its name from the 680x0 @code{rtd} instruction.
9084 Use a different function-calling convention where the first two arguments
9085 are passed in registers.
9087 This calling convention is incompatible with the one normally
9088 used on Unix, so you cannot use it if you need to call libraries
9089 compiled with the Unix compiler.
9092 @opindex mnoregparam
9093 Do not pass any arguments in registers. This is the default for all
9098 It is OK to use the sb as an index register which is always loaded with
9099 zero. This is the default for the pc532-netbsd target.
9103 The sb register is not available for use or has not been initialized to
9104 zero by the run time system. This is the default for all targets except
9105 the pc532-netbsd. It is also implied whenever @option{-mhimem} or
9106 @option{-fpic} is set.
9110 Many ns32000 series addressing modes use displacements of up to 512MB@.
9111 If an address is above 512MB then displacements from zero can not be used.
9112 This option causes code to be generated which can be loaded above 512MB@.
9113 This may be useful for operating systems or ROM code.
9117 Assume code will be loaded in the first 512MB of virtual address space.
9118 This is the default for all platforms.
9124 @subsection AVR Options
9127 These options are defined for AVR implementations:
9130 @item -mmcu=@var{mcu}
9132 Specify ATMEL AVR instruction set or MCU type.
9134 Instruction set avr1 is for the minimal AVR core, not supported by the C
9135 compiler, only for assembler programs (MCU types: at90s1200, attiny10,
9136 attiny11, attiny12, attiny15, attiny28).
9138 Instruction set avr2 (default) is for the classic AVR core with up to
9139 8K program memory space (MCU types: at90s2313, at90s2323, attiny22,
9140 at90s2333, at90s2343, at90s4414, at90s4433, at90s4434, at90s8515,
9141 at90c8534, at90s8535).
9143 Instruction set avr3 is for the classic AVR core with up to 128K program
9144 memory space (MCU types: atmega103, atmega603).
9146 Instruction set avr4 is for the enhanced AVR core with up to 8K program
9147 memory space (MCU types: atmega83, atmega85).
9149 Instruction set avr5 is for the enhanced AVR core with up to 128K program
9150 memory space (MCU types: atmega161, atmega163, atmega32, at94k).
9154 Output instruction sizes to the asm file.
9156 @item -minit-stack=@var{N}
9157 @opindex minit-stack
9158 Specify the initial stack address, which may be a symbol or numeric value,
9159 @samp{__stack} is the default.
9161 @item -mno-interrupts
9162 @opindex mno-interrupts
9163 Generated code is not compatible with hardware interrupts.
9164 Code size will be smaller.
9166 @item -mcall-prologues
9167 @opindex mcall-prologues
9168 Functions prologues/epilogues expanded as call to appropriate
9169 subroutines. Code size will be smaller.
9171 @item -mno-tablejump
9172 @opindex mno-tablejump
9173 Do not generate tablejump insns which sometimes increase code size.
9176 @opindex mtiny-stack
9177 Change only the low 8 bits of the stack pointer.
9181 @subsection MCore Options
9182 @cindex MCore options
9184 These are the @samp{-m} options defined for the Motorola M*Core
9194 @opindex mno-hardlit
9195 Inline constants into the code stream if it can be done in two
9196 instructions or less.
9204 Use the divide instruction. (Enabled by default).
9206 @item -mrelax-immediate
9207 @itemx -mrelax-immediate
9208 @itemx -mno-relax-immediate
9209 @opindex mrelax-immediate
9210 @opindex mrelax-immediate
9211 @opindex mno-relax-immediate
9212 Allow arbitrary sized immediates in bit operations.
9214 @item -mwide-bitfields
9215 @itemx -mwide-bitfields
9216 @itemx -mno-wide-bitfields
9217 @opindex mwide-bitfields
9218 @opindex mwide-bitfields
9219 @opindex mno-wide-bitfields
9220 Always treat bit-fields as int-sized.
9222 @item -m4byte-functions
9223 @itemx -m4byte-functions
9224 @itemx -mno-4byte-functions
9225 @opindex m4byte-functions
9226 @opindex m4byte-functions
9227 @opindex mno-4byte-functions
9228 Force all functions to be aligned to a four byte boundary.
9230 @item -mcallgraph-data
9231 @itemx -mcallgraph-data
9232 @itemx -mno-callgraph-data
9233 @opindex mcallgraph-data
9234 @opindex mcallgraph-data
9235 @opindex mno-callgraph-data
9236 Emit callgraph information.
9240 @itemx -mno-slow-bytes
9241 @opindex mslow-bytes
9242 @opindex mslow-bytes
9243 @opindex mno-slow-bytes
9244 Prefer word access when reading byte quantities.
9246 @item -mlittle-endian
9247 @itemx -mlittle-endian
9249 @opindex mlittle-endian
9250 @opindex mlittle-endian
9251 @opindex mbig-endian
9252 Generate code for a little endian target.
9260 Generate code for the 210 processor.
9264 @subsection IA-64 Options
9265 @cindex IA-64 Options
9267 These are the @samp{-m} options defined for the Intel IA-64 architecture.
9271 @opindex mbig-endian
9272 Generate code for a big endian target. This is the default for HPUX@.
9274 @item -mlittle-endian
9275 @opindex mlittle-endian
9276 Generate code for a little endian target. This is the default for AIX5
9283 Generate (or don't) code for the GNU assembler. This is the default.
9284 @c Also, this is the default if the configure option @option{--with-gnu-as}
9291 Generate (or don't) code for the GNU linker. This is the default.
9292 @c Also, this is the default if the configure option @option{--with-gnu-ld}
9297 Generate code that does not use a global pointer register. The result
9298 is not position independent code, and violates the IA-64 ABI@.
9300 @item -mvolatile-asm-stop
9301 @itemx -mno-volatile-asm-stop
9302 @opindex mvolatile-asm-stop
9303 @opindex mno-volatile-asm-stop
9304 Generate (or don't) a stop bit immediately before and after volatile asm
9309 Generate code that works around Itanium B step errata.
9311 @item -mregister-names
9312 @itemx -mno-register-names
9313 @opindex mregister-names
9314 @opindex mno-register-names
9315 Generate (or don't) @samp{in}, @samp{loc}, and @samp{out} register names for
9316 the stacked registers. This may make assembler output more readable.
9322 Disable (or enable) optimizations that use the small data section. This may
9323 be useful for working around optimizer bugs.
9326 @opindex mconstant-gp
9327 Generate code that uses a single constant global pointer value. This is
9328 useful when compiling kernel code.
9332 Generate code that is self-relocatable. This implies @option{-mconstant-gp}.
9333 This is useful when compiling firmware code.
9335 @item -minline-divide-min-latency
9336 @opindex minline-divide-min-latency
9337 Generate code for inline divides using the minimum latency algorithm.
9339 @item -minline-divide-max-throughput
9340 @opindex minline-divide-max-throughput
9341 Generate code for inline divides using the maximum throughput algorithm.
9343 @item -mno-dwarf2-asm
9345 @opindex mno-dwarf2-asm
9346 @opindex mdwarf2-asm
9347 Don't (or do) generate assembler code for the DWARF2 line number debugging
9348 info. This may be useful when not using the GNU assembler.
9350 @item -mfixed-range=@var{register-range}
9351 @opindex mfixed-range
9352 Generate code treating the given register range as fixed registers.
9353 A fixed register is one that the register allocator can not use. This is
9354 useful when compiling kernel code. A register range is specified as
9355 two registers separated by a dash. Multiple register ranges can be
9356 specified separated by a comma.
9360 @subsection D30V Options
9361 @cindex D30V Options
9363 These @samp{-m} options are defined for D30V implementations:
9368 Link the @samp{.text}, @samp{.data}, @samp{.bss}, @samp{.strings},
9369 @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections into external
9370 memory, which starts at location @code{0x80000000}.
9374 Same as the @option{-mextmem} switch.
9378 Link the @samp{.text} section into onchip text memory, which starts at
9379 location @code{0x0}. Also link @samp{.data}, @samp{.bss},
9380 @samp{.strings}, @samp{.rodata}, @samp{.rodata1}, @samp{.data1} sections
9381 into onchip data memory, which starts at location @code{0x20000000}.
9383 @item -mno-asm-optimize
9384 @itemx -masm-optimize
9385 @opindex mno-asm-optimize
9386 @opindex masm-optimize
9387 Disable (enable) passing @option{-O} to the assembler when optimizing.
9388 The assembler uses the @option{-O} option to automatically parallelize
9389 adjacent short instructions where possible.
9391 @item -mbranch-cost=@var{n}
9392 @opindex mbranch-cost
9393 Increase the internal costs of branches to @var{n}. Higher costs means
9394 that the compiler will issue more instructions to avoid doing a branch.
9397 @item -mcond-exec=@var{n}
9399 Specify the maximum number of conditionally executed instructions that
9400 replace a branch. The default is 4.
9403 @node S/390 and zSeries Options
9404 @subsection S/390 and zSeries Options
9405 @cindex S/390 and zSeries Options
9407 These are the @samp{-m} options defined for the S/390 and zSeries architecture.
9412 @opindex mhard-float
9413 @opindex msoft-float
9414 Use (do not use) the hardware floating-point instructions and registers
9415 for floating-point operations. When @option{-msoft-float} is specified,
9416 functions in @file{libgcc.a} will be used to perform floating-point
9417 operations. When @option{-mhard-float} is specified, the compiler
9418 generates IEEE floating-point instructions. This is the default.
9421 @itemx -mno-backchain
9423 @opindex mno-backchain
9424 Generate (or do not generate) code which maintains an explicit
9425 backchain within the stack frame that points to the caller's frame.
9426 This is currently needed to allow debugging. The default is to
9427 generate the backchain.
9430 @itemx -mno-small-exec
9431 @opindex msmall-exec
9432 @opindex mno-small-exec
9433 Generate (or do not generate) code using the @code{bras} instruction
9434 to do subroutine calls.
9435 This only works reliably if the total executable size does not
9436 exceed 64k. The default is to use the @code{basr} instruction instead,
9437 which does not have this limitation.
9443 When @option{-m31} is specified, generate code compliant to the
9444 Linux for S/390 ABI@. When @option{-m64} is specified, generate
9445 code compliant to the Linux for zSeries ABI@. This allows GCC in
9446 particular to generate 64-bit instructions. For the @samp{s390}
9447 targets, the default is @option{-m31}, while the @samp{s390x}
9448 targets default to @option{-m64}.
9454 Generate (or do not generate) code using the @code{mvcle} instruction
9455 to perform block moves. When @option{-mno-mvcle} is specifed,
9456 use a @code{mvc} loop instead. This is the default.
9462 Print (or do not print) additional debug information when compiling.
9463 The default is to not print debug information.
9468 @node Code Gen Options
9469 @section Options for Code Generation Conventions
9470 @cindex code generation conventions
9471 @cindex options, code generation
9472 @cindex run-time options
9474 These machine-independent options control the interface conventions
9475 used in code generation.
9477 Most of them have both positive and negative forms; the negative form
9478 of @option{-ffoo} would be @option{-fno-foo}. In the table below, only
9479 one of the forms is listed---the one which is not the default. You
9480 can figure out the other form by either removing @samp{no-} or adding
9485 @opindex fexceptions
9486 Enable exception handling. Generates extra code needed to propagate
9487 exceptions. For some targets, this implies GCC will generate frame
9488 unwind information for all functions, which can produce significant data
9489 size overhead, although it does not affect execution. If you do not
9490 specify this option, GCC will enable it by default for languages like
9491 C++ which normally require exception handling, and disable it for
9492 languages like C that do not normally require it. However, you may need
9493 to enable this option when compiling C code that needs to interoperate
9494 properly with exception handlers written in C++. You may also wish to
9495 disable this option if you are compiling older C++ programs that don't
9496 use exception handling.
9498 @item -fnon-call-exceptions
9499 @opindex fnon-call-exceptions
9500 Generate code that allows trapping instructions to throw exceptions.
9501 Note that this requires platform-specific runtime support that does
9502 not exist everywhere. Moreover, it only allows @emph{trapping}
9503 instructions to throw exceptions, i.e.@: memory references or floating
9504 point instructions. It does not allow exceptions to be thrown from
9505 arbitrary signal handlers such as @code{SIGALRM}.
9507 @item -funwind-tables
9508 @opindex funwind-tables
9509 Similar to @option{-fexceptions}, except that it will just generate any needed
9510 static data, but will not affect the generated code in any other way.
9511 You will normally not enable this option; instead, a language processor
9512 that needs this handling would enable it on your behalf.
9514 @item -fpcc-struct-return
9515 @opindex fpcc-struct-return
9516 Return ``short'' @code{struct} and @code{union} values in memory like
9517 longer ones, rather than in registers. This convention is less
9518 efficient, but it has the advantage of allowing intercallability between
9519 GCC-compiled files and files compiled with other compilers.
9521 The precise convention for returning structures in memory depends
9522 on the target configuration macros.
9524 Short structures and unions are those whose size and alignment match
9525 that of some integer type.
9527 @item -freg-struct-return
9528 @opindex freg-struct-return
9529 Return @code{struct} and @code{union} values in registers when possible.
9530 This is more efficient for small structures than
9531 @option{-fpcc-struct-return}.
9533 If you specify neither @option{-fpcc-struct-return} nor
9534 @option{-freg-struct-return}, GCC defaults to whichever convention is
9535 standard for the target. If there is no standard convention, GCC
9536 defaults to @option{-fpcc-struct-return}, except on targets where GCC is
9537 the principal compiler. In those cases, we can choose the standard, and
9538 we chose the more efficient register return alternative.
9541 @opindex fshort-enums
9542 Allocate to an @code{enum} type only as many bytes as it needs for the
9543 declared range of possible values. Specifically, the @code{enum} type
9544 will be equivalent to the smallest integer type which has enough room.
9546 @item -fshort-double
9547 @opindex fshort-double
9548 Use the same size for @code{double} as for @code{float}.
9551 @opindex fshared-data
9552 Requests that the data and non-@code{const} variables of this
9553 compilation be shared data rather than private data. The distinction
9554 makes sense only on certain operating systems, where shared data is
9555 shared between processes running the same program, while private data
9556 exists in one copy per process.
9560 In C, allocate even uninitialized global variables in the data section of the
9561 object file, rather than generating them as common blocks. This has the
9562 effect that if the same variable is declared (without @code{extern}) in
9563 two different compilations, you will get an error when you link them.
9564 The only reason this might be useful is if you wish to verify that the
9565 program will work on other systems which always work this way.
9569 Ignore the @samp{#ident} directive.
9571 @item -fno-gnu-linker
9572 @opindex fno-gnu-linker
9573 Do not output global initializations (such as C++ constructors and
9574 destructors) in the form used by the GNU linker (on systems where the GNU
9575 linker is the standard method of handling them). Use this option when
9576 you want to use a non-GNU linker, which also requires using the
9577 @command{collect2} program to make sure the system linker includes
9578 constructors and destructors. (@command{collect2} is included in the GCC
9579 distribution.) For systems which @emph{must} use @command{collect2}, the
9580 compiler driver @command{gcc} is configured to do this automatically.
9582 @item -finhibit-size-directive
9583 @opindex finhibit-size-directive
9584 Don't output a @code{.size} assembler directive, or anything else that
9585 would cause trouble if the function is split in the middle, and the
9586 two halves are placed at locations far apart in memory. This option is
9587 used when compiling @file{crtstuff.c}; you should not need to use it
9591 @opindex fverbose-asm
9592 Put extra commentary information in the generated assembly code to
9593 make it more readable. This option is generally only of use to those
9594 who actually need to read the generated assembly code (perhaps while
9595 debugging the compiler itself).
9597 @option{-fno-verbose-asm}, the default, causes the
9598 extra information to be omitted and is useful when comparing two assembler
9603 Consider all memory references through pointers to be volatile.
9605 @item -fvolatile-global
9606 @opindex fvolatile-global
9607 Consider all memory references to extern and global data items to
9608 be volatile. GCC does not consider static data items to be volatile
9609 because of this switch.
9611 @item -fvolatile-static
9612 @opindex fvolatile-static
9613 Consider all memory references to static data to be volatile.
9617 @cindex global offset table
9619 Generate position-independent code (PIC) suitable for use in a shared
9620 library, if supported for the target machine. Such code accesses all
9621 constant addresses through a global offset table (GOT)@. The dynamic
9622 loader resolves the GOT entries when the program starts (the dynamic
9623 loader is not part of GCC; it is part of the operating system). If
9624 the GOT size for the linked executable exceeds a machine-specific
9625 maximum size, you get an error message from the linker indicating that
9626 @option{-fpic} does not work; in that case, recompile with @option{-fPIC}
9627 instead. (These maximums are 16k on the m88k, 8k on the Sparc, and 32k
9628 on the m68k and RS/6000. The 386 has no such limit.)
9630 Position-independent code requires special support, and therefore works
9631 only on certain machines. For the 386, GCC supports PIC for System V
9632 but not for the Sun 386i. Code generated for the IBM RS/6000 is always
9633 position-independent.
9637 If supported for the target machine, emit position-independent code,
9638 suitable for dynamic linking and avoiding any limit on the size of the
9639 global offset table. This option makes a difference on the m68k, m88k,
9642 Position-independent code requires special support, and therefore works
9643 only on certain machines.
9645 @item -ffixed-@var{reg}
9647 Treat the register named @var{reg} as a fixed register; generated code
9648 should never refer to it (except perhaps as a stack pointer, frame
9649 pointer or in some other fixed role).
9651 @var{reg} must be the name of a register. The register names accepted
9652 are machine-specific and are defined in the @code{REGISTER_NAMES}
9653 macro in the machine description macro file.
9655 This flag does not have a negative form, because it specifies a
9658 @item -fcall-used-@var{reg}
9660 Treat the register named @var{reg} as an allocable register that is
9661 clobbered by function calls. It may be allocated for temporaries or
9662 variables that do not live across a call. Functions compiled this way
9663 will not save and restore the register @var{reg}.
9665 It is an error to used this flag with the frame pointer or stack pointer.
9666 Use of this flag for other registers that have fixed pervasive roles in
9667 the machine's execution model will produce disastrous results.
9669 This flag does not have a negative form, because it specifies a
9672 @item -fcall-saved-@var{reg}
9673 @opindex fcall-saved
9674 Treat the register named @var{reg} as an allocable register saved by
9675 functions. It may be allocated even for temporaries or variables that
9676 live across a call. Functions compiled this way will save and restore
9677 the register @var{reg} if they use it.
9679 It is an error to used this flag with the frame pointer or stack pointer.
9680 Use of this flag for other registers that have fixed pervasive roles in
9681 the machine's execution model will produce disastrous results.
9683 A different sort of disaster will result from the use of this flag for
9684 a register in which function values may be returned.
9686 This flag does not have a negative form, because it specifies a
9690 @opindex fpack-struct
9691 Pack all structure members together without holes. Usually you would
9692 not want to use this option, since it makes the code suboptimal, and
9693 the offsets of structure members won't agree with system libraries.
9695 @item -fcheck-memory-usage
9696 @opindex fcheck-memory-usage
9697 Generate extra code to check each memory access. GCC will generate
9698 code that is suitable for a detector of bad memory accesses such as
9701 Normally, you should compile all, or none, of your code with this option.
9703 If you do mix code compiled with and without this option,
9704 you must ensure that all code that has side effects
9705 and that is called by code compiled with this option
9706 is, itself, compiled with this option.
9707 If you do not, you might get erroneous messages from the detector.
9709 If you use functions from a library that have side-effects (such as
9710 @code{read}), you might not be able to recompile the library and
9711 specify this option. In that case, you can enable the
9712 @option{-fprefix-function-name} option, which requests GCC to encapsulate
9713 your code and make other functions look as if they were compiled with
9714 @option{-fcheck-memory-usage}. This is done by calling ``stubs'',
9715 which are provided by the detector. If you cannot find or build
9716 stubs for every function you call, you might have to specify
9717 @option{-fcheck-memory-usage} without @option{-fprefix-function-name}.
9719 If you specify this option, you can not use the @code{asm} or
9720 @code{__asm__} keywords in functions with memory checking enabled. GCC
9721 cannot understand what the @code{asm} statement may do, and therefore
9722 cannot generate the appropriate code, so it will reject it. However, if
9723 you specify the function attribute @code{no_check_memory_usage}
9724 (@pxref{Function Attributes}), GCC will disable memory checking within a
9725 function; you may use @code{asm} statements inside such functions. You
9726 may have an inline expansion of a non-checked function within a checked
9727 function; in that case GCC will not generate checks for the inlined
9728 function's memory accesses.
9730 If you move your @code{asm} statements to non-checked inline functions
9731 and they do access memory, you can add calls to the support code in your
9732 inline function, to indicate any reads, writes, or copies being done.
9733 These calls would be similar to those done in the stubs described above.
9735 @item -fprefix-function-name
9736 @opindex fprefix-function-name
9737 Request GCC to add a prefix to the symbols generated for function names.
9738 GCC adds a prefix to the names of functions defined as well as
9739 functions called. Code compiled with this option and code compiled
9740 without the option can't be linked together, unless stubs are used.
9742 If you compile the following code with @option{-fprefix-function-name}
9744 extern void bar (int);
9753 GCC will compile the code as if it was written:
9755 extern void prefix_bar (int);
9759 return prefix_bar (a + 5);
9762 This option is designed to be used with @option{-fcheck-memory-usage}.
9764 @item -finstrument-functions
9765 @opindex finstrument-functions
9766 Generate instrumentation calls for entry and exit to functions. Just
9767 after function entry and just before function exit, the following
9768 profiling functions will be called with the address of the current
9769 function and its call site. (On some platforms,
9770 @code{__builtin_return_address} does not work beyond the current
9771 function, so the call site information may not be available to the
9772 profiling functions otherwise.)
9775 void __cyg_profile_func_enter (void *this_fn,
9777 void __cyg_profile_func_exit (void *this_fn,
9781 The first argument is the address of the start of the current function,
9782 which may be looked up exactly in the symbol table.
9784 This instrumentation is also done for functions expanded inline in other
9785 functions. The profiling calls will indicate where, conceptually, the
9786 inline function is entered and exited. This means that addressable
9787 versions of such functions must be available. If all your uses of a
9788 function are expanded inline, this may mean an additional expansion of
9789 code size. If you use @samp{extern inline} in your C code, an
9790 addressable version of such functions must be provided. (This is
9791 normally the case anyways, but if you get lucky and the optimizer always
9792 expands the functions inline, you might have gotten away without
9793 providing static copies.)
9795 A function may be given the attribute @code{no_instrument_function}, in
9796 which case this instrumentation will not be done. This can be used, for
9797 example, for the profiling functions listed above, high-priority
9798 interrupt routines, and any functions from which the profiling functions
9799 cannot safely be called (perhaps signal handlers, if the profiling
9800 routines generate output or allocate memory).
9803 @opindex fstack-check
9804 Generate code to verify that you do not go beyond the boundary of the
9805 stack. You should specify this flag if you are running in an
9806 environment with multiple threads, but only rarely need to specify it in
9807 a single-threaded environment since stack overflow is automatically
9808 detected on nearly all systems if there is only one stack.
9810 Note that this switch does not actually cause checking to be done; the
9811 operating system must do that. The switch causes generation of code
9812 to ensure that the operating system sees the stack being extended.
9814 @item -fstack-limit-register=@var{reg}
9815 @itemx -fstack-limit-symbol=@var{sym}
9816 @itemx -fno-stack-limit
9817 @opindex fstack-limit-register
9818 @opindex fstack-limit-symbol
9819 @opindex fno-stack-limit
9820 Generate code to ensure that the stack does not grow beyond a certain value,
9821 either the value of a register or the address of a symbol. If the stack
9822 would grow beyond the value, a signal is raised. For most targets,
9823 the signal is raised before the stack overruns the boundary, so
9824 it is possible to catch the signal without taking special precautions.
9826 For instance, if the stack starts at absolute address @samp{0x80000000}
9827 and grows downwards, you can use the flags
9828 @option{-fstack-limit-symbol=__stack_limit} and
9829 @option{-Wl,--defsym,__stack_limit=0x7ffe0000} to enforce a stack limit
9830 of 128KB@. Note that this may only work with the GNU linker.
9832 @cindex aliasing of parameters
9833 @cindex parameters, aliased
9834 @item -fargument-alias
9835 @itemx -fargument-noalias
9836 @itemx -fargument-noalias-global
9837 @opindex fargument-alias
9838 @opindex fargument-noalias
9839 @opindex fargument-noalias-global
9840 Specify the possible relationships among parameters and between
9841 parameters and global data.
9843 @option{-fargument-alias} specifies that arguments (parameters) may
9844 alias each other and may alias global storage.@*
9845 @option{-fargument-noalias} specifies that arguments do not alias
9846 each other, but may alias global storage.@*
9847 @option{-fargument-noalias-global} specifies that arguments do not
9848 alias each other and do not alias global storage.
9850 Each language will automatically use whatever option is required by
9851 the language standard. You should not need to use these options yourself.
9853 @item -fleading-underscore
9854 @opindex fleading-underscore
9855 This option and its counterpart, @option{-fno-leading-underscore}, forcibly
9856 change the way C symbols are represented in the object file. One use
9857 is to help link with legacy assembly code.
9859 Be warned that you should know what you are doing when invoking this
9860 option, and that not all targets provide complete support for it.
9865 @node Environment Variables
9866 @section Environment Variables Affecting GCC
9867 @cindex environment variables
9869 @c man begin ENVIRONMENT
9871 This section describes several environment variables that affect how GCC
9872 operates. Some of them work by specifying directories or prefixes to use
9873 when searching for various kinds of files. Some are used to specify other
9874 aspects of the compilation environment.
9877 Note that you can also specify places to search using options such as
9878 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
9879 take precedence over places specified using environment variables, which
9880 in turn take precedence over those specified by the configuration of GCC@.
9884 Note that you can also specify places to search using options such as
9885 @option{-B}, @option{-I} and @option{-L} (@pxref{Directory Options}). These
9886 take precedence over places specified using environment variables, which
9887 in turn take precedence over those specified by the configuration of GCC@.
9894 @c @itemx LC_COLLATE
9896 @c @itemx LC_MONETARY
9897 @c @itemx LC_NUMERIC
9902 @c @findex LC_COLLATE
9904 @c @findex LC_MONETARY
9905 @c @findex LC_NUMERIC
9909 These environment variables control the way that GCC uses
9910 localization information that allow GCC to work with different
9911 national conventions. GCC inspects the locale categories
9912 @env{LC_CTYPE} and @env{LC_MESSAGES} if it has been configured to do
9913 so. These locale categories can be set to any value supported by your
9914 installation. A typical value is @samp{en_UK} for English in the United
9917 The @env{LC_CTYPE} environment variable specifies character
9918 classification. GCC uses it to determine the character boundaries in
9919 a string; this is needed for some multibyte encodings that contain quote
9920 and escape characters that would otherwise be interpreted as a string
9923 The @env{LC_MESSAGES} environment variable specifies the language to
9924 use in diagnostic messages.
9926 If the @env{LC_ALL} environment variable is set, it overrides the value
9927 of @env{LC_CTYPE} and @env{LC_MESSAGES}; otherwise, @env{LC_CTYPE}
9928 and @env{LC_MESSAGES} default to the value of the @env{LANG}
9929 environment variable. If none of these variables are set, GCC
9930 defaults to traditional C English behavior.
9934 If @env{TMPDIR} is set, it specifies the directory to use for temporary
9935 files. GCC uses temporary files to hold the output of one stage of
9936 compilation which is to be used as input to the next stage: for example,
9937 the output of the preprocessor, which is the input to the compiler
9940 @item GCC_EXEC_PREFIX
9941 @findex GCC_EXEC_PREFIX
9942 If @env{GCC_EXEC_PREFIX} is set, it specifies a prefix to use in the
9943 names of the subprograms executed by the compiler. No slash is added
9944 when this prefix is combined with the name of a subprogram, but you can
9945 specify a prefix that ends with a slash if you wish.
9947 If @env{GCC_EXEC_PREFIX} is not set, GCC will attempt to figure out
9948 an appropriate prefix to use based on the pathname it was invoked with.
9950 If GCC cannot find the subprogram using the specified prefix, it
9951 tries looking in the usual places for the subprogram.
9953 The default value of @env{GCC_EXEC_PREFIX} is
9954 @file{@var{prefix}/lib/gcc-lib/} where @var{prefix} is the value
9955 of @code{prefix} when you ran the @file{configure} script.
9957 Other prefixes specified with @option{-B} take precedence over this prefix.
9959 This prefix is also used for finding files such as @file{crt0.o} that are
9962 In addition, the prefix is used in an unusual way in finding the
9963 directories to search for header files. For each of the standard
9964 directories whose name normally begins with @samp{/usr/local/lib/gcc-lib}
9965 (more precisely, with the value of @env{GCC_INCLUDE_DIR}), GCC tries
9966 replacing that beginning with the specified prefix to produce an
9967 alternate directory name. Thus, with @option{-Bfoo/}, GCC will search
9968 @file{foo/bar} where it would normally search @file{/usr/local/lib/bar}.
9969 These alternate directories are searched first; the standard directories
9973 @findex COMPILER_PATH
9974 The value of @env{COMPILER_PATH} is a colon-separated list of
9975 directories, much like @env{PATH}. GCC tries the directories thus
9976 specified when searching for subprograms, if it can't find the
9977 subprograms using @env{GCC_EXEC_PREFIX}.
9980 @findex LIBRARY_PATH
9981 The value of @env{LIBRARY_PATH} is a colon-separated list of
9982 directories, much like @env{PATH}. When configured as a native compiler,
9983 GCC tries the directories thus specified when searching for special
9984 linker files, if it can't find them using @env{GCC_EXEC_PREFIX}. Linking
9985 using GCC also uses these directories when searching for ordinary
9986 libraries for the @option{-l} option (but directories specified with
9987 @option{-L} come first).
9989 @item C_INCLUDE_PATH
9990 @itemx CPLUS_INCLUDE_PATH
9991 @itemx OBJC_INCLUDE_PATH
9992 @findex C_INCLUDE_PATH
9993 @findex CPLUS_INCLUDE_PATH
9994 @findex OBJC_INCLUDE_PATH
9995 @c @itemx OBJCPLUS_INCLUDE_PATH
9996 These environment variables pertain to particular languages. Each
9997 variable's value is a colon-separated list of directories, much like
9998 @env{PATH}. When GCC searches for header files, it tries the
9999 directories listed in the variable for the language you are using, after
10000 the directories specified with @option{-I} but before the standard header
10003 @item DEPENDENCIES_OUTPUT
10004 @findex DEPENDENCIES_OUTPUT
10005 @cindex dependencies for make as output
10006 If this variable is set, its value specifies how to output dependencies
10007 for Make based on the header files processed by the compiler. This
10008 output looks much like the output from the @option{-M} option
10009 (@pxref{Preprocessor Options}), but it goes to a separate file, and is
10010 in addition to the usual results of compilation.
10012 The value of @env{DEPENDENCIES_OUTPUT} can be just a file name, in
10013 which case the Make rules are written to that file, guessing the target
10014 name from the source file name. Or the value can have the form
10015 @samp{@var{file} @var{target}}, in which case the rules are written to
10016 file @var{file} using @var{target} as the target name.
10020 @cindex locale definition
10021 This variable is used to pass locale information to the compiler. One way in
10022 which this information is used is to determine the character set to be used
10023 when character literals, string literals and comments are parsed in C and C++.
10024 When the compiler is configured to allow multibyte characters,
10025 the following values for @env{LANG} are recognized:
10029 Recognize JIS characters.
10031 Recognize SJIS characters.
10033 Recognize EUCJP characters.
10036 If @env{LANG} is not defined, or if it has some other value, then the
10037 compiler will use mblen and mbtowc as defined by the default locale to
10038 recognize and translate multibyte characters.
10043 @node Running Protoize
10044 @section Running Protoize
10046 The program @code{protoize} is an optional part of GCC@. You can use
10047 it to add prototypes to a program, thus converting the program to ISO
10048 C in one respect. The companion program @code{unprotoize} does the
10049 reverse: it removes argument types from any prototypes that are found.
10051 When you run these programs, you must specify a set of source files as
10052 command line arguments. The conversion programs start out by compiling
10053 these files to see what functions they define. The information gathered
10054 about a file @var{foo} is saved in a file named @file{@var{foo}.X}.
10056 After scanning comes actual conversion. The specified files are all
10057 eligible to be converted; any files they include (whether sources or
10058 just headers) are eligible as well.
10060 But not all the eligible files are converted. By default,
10061 @code{protoize} and @code{unprotoize} convert only source and header
10062 files in the current directory. You can specify additional directories
10063 whose files should be converted with the @option{-d @var{directory}}
10064 option. You can also specify particular files to exclude with the
10065 @option{-x @var{file}} option. A file is converted if it is eligible, its
10066 directory name matches one of the specified directory names, and its
10067 name within the directory has not been excluded.
10069 Basic conversion with @code{protoize} consists of rewriting most
10070 function definitions and function declarations to specify the types of
10071 the arguments. The only ones not rewritten are those for varargs
10074 @code{protoize} optionally inserts prototype declarations at the
10075 beginning of the source file, to make them available for any calls that
10076 precede the function's definition. Or it can insert prototype
10077 declarations with block scope in the blocks where undeclared functions
10080 Basic conversion with @code{unprotoize} consists of rewriting most
10081 function declarations to remove any argument types, and rewriting
10082 function definitions to the old-style pre-ISO form.
10084 Both conversion programs print a warning for any function declaration or
10085 definition that they can't convert. You can suppress these warnings
10088 The output from @code{protoize} or @code{unprotoize} replaces the
10089 original source file. The original file is renamed to a name ending
10090 with @samp{.save} (for DOS, the saved filename ends in @samp{.sav}
10091 without the original @samp{.c} suffix). If the @samp{.save} (@samp{.sav}
10092 for DOS) file already exists, then the source file is simply discarded.
10094 @code{protoize} and @code{unprotoize} both depend on GCC itself to
10095 scan the program and collect information about the functions it uses.
10096 So neither of these programs will work until GCC is installed.
10098 Here is a table of the options you can use with @code{protoize} and
10099 @code{unprotoize}. Each option works with both programs unless
10103 @item -B @var{directory}
10104 Look for the file @file{SYSCALLS.c.X} in @var{directory}, instead of the
10105 usual directory (normally @file{/usr/local/lib}). This file contains
10106 prototype information about standard system functions. This option
10107 applies only to @code{protoize}.
10109 @item -c @var{compilation-options}
10110 Use @var{compilation-options} as the options when running @code{gcc} to
10111 produce the @samp{.X} files. The special option @option{-aux-info} is
10112 always passed in addition, to tell @code{gcc} to write a @samp{.X} file.
10114 Note that the compilation options must be given as a single argument to
10115 @code{protoize} or @code{unprotoize}. If you want to specify several
10116 @code{gcc} options, you must quote the entire set of compilation options
10117 to make them a single word in the shell.
10119 There are certain @code{gcc} arguments that you cannot use, because they
10120 would produce the wrong kind of output. These include @option{-g},
10121 @option{-O}, @option{-c}, @option{-S}, and @option{-o} If you include these in
10122 the @var{compilation-options}, they are ignored.
10125 Rename files to end in @samp{.C} (@samp{.cc} for DOS-based file
10126 systems) instead of @samp{.c}. This is convenient if you are converting
10127 a C program to C++. This option applies only to @code{protoize}.
10130 Add explicit global declarations. This means inserting explicit
10131 declarations at the beginning of each source file for each function
10132 that is called in the file and was not declared. These declarations
10133 precede the first function definition that contains a call to an
10134 undeclared function. This option applies only to @code{protoize}.
10136 @item -i @var{string}
10137 Indent old-style parameter declarations with the string @var{string}.
10138 This option applies only to @code{protoize}.
10140 @code{unprotoize} converts prototyped function definitions to old-style
10141 function definitions, where the arguments are declared between the
10142 argument list and the initial @samp{@{}. By default, @code{unprotoize}
10143 uses five spaces as the indentation. If you want to indent with just
10144 one space instead, use @option{-i " "}.
10147 Keep the @samp{.X} files. Normally, they are deleted after conversion
10151 Add explicit local declarations. @code{protoize} with @option{-l} inserts
10152 a prototype declaration for each function in each block which calls the
10153 function without any declaration. This option applies only to
10157 Make no real changes. This mode just prints information about the conversions
10158 that would have been done without @option{-n}.
10161 Make no @samp{.save} files. The original files are simply deleted.
10162 Use this option with caution.
10164 @item -p @var{program}
10165 Use the program @var{program} as the compiler. Normally, the name
10166 @file{gcc} is used.
10169 Work quietly. Most warnings are suppressed.
10172 Print the version number, just like @option{-v} for @code{gcc}.
10175 If you need special compiler options to compile one of your program's
10176 source files, then you should generate that file's @samp{.X} file
10177 specially, by running @code{gcc} on that source file with the
10178 appropriate options and the option @option{-aux-info}. Then run
10179 @code{protoize} on the entire set of files. @code{protoize} will use
10180 the existing @samp{.X} file because it is newer than the source file.
10184 gcc -Dfoo=bar file1.c -aux-info file1.X
10189 You need to include the special files along with the rest in the
10190 @code{protoize} command, even though their @samp{.X} files already
10191 exist, because otherwise they won't get converted.
10193 @xref{Protoize Caveats}, for more information on how to use
10194 @code{protoize} successfully.